WO2018121368A1 - 一种歌词的配乐生成方法和相关装置 - Google Patents

一种歌词的配乐生成方法和相关装置 Download PDF

Info

Publication number
WO2018121368A1
WO2018121368A1 PCT/CN2017/117358 CN2017117358W WO2018121368A1 WO 2018121368 A1 WO2018121368 A1 WO 2018121368A1 CN 2017117358 W CN2017117358 W CN 2017117358W WO 2018121368 A1 WO2018121368 A1 WO 2018121368A1
Authority
WO
WIPO (PCT)
Prior art keywords
soundtrack
segment
music
lyrics
word
Prior art date
Application number
PCT/CN2017/117358
Other languages
English (en)
French (fr)
Inventor
叶舟
王瑜
张亚楠
苏飞
杨洋
Original Assignee
阿里巴巴集团控股有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 阿里巴巴集团控股有限公司 filed Critical 阿里巴巴集团控股有限公司
Publication of WO2018121368A1 publication Critical patent/WO2018121368A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/60Information retrieval; Database structures therefor; File system structures therefor of audio data
    • G06F16/63Querying
    • G06F16/632Query formulation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/60Information retrieval; Database structures therefor; File system structures therefor of audio data
    • G06F16/68Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/683Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G06F16/685Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using automatically derived transcript of audio data, e.g. lyrics

Definitions

  • the present invention relates to the field of data processing, and in particular to a soundtrack generating method and related apparatus for lyrics.
  • the lyrics are created first, and then the lyrics are composed according to the lyrics, that is, the lyrics complete the corresponding soundtrack, thereby forming a song.
  • the present invention provides a soundtrack generating method and related device for lyrics, which automatically generates a soundtrack of a to-be-matched lyrics by using the matched soundtrack segment and the position of the word segmentation in the lyrics to be played, effectively improving the score.
  • the efficiency of the lyrics soundtrack is a soundtrack generating method and related device for lyrics, which automatically generates a soundtrack of a to-be-matched lyrics by using the matched soundtrack segment and the position of the word segmentation in the lyrics to be played, effectively improving the score.
  • the present invention provides a soundtrack generating method for a lyric, which is applied to a soundtrack server, the soundtrack server including a music library, the soundtrack library including a first correspondence between a word segmentation and a music score, wherein any one of the word segments has a A set of scores including at least one soundtrack segment, the method comprising:
  • the to-be-matched music lyrics including at least one text segment, wherein one of the at least one text segment includes at least one word segment;
  • the soundtrack of the to-be-matched lyrics is generated by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the first text segment is one of the at least one text segment
  • the at least one text is determined according to the music score segment set corresponding to the plurality of word segments and the word segment included in the at least one text segment.
  • the corresponding sub-tracks of the clips include:
  • a set of soundtracks whose degree of smoothness meets a preset condition is used as a sub-track of the first text segment.
  • the first participle and the second participle are adjacent participles in the first piece of text, and determining, according to the adjacent relationship of the participles included in the first piece of text, the participles included from the first piece of text
  • the fluency between the selected scores in the corresponding score collection including:
  • the group of music scores that satisfy the preset threshold value as the sub-track of the first text segment includes:
  • a set of soundtrack segments corresponding to a stitching total score is selected from the stitching total score exceeding the preset threshold as a sub-track of the first text segment.
  • the calculating a degree of splicing between the soundtrack segment in the soundtrack segment set corresponding to the first segment of the first segment of the first segment and the score segment in the score segment corresponding to the second segment segment, to obtain the first segmentation At least one stitch score between the second participle and the second participle, including:
  • a degree of splicing between the two score segments is calculated, and a stitch score determined by the two score segments between the first word segment and the second word segment is obtained.
  • the soundtrack server further includes a type library, wherein the type library includes possibility information that the word segment belongs to different song types, and the music library further includes a second correspondence relationship between the soundtrack segment and the song type;
  • the method further includes:
  • it also includes:
  • the three-part word is a participle obtained from the historical lyrics
  • the type library is created based on the possibility information that the word segmentation obtained in the historical lyrics belongs to different song types.
  • it also includes:
  • the obtaining a plurality of word segments obtained by performing word segmentation processing on the music lyrics includes:
  • the present invention provides a soundtrack generating apparatus for a lyrics, which is applied to a soundtrack server, the soundtrack server including a music library, the soundtrack library including a first correspondence between a word segmentation and a music score, wherein any one of the word segments has one A soundtrack segment set including at least one soundtrack segment, the device comprising an acquisition unit, a lookup unit, a determination unit, and a generation unit:
  • the acquiring unit is configured to acquire a plurality of participles obtained by performing word segmentation processing on the music lyrics, wherein the to-be-matched music lyrics include at least one text segment, and one of the at least one text segment includes at least one word segment;
  • the searching unit is configured to search the music library according to the plurality of word segments, and determine, by using the first correspondence relationship, a soundtrack segment set corresponding to each word segment;
  • the determining unit is configured to determine, according to the music score segment set corresponding to the plurality of word segments and the word segment included in the at least one text segment, the sub-tracks corresponding to the at least one text segment respectively;
  • the generating unit is configured to generate a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the first text segment is one of the at least one text segment
  • the determining unit is specifically configured to acquire a music score segment set corresponding to the word segment included in the first text segment; and according to the word segment included in the first text segment a neighboring relationship, determining a degree of smoothness between the selected soundtrack segments from the set of scores corresponding to the participles included in the first text segment; and a set of soundtrack segments satisfying the preset condition as the first text segment Sub-track.
  • the first participle and the second participle are adjacent participles in the first piece of text
  • the determining unit comprises a calculating subunit, obtaining a subunit, and as a subunit:
  • the calculating sub-unit is configured to calculate a splicing degree between the soundtrack segment in the soundtrack segment set corresponding to the first segment of the first segment and the score segment in the score segment corresponding to the second segment segment, to obtain the At least one stitching score between a participle and a second participle;
  • the obtaining subunit is configured to calculate, according to the splicing score of the adjacent word segment in the first text segment, at least one splicing total score of the first text segment;
  • the subunit is configured to select a group of music scores corresponding to a spliced total score from a splicing total score exceeding a preset threshold as a sub-track of the first text segment.
  • the calculating sub-unit is specifically configured to acquire one of the soundtrack segments in the set of music scores corresponding to the first word segment, and obtain one of the soundtrack segments in the set of music score segments corresponding to the second word segment; The degree of splicing between the score segments yields a stitch score determined by the two score segments between the first segment and the second segment.
  • the soundtrack server further includes a type library, wherein the type library includes possibility information that the word segment belongs to different song types, and the music library further includes a second correspondence between the soundtrack segment and the song type;
  • the device also includes determining a song type unit:
  • the determining a song type unit configured to search the type library according to the plurality of word segments, and determine a song type to which the to-be-matched music lyric belongs;
  • the lookup unit includes a screening subunit and a determining subunit:
  • the screening subunit is configured to filter, according to the second correspondence relationship, a soundtrack segment corresponding to a song type to which the to-be-matched music lyrics belongs according to the music library;
  • the determining subunit is configured to determine, according to the first correspondence, a set of music scores corresponding to the plurality of word segments from the soundtrack segments corresponding to the song type to which the music to be played.
  • the device further includes: acquiring a historical song unit, a word segmentation unit, a statistics unit, a determining information unit, and an establishing unit:
  • the acquiring a historical song unit for acquiring a historical song, the historical song including a historical lyric, a historical soundtrack, and a song type;
  • the word segmentation unit is configured to segment the historical lyrics
  • the statistical unit is configured to count the number of occurrences of the third participle in the word segmentation process, and the correspondence between the third participle and the song type of the belonging historical song, thereby determining the third participle in the same song type
  • the number of occurrences, the third participle is a participle obtained from the historical lyrics
  • the determining information unit is configured to determine, according to the number of occurrences of the third word segmentation in the word segmentation process and the number of occurrences of the third word segment in the same song type, the possibility that the third word segment belongs to different song types information;
  • the establishing unit is configured to establish the type library according to the possibility information that the word segmentation obtained in the historical lyrics belongs to different song types.
  • the device further includes: acquiring a historical song unit, a word segmentation unit, determining a soundtrack segment unit, and establishing the unit:
  • the acquiring a historical song unit for acquiring a historical song, the historical song including a historical lyric, a historical soundtrack, and a song type;
  • the word segmentation unit is configured to segment the historical lyrics
  • the establishing unit is configured to establish the music library according to the first correspondence between the obtained word segment and the music score in the historical soundtrack, and the second correspondence between the music score and the song type in the historical soundtrack.
  • the obtaining unit is specifically configured to perform word segmentation on the acquired lyrics to be scored to obtain a plurality of word segments.
  • the present invention provides a soundtrack generating method for a lyrics, which is applied to a soundtrack server, the soundtrack server including a music library, the soundtrack library including a first correspondence relationship between a segment and a soundtrack segment, wherein any one of the segments has a A set of scores including at least one soundtrack segment, the method comprising:
  • a soundtrack of the to-be-matched lyrics is generated by stitching the determined soundtrack segments.
  • the segment is a participle
  • the acquiring at least one segment obtained by performing word segmentation processing on the music lyrics includes:
  • the segment is a text segment
  • the to-be-matched music lyric includes at least one text segment
  • the to-be-matched music lyrics include at least one text segment, and one of the at least one text segment includes at least one word segment, and the soundtrack corresponding to the at least one word segment according to the to-be-matched music lyrics respectively Determining, in the set of fragments, a soundtrack segment corresponding to each of the at least one participle, including:
  • the soundtrack segment determined by the splicing generates the soundtrack of the to-be-matched music lyrics, including:
  • the soundtrack of the to-be-matched lyrics is generated by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the present invention provides a soundtrack generating apparatus for a lyric, which is applied to a soundtrack server, and the soundtrack server includes a music library, and the soundtrack library includes a first correspondence relationship between a segment and a soundtrack segment, wherein any one of the segments has a A set of scores including at least one soundtrack segment, the apparatus comprising an acquisition unit, a lookup unit, a determination unit, and a splicing unit:
  • the obtaining unit is configured to acquire at least one segment obtained by performing word segmentation processing on the music lyrics;
  • the searching unit is configured to search the music library according to the at least one segment, and determine, by using the first correspondence, a soundtrack segment set corresponding to the at least one segment respectively;
  • the determining unit is configured to determine, according to the to-be-matched music lyrics, a soundtrack segment corresponding to each of the at least one segment from the set of soundtrack segments respectively corresponding to the at least one segment;
  • the splicing unit is configured to generate a soundtrack of the to-be-matched lyrics by stitching the determined soundtrack segments.
  • the obtaining unit is specifically configured to acquire a plurality of word segments obtained by performing word segmentation processing on the to-be-matched music lyrics.
  • the determining unit is configured to determine, according to the music score segment corresponding to the plurality of word segments and the word segment included in the at least one text segment, a sub-track corresponding to the at least one text segment, where The sub-score corresponding to the text segment is obtained from the soundtrack segment corresponding to the word segment included in the text segment;
  • the splicing unit is specifically configured to generate a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the present invention provides a method for acquiring a soundtrack of a lyric, which is applied to an interactive terminal, and the method includes:
  • the to-be-matched music lyrics include at least one text segment, one of the at least one text segment includes at least one word segment, the soundtrack server includes a music library, and the music library includes a first correspondence between the word segmentation and the score segment, wherein any one of the segmentation words has a set of music score segments including at least one soundtrack segment;
  • the corresponding soundtrack set is obtained by the soundtrack server searching the soundtrack library according to the first correspondence.
  • the obtaining, by the soundtrack server, the soundtrack corresponding to the to-be-matched music lyrics includes:
  • the to-be-matched music that meets the demand is selected as the soundtrack of the to-be-matched music lyrics.
  • the soundtrack information includes a score of a to-be-scheduled music piece carrying the music score information and/or a soundtrack type.
  • the method further includes:
  • the feedback information is sent to the soundtrack server, so that the server regenerates the soundtrack of the to-be-matched music lyrics according to the feedback information.
  • the feedback information includes information describing the requirement.
  • the present invention provides a soundtrack obtaining apparatus for a lyric, which is applied to an interactive terminal, and the apparatus includes a transmitting unit and an acquiring unit:
  • the sending unit is configured to send the acquired to-be-matched music lyrics to the soundtrack server, the to-be-matched music lyrics include at least one text segment, one of the at least one text segment includes at least one word segment, and the soundtrack server includes a soundtrack a library, the music library comprising a first correspondence of word segmentation and music scores, wherein any one of the word segments has a soundtrack segment set including at least one soundtrack segment;
  • the obtaining unit is configured to obtain, from the soundtrack server, a soundtrack corresponding to the to-be-matched music lyrics, where the soundtrack is obtained by stitching the music scores in the music score segment corresponding to the word segmentation in the to-be-matched music lyrics.
  • the soundtrack segment set corresponding to the word segmentation in the to-be-matched music lyrics is obtained by the soundtrack server searching the music library according to the first correspondence relationship.
  • the obtaining unit includes an obtaining subunit and a selecting subunit:
  • the obtaining subunit is configured to acquire, from the soundtrack server, a plurality of to-be-sorted music corresponding to the to-be-matched music lyrics, where the to-be-scheduled musictrack carries the music score information;
  • the selection sub-unit is configured to select, according to the music information of the plurality of to-be-determined music scores, a to-be-sorted music that meets the requirement as a soundtrack of the to-be-matched music lyrics.
  • the soundtrack information includes a score of a to-be-scheduled music piece carrying the music score information and/or a soundtrack type.
  • the device further includes a feedback unit:
  • the feedback unit is triggered, and the feedback unit is configured to send feedback information to the soundtrack server, so that the server regenerates the soundtrack of the to-be-matched music lyrics according to the feedback information.
  • the feedback information includes information describing the requirement.
  • the present invention provides a soundtrack system for a lyric, the soundtrack system comprising a soundtrack server and an interactive end:
  • the soundtrack server includes a music library including a first correspondence of word segmentation and music scores, wherein any one of the word segments has a soundtrack segment set including at least one soundtrack segment, and the soundtrack server is configured to obtain a soundtrack through treatment
  • the lyrics are subjected to word segmentation processing to obtain a plurality of word segments, the to-be-matched music lyrics include at least one text segment, and one of the at least one text segment includes at least one word segment; searching the music library according to the plurality of word segments, and utilizing Determining, by the first correspondence, a set of scores corresponding to each of the word segments; determining, according to the set of scores corresponding to the plurality of particials and the participles included in the at least one text segment, the at least one text segment respectively a corresponding sub-track; generating a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively;
  • the interaction end is configured to send the acquired to-be-matched music lyrics to the soundtrack server; and acquire a soundtrack corresponding to the to-be-matched music lyrics from the soundtrack server.
  • the present invention provides a music editor having an input interface for acquiring a lyrics to be input, and an editing interface for displaying a soundtrack;
  • the input interface is configured to obtain a lyrics to be played, and the acquired lyrics to be played are lyrics input in the input interface, or lyrics obtained after identifying the input voice;
  • the music editor is configured to send, to the soundtrack server, the to-be-matched lyrics acquired through the input interface, the soundtrack server includes a music library, and the music library includes a first correspondence between the word segmentation and the music score segment. Any one of the word segments has a set of soundtrack segments including at least one soundtrack segment;
  • the music editor is configured to display the soundtrack corresponding to the to-be-matched music lyrics in the editing interface, and the soundtrack corresponding to the to-be-matched music lyrics is used as the soundtrack corresponding to the to-be-matched music lyrics.
  • the soundtrack server is spliced according to the music scores in the set of music scores corresponding to the word segmentation in the music lyrics to be scored, and the soundtrack segment corresponding to the word segmentation in the music to be lyrics is used by the soundtrack server to search the music library according to the first correspondence relationship. owned.
  • the editing interface has an editing button, and when the triggering of the editing button is received, the editing interface is configured to set the displayed soundtrack to be editable;
  • the soundtrack displayed by the editing interface also includes a soundtrack obtained by recognizing the recorded audio data.
  • the present invention provides a music player, the music player having a hybrid interface and a play interface;
  • the mixing interface is configured to mix the music to be played and the soundtrack corresponding to the music to be played to obtain a song
  • the soundtrack corresponding to the music to be played is a soundtrack server according to the music score segment corresponding to the word segmentation in the to-be-matched music lyrics.
  • the soundtrack segments are stitched together, and the soundtrack segment set corresponding to the word segmentation in the to-be-matched music lyrics is obtained by the soundtrack server searching for the music library according to the first correspondence relationship, and the soundtrack server includes a music library, and the music library includes word segmentation and soundtrack segments. a first correspondence relationship, wherein any one of the word segments has a soundtrack segment set including at least one soundtrack segment;
  • the play interface is used to play the song.
  • the soundtrack server includes a soundtrack library including a first correspondence relationship between the word segmentation and the soundtrack segment, and after acquiring a plurality of word segments obtained by performing word segmentation processing on the music scores, the plurality of word segments can be utilized.
  • the participles determine the set of scores corresponding to the plurality of participles from the score library according to the first correspondence, and then combine the scores of the scores corresponding to the participles in the text segment by using the text segment of the lyrics to be matched.
  • the sub-track corresponding to the text segment, and the sub-track of the text segment is spliced into the soundtrack of the to-be-matched lyrics.
  • the music score corresponding to the word segment is pre-stored, the music score corresponding to the word segment can be matched after the word segmentation of the music score is processed. Therefore, the soundtrack of the to-be-matched lyrics can be automatically generated by using the matched scores and the position of the participle in the lyrics to be played, which effectively improves the efficiency of the lyrics.
  • FIG. 1 is a schematic diagram of a system for generating a soundtrack of a lyric according to an embodiment of the present invention
  • FIG. 2 is a flowchart of a method for generating a soundtrack of a lyric according to an embodiment of the present invention
  • FIG. 2a is a flowchart of a method for determining a sub-track corresponding to a first text segment according to an embodiment of the present invention.
  • FIG. 3 is a schematic flowchart of a process for establishing a music library according to an embodiment of the present invention
  • FIG. 4 is a schematic flowchart of a type library establishment process according to an embodiment of the present invention.
  • FIG. 5 is a flowchart of a method for generating a soundtrack of another lyric according to an embodiment of the present invention
  • FIG. 6 is a flowchart of a method for acquiring a soundtrack of a lyric according to an embodiment of the present invention
  • FIG. 7 is a structural diagram of a device for generating a soundtrack of a lyric according to an embodiment of the present invention.
  • FIG. 8 is a structural diagram of an apparatus for generating a soundtrack of another lyric according to an embodiment of the present invention.
  • FIG. 9 is a structural diagram of a device for acquiring a soundtrack of a lyric according to an embodiment of the present invention.
  • FIG. 10 is a system structural diagram of a soundtrack system for a lyric according to an embodiment of the present invention.
  • the completion of a song is generally a vocabulary, that is, the lyrics are first created, and then the lyrics are scored (soundtrack) to form a song. It takes a lot of time to score the lyrics. With the development of deep learning in the field of speech, more and more tasks can be automated through the machine. If the automatic music score based on the content of the lyrics will greatly improve the efficiency of music creation.
  • an embodiment of the present invention provides a soundtrack generating method for a lyrics.
  • the soundtrack server acquires a plurality of word segments obtained by performing word segmentation processing on the music lyrics, and matches the word segmentation according to the correspondence between the pre-saved word segmentation and the music score segment.
  • the soundtrack segment can automatically generate the soundtrack of the to-be-matched lyrics by using the matched score segment and the position of the word segmentation in the lyrics to be scored.
  • the embodiment of the present invention can be implemented by a soundtrack server.
  • the soundtrack server includes a music library 200 in which the correspondence between the word segmentation and the music score segment is stored in the music library 200.
  • the word segment included in the first text segment 100 such as the first word segment, the second word segment, and the third word segment, can be obtained.
  • the set of scores 300 corresponding to the participles can be determined from the correspondence between the participles stored in the score library 200 and the scores. As shown in FIG.
  • each of the soundtrack clip sets includes at least one soundtrack clip, for example, the score set A may include soundtracks a1, a2, and a3, the score set B may include soundtracks b1 and b2, and the score set C may include soundtracks c1, c2 And c3.
  • the soundtrack server selects a suitable score segment from the set of scores corresponding to each participle as the sub-track 400 of the first text segment, and if the score segment a1-b2-c3 can be selected as the sub-track 400 of the text segment.
  • a lyric can be understood as a lyric for forming a complete song, and a lyric may include at least one text segment, and a text segment may be a paragraph of a word structure category, or may be Punctuation conforms to a sentence or a paragraph of text as a division.
  • a piece of text can consist of at least one participle.
  • the first text segment is a text segment of the at least one text segment.
  • the invention does not limit the specific implementation manner of the word segmentation of the lyrics, and the word segmentation having the independent word meaning or the word structure may be divided from the lyrics by word meaning or word structure, etc., and the segmented word segment may be a word or a word. .
  • a text segment "I love Beijing Tiananmen” can be divided into “I", “Love”, “Beijing” and "Tiananmen”.
  • a score segment can be understood as a score segment determined for a participle.
  • a participle is in a lyric. This lyric has an already assigned score.
  • the lyrics are identified by the time range corresponding to the participle on the timeline.
  • a part of the soundtrack corresponding to the position is intercepted from the already arranged soundtrack, for example, the soundtrack is in a part of the soundtrack in the time range, and the partial soundtrack may be the soundtrack segment corresponding to the word segmentation.
  • a song may include a lyric and a soundtrack corresponding to the lyric, and a text segment of the lyric has a sub-track corresponding to the text segment, and the sub-track is a part of the score corresponding to the text segment in the tempo.
  • the sub-track corresponding to this text segment can be composed of the scores corresponding to the word segmentation in this text segment.
  • an unscore text segment "I love Beijing Tiananmen” can be divided into “I”, “Love”, “Beijing” And “Tiananmen.”
  • the participle “I” has a corresponding score segment A including the score segments a1, a2 and a3, the segmentation “love” has a corresponding score segment B including the score segments b1 and b2, and the segment “Beijing” has a corresponding score segment set.
  • C includes the music scores c1, c2 and c3, and the participle "Tiananmen” has a corresponding score segment D including the soundtrack segment d1, then the sub-track of the text segment "I love Beijing Tiananmen” may be respectively composed of the score segment A, One of B, C, and D is composed of a musical score, such as a1-b2-c3-d1, or a2-b1-c3-d1, and the like.
  • the lyrics that need to be scored can be referred to as the lyrics to be played.
  • the lyrics to be played can be known, and the lyrics can be divided into two levels, namely, the word segmentation and the text segment, and the soundtrack provided by the embodiment of the present invention.
  • the generation method can be applied to such a to-be-sorted lyrics or the division granularity of the lyrics to be dealt with, and can also be applied to the to-be-matched lyrics having other constituent forms, or to the case of other division granularity of the lyrics to be processed.
  • FIG. 2 is a flowchart of a method for generating a soundtrack of a lyric according to an embodiment of the present invention, where the method includes:
  • S201 Acquire a plurality of word segments obtained by performing word segmentation processing on the soundtrack lyrics.
  • the lyrics can be manually input or voiced into the relevant device.
  • the lyrics can be scored by the soundtrack server.
  • the soundtrack server includes a music library in which the first correspondence between the word segmentation and the soundtrack segment is pre-stored. In order to facilitate the soundtrack server to perform automatic soundtracking on the soundtrack lyrics, it is first necessary to perform word segmentation processing on the soundtrack lyrics.
  • the specific device for performing word segmentation processing on the music lyrics in the embodiment of the present invention is not limited, and the soundtrack server may perform word segmentation processing on the music lyrics to obtain a plurality of word segments. It is also possible that the other device processes the word distribution lyrics to obtain a plurality of word segments, and the soundtrack server can obtain a plurality of word segments obtained after the word segmentation process.
  • the word lyrics can be processed in a progressive manner for word segmentation, that is, the lyrics to be scored can be first divided into text segments, and then the segmentation of the segment is processed.
  • the punctuation marks can be used as the basis for dividing the text segments, or the paragraphs can be used as the basis for dividing the text segments.
  • the word segmentation is processed by word segmentation, and the word segmentation with independent word meaning or word structure can be divided from the text segment by word meaning or word structure. It is also possible to directly treat the lyrics of the soundtrack for word segmentation, and then determine the segment of the text to which the segmentation belongs.
  • the lyrics of the music to be played "I love Beijing Tiananmen, the sun rises on Tiananmen Square".
  • the lyrics can be divided into two words: “I love Beijing Tiananmen” and “Sun on the Tiananmen Square”. Fragment.
  • word meaning or word structure the word segmentation with independent meaning or word structure is divided from the lyrics.
  • the word segmentation "I love Beijing Tiananmen” is divided into words, and I can get "I", “Love”, “Beijing", "Tiananmen” "These four participles are divided into four parts of "Tiananmen", “upper”, “sun” and “sheng”.
  • S202 Searching the music library according to the plurality of word segments, and determining, by using the first correspondence relationship, a soundtrack segment set corresponding to each word segment.
  • the soundtrack server may pre-establish a soundtrack library in which the correspondence between the word segmentation and the soundtrack segment is stored, and the correspondence relationship may be referred to as a first correspondence relationship.
  • a participle may have a corresponding at least one musical score.
  • a piece of music to be played can be divided into a plurality of participles. Since a participle in the music library can have at least one piece of music corresponding to each other, when the word segment is matched by the divided word, one part can match one or more soundtracks. For a segment, a score segment matched to a segmented word segment may be used as a set of score segments corresponding to the segment, that is, a segment has a corresponding set of score segments. For example, in a song, the participle "I" matches three scores, a1, a2, and a3, which can be used as a set of scores for the participle "I".
  • determining the set of the scores corresponding to the first participle specifically includes: searching the score library according to the first participle, and determining the first participle A corresponding set of scores including at least one soundtrack segment.
  • the method for determining the corresponding soundtrack segment set for other word segments is the same as the first segment word determining corresponding music score segment set, and details are not described herein again.
  • S203 Determine a sub-track corresponding to the at least one text segment according to the music score segment set corresponding to the plurality of word segments and the word segment included in the at least one text segment.
  • the music lyrics include at least one text segment, and the first text segment, that is, one of the at least one text segment, is taken as an example. If only one participle is included in the first text segment, then a soundtrack segment can be directly determined from the segmentation as a sub-track of the first text segment. If a plurality of word segments are included in the first text segment, the method for determining the sub-score corresponding to the first text segment may be as shown in FIG. 2a, and specifically includes S2031-S2033.
  • S2031 Acquire a set of music scores corresponding to the participles included in the first text segment.
  • At least one participle included in the first text segment can be obtained by the word segmentation processing of S201, and the set of the music score segments corresponding to the at least one word segment included in the first text segment can be determined through S202.
  • S2032 Determine, according to the adjacent relationship of the participles included in the first text segment, a degree of smoothness between the selected musical score segments in the set of musical score segments corresponding to the participles included in the first text segment.
  • a score segment is selected from a set of scores corresponding to each participle.
  • the first segment includes a plurality of participles, and corresponding scores may be selected, and the score segment may be used as a set of scores, for example,
  • a text segment includes three word segments, a segment word 1, a segment word 2, and a segment word 3.
  • the word segment 1 corresponds to a soundtrack segment set, and the soundtrack segment set includes three soundtrack segments; the segment word 2 corresponds to a soundtrack segment set in the soundtrack segment.
  • the set includes a soundtrack segment; the segmentation 3 corresponds to a soundtrack segment set, and the soundtrack segment set includes two soundtrack segments, and a soundtrack segment is selected from the soundtrack segment set corresponding to each segmentation word, which can be used as a group
  • six sets of music pieces can be selected for the first piece of text.
  • whether the set of soundtracks is appropriate or not can be determined based on the smoothness between the scores.
  • the fluency can reflect the comfort between the scores, the higher the fluency, the better the comfort between the scores, that is, the more beautiful the generated tunes, can give users a better sense of hearing.
  • the degree of smoothness between the score segments refers to the degree of smoothness between a plurality of (at least two) soundtrack segments having adjacent relationships.
  • the adjacent relationship between the score segments may be determined according to the adjacent relationship between the word segments in the first text segment, for example, two adjacent word segments in the first text segment, the word segment 1 and the segment word 2, and the segment word 1 corresponds to The soundtrack segment has a neighboring relationship with the score segment corresponding to the word segment 2.
  • the level of fluency it can be expressed by the degree of splicing between the score segments.
  • the music principle can be used as a basis for judging the degree of splicing between two music score segments.
  • the degree of splicing can be understood as the degree of convergence of the scores of two adjacent participles. The higher the degree of convergence, the more graceful the melody between the two parts.
  • songs and noise the most essential difference between the two is that the song has a beautiful melody that can give people an auditory enjoyment, and the poor melody of the noise makes people feel harsh. Therefore, the degree of stitching of the scores in the song is one of the main factors affecting the melody of the song.
  • the better the stitching degree of the two scores the more the music of the two scores is spliced together with the music principle, the more beautiful the melody, the more The better the hearing effect is.
  • the degree of stitching between the scores can be calculated based on the structure of the music, such as melody, rhythm, harmony, alignment, polyphony, music, instrumental method, orchestration method.
  • the degree of splicing between the soundtrack segments can be calculated by the cyclic neural network model, so that the stitching score corresponding to the stitching degree can be obtained efficiently and accurately.
  • the two word segments can be referred to as the first word segment and the second segment word respectively.
  • the method for calculating the stitching score between the two word segments is as follows:
  • Sa calculating a degree of splicing between the soundtrack segment in the music score segment corresponding to the first segment of the first segment and the music segment in the score segment corresponding to the second segment, and obtaining the first segment and the second segment At least one stitching score between.
  • a participle can correspond to at least one score segment, and the scores of two adjacent segments can form a plurality of combinations, each combination corresponding to one stitch score, and how many combinations can be combined to obtain how many stitch scores. For example, in the first text segment, two adjacent participles “I” and “Love”, “I” has two soundtrack segments, “Love” has three soundtrack segments, and the two adjacent segmentation soundtrack segments have six combinations. In the same way, six stitching scores can be obtained.
  • the specific calculation method of the stitching score is as follows:
  • a degree of splicing between the two score segments is calculated, and a stitch score determined by the two score segments between the first word segment and the second word segment is obtained.
  • all the stitching scores corresponding to the two adjacent word segments can be calculated.
  • the participle "I” and the participle “love” are two adjacent participles, the participle “I” has a corresponding score set A including the score sections a1, a2 and a3, and the participle "love” has a corresponding score set.
  • B includes the music scores b1 and b2.
  • the splicing scores of the six combinations can be obtained by the cyclic neural network model. For: 3, 5, 8, 4, 7, and 2.
  • the first participle and the second participle are the adjacent two participles of the first text segment, and the first text segment may include a plurality of participles, each of the participles having a participle adjacent thereto, referring to the first participle and the first participle
  • the calculation method of the splicing score of the binary word can calculate the splicing score of other adjacent word segments in the first text segment, thereby obtaining the splicing score corresponding to all adjacent word segments in the text segment.
  • a text segment in the lyrics of the music to be scored "I love Beijing Tiananmen” can be divided into four sub-words "I”, “Love”, “Beijing” and “Tiananmen”, with three groups of adjacent participles: "I” And “love”, “love” and “Beijing”, “Beijing” and “Tiananmen”, according to the above method of calculating the stitching score, can calculate at least one stitch score corresponding to "I” and “Love”, “Love” At least one stitch score corresponding to "Beijing”, at least one stitch score corresponding to "Beijing” and "Tiananmen".
  • Sb Calculating at least one splicing total score of the first text segment according to a splicing score of adjacent word segments in the first text segment.
  • the splicing total score calculated by the splicing score can be used to indicate the degree of splicing between the score segments corresponding to each segment of a piece of text, and the better the splicing degree, the higher the splicing total score.
  • the calculation method of the total score of the stitching can be various, and the total score of the stitching can be calculated by adding the stitching scores, or the stitching total score can be calculated by multiplying the stitching scores, and the stitching scores can be added according to a certain weight ratio. Or multiply to calculate the total score.
  • the calculation method can be selected according to the specific requirements of the lyrics score.
  • the splicing total score is calculated by adding the splicing scores.
  • a text segment "I love Beijing Tiananmen” is divided into four sub-words "I”, “Love”, “Beijing", “Tiananmen”, and the participle “I” has a corresponding score segment A including the soundtrack segment a1.
  • A2 and a3 the participle “love” has a corresponding score set B including the score sections b1 and b2, the participle "Beijing” has a corresponding score set C including the score sections c1, c2 and c3, and the participle "Tiananmen” has a correspondence
  • the soundtrack segment set D includes a soundtrack segment d1.
  • "I” and “love” are two adjacent participles.
  • the soundtrack segments corresponding to all the word segments can constitute 18 combinations, and correspondingly, 18 stitching total scores can be calculated.
  • the splicing score of a1b1 is 6 points
  • the splicing score of b1c3 is 4 points
  • the splicing score of c3d1 is 5 points. If the total score of splicing is determined by adding way , the splicing score of the group of scores is 15 points.
  • the splicing total score corresponding to all the soundtrack segment combinations of the text segment can be calculated.
  • S2033 A set of soundtracks whose fluency meets a preset condition is used as a sub-track of the first text segment.
  • the preset condition may be a condition for judging the level of fluency, and the preset condition may be preset, and may be a relevant condition when the fluency is measured according to the music principle.
  • the fluency between the soundtrack segments in a set of music scores satisfies the preset condition, the fluency is high enough that the set of scores can be used as a sub-track of the first text segment.
  • At least one splicing total score of the first text segment may be calculated, and correspondingly, the preset condition may be a preset threshold. Specifically, a set of music scores corresponding to a spliced total score may be selected from the spliced total score exceeding the preset threshold as a sub-track of the first text segment.
  • a text segment may include a plurality of word segments, each word segment may correspond to a plurality of music score segments, and selecting one soundtrack segment from each word segment may constitute a group of soundtrack segments.
  • a set of music scores corresponds to a total score of a stitch, and a text segment may have multiple soundtrack segments, that is, a text segment may include multiple stitching total scores.
  • a set of soundtrack clips needs to be selected from the plurality of sets of scores, and the selected set of scores can be used as the sub-track corresponding to the text segment.
  • the sub-track of a text segment can be selected based on the total score of the stitching. Since a text segment generally includes a plurality of splicing total scores, it can be selected by setting a preset threshold.
  • a splicing total score exceeding a preset threshold is selected, and then a suitable splicing total score is selected from the splicing total score exceeding the preset threshold, and the splicing total score corresponds to
  • a set of soundtracks can be used as a sub-score of the text segment.
  • the preset threshold may be a fixed value, and the splicing total score exceeding the preset threshold is selected, that is, the splicing total score of the splicing total score exceeds the fixed value.
  • the total score of a text segment is 1, 3, 2, 9, 7, 5, 4, 8, and 6.
  • the preset threshold is 5, and the total scores exceeding the preset threshold are 6, 7, respectively. 8,9.
  • a set of soundtracks can be arbitrarily selected from the four groups for stitching out the sub-tracks of the text segment, or a group of soundtracks with the highest stitching total score can be selected from the four groups for stitching out the sub-tracks of the text segment. .
  • the timing for setting the preset threshold can be variously selected, for example, it can be set before the lyrics are scored, or after the splicing total score is calculated, before the splicing total score is selected.
  • the first character segment is a character segment arbitrarily selected from the plurality of character segments, and it can be seen that the method for generating the sub-track of the remaining character segments is the same as the method for generating the first character segment, and details are not described herein again.
  • S204 Generate a music of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the sub-track corresponding to all the text segments included in the music lyrics can be calculated by S203, and each of the text segments has a fixed position in the lyrics, so that the sub-tracks corresponding to all the text segments in the lyrics can be followed.
  • the position of the text segment is spliced to generate a complete soundtrack, which is the soundtrack of the lyrics to be scored.
  • the above content details the lyrics generation method of the lyrics. Since the scores corresponding to the participles are pre-stored in the music library, the scores corresponding to the participles can be matched after the lyrics of the lyrics are processed, so that the matched scores and participles can be utilized. The position in the lyrics of the music genre automatically generates the soundtrack of the lyrics to be played, which effectively improves the efficiency of the lyrics.
  • the basis for establishing the soundtrack library may be historical music, or it may be a music score that may be set in advance by a music principle.
  • the present invention will explain the manner in which a music library is created by historical music.
  • FIG. 3 is a schematic flowchart of a process for establishing a music library according to an embodiment of the present invention, where the establishing process includes:
  • S301 Acquire a historical song, the historical song including historical lyrics, historical soundtrack and song type.
  • Historical songs can be songs that currently exist, including full lyrics and soundtracks. This historical song can be stored in the cloud or server, and can be obtained from the cloud or server when these historical songs are needed. In order to facilitate subsequent processing of historical songs, the historical songs may include historical lyrics, historical soundtracks, and song genres.
  • the song type can be divided according to the melody of the song, the content of the lyrics, the rhythm of the soundtrack, and the like.
  • the song types can be roughly divided into ten categories: R&B, rap, rock, jazz, country music, new century music, classical pop, folk song, disco, British rock. (Trip-Hop).
  • the types of songs that can be divided according to the melody of the song include lyrics, sadness, and the like.
  • the word segmentation of the lyrics may be the same as the method of obtaining a plurality of word segments in S201, and details are not described herein again.
  • the position of the participle in the lyrics may be determined first, and the position may be identified by a time range corresponding to the participle on the timeline of the lyric, and then the part corresponding to the position is intercepted from the soundtrack of the lyric according to the position.
  • a soundtrack as a score segment corresponding to a participle.
  • the participle "Tiananmen” is in the 50th second of the song playing, and the time range corresponding to the participle can be 45 seconds - 55 seconds, which can be intercepted in the soundtrack for 45 seconds - 55 seconds.
  • the soundtrack as a soundtrack for the participle.
  • a participle may appear multiple times in a song, and each time it appears, it can intercept a corresponding score. Therefore, in a song, a participle can correspond to multiple scores. In addition, the same participle may appear in different songs.
  • the score segment corresponding to each participle may not only come from one song, but also may come from other songs. Therefore, the more historical songs are acquired, the more word segments are obtained, and the more the scores corresponding to the word segments, the more complete the word segmentation and the corresponding soundtrack segments.
  • S304 The music library is established according to the first correspondence between the obtained word segment and the music score in the historical soundtrack, and the second correspondence between the music score and the song type in the historical soundtrack.
  • the score segment corresponding to the word segmentation can be obtained by S303, and the correspondence between the obtained segmentation word and the music score segment in the historical soundtrack can be referred to as a first correspondence relationship.
  • each score has its own historical song. According to the type of song included in a historical song, the score of the score from this historical song can be determined.
  • the correspondence between the song types of the historical songs is different from the first correspondence relationship described above, and the correspondence between the scores of the historical soundtracks and the song types may be referred to as a second correspondence.
  • a music library may be established according to the first correspondence between the obtained word segment and the music score in the historical soundtrack, and the second correspondence between the music score and the song type in the historical soundtrack. Once the score library is created, it can be saved in the soundtrack server.
  • the soundtrack segment can be initially screened, and then the appropriate soundtrack segment is selected from the selected soundtrack segments.
  • the initial screening of the scores can be based on the degree of association between the lyrics and the song types.
  • the type of the song to which the to-be-matched music lyrics belongs may be determined first, and then the soundtrack segment corresponding to the song type to which the to-be-matched music lyric belongs belongs is selected from the music library according to the second correspondence stored in the music library. .
  • the music score segment corresponding to the plurality of word segments is determined from the soundtrack segments corresponding to the song type to which the music to be played.
  • the song type of the music lyrics to be played may be preset, or may be determined according to the content of the lyrics to be played, or may be determined according to a type library established in the soundtrack server, and the type library may include word breakers belonging to different song types. Possibility information. Specifically, the type library may be searched according to a plurality of participles included in the lyrics to be scored, and the type of the song to which the to-be-matched lyrics belongs may be determined, that is, the type of the song to which the to-be-matched lyrics belongs is determined according to the possibility information in the type library.
  • the accuracy of the determined song type directly affects the quality of the generated soundtrack.
  • the more accurate the song type the better the quality of the generated soundtrack, so that the final generated song can give the user a better hearing effect.
  • the Naive Bayesian model is one of the most widely used classification models. It originated from classical mathematical theory and has a solid mathematical foundation and stable classification efficiency. Therefore, in the embodiment of the present invention, the type library may specifically determine the type of song to which the music to be played belongs according to the naive Bayesian model.
  • the naive Bayesian model can be trained by the type of songs corresponding to the participle and the participle to improve the accuracy of judging the type of songs to which the participle belongs.
  • the soundtrack of the song type can be used to generate a soundtrack for the to-be-scoring participle.
  • the soundtracks belonging to this song type can better reflect the characteristics of this song type, so the soundtrack generated by this song can also conform to the style that the to-be-matched music lyrics are intended to embody. Further, the screening of the scores by the song type can effectively reduce the scores used for the soundtrack of the to-be-matched lyrics, thereby reducing the computational difficulty of the automatic score.
  • the song type can be determined first based on the type library.
  • FIG. 4 is a schematic flowchart of a process of establishing a type library according to an embodiment of the present invention, where the process includes:
  • S401 Acquire a historical song, the historical song including historical lyrics, historical soundtrack and song type.
  • the S401-S402 is the same as the S301-S302 in FIG. 3, and details are not described herein again.
  • S403 Counting the number of occurrences of the third participle in the word segmentation process, and the correspondence between the third participle and the song type of the belonging historical song, thereby determining the number of occurrences of the third participle in the same song type.
  • the third participle is a participle obtained from the historical lyrics.
  • the word segmentation of historical songs is a word segmentation of historical lyrics in units of each song, so that the number of occurrences of word segmentation in each song can be obtained.
  • the third participle as an example, after the historical lyrics are segmented, the number of occurrences of the third participle in the historical song can be counted, and the correspondence between the third participle and the song type can be counted by the type of the song included in the historical song. Thereby, the number of occurrences of the third participle in the same song type can be determined.
  • the third participle appears 10 times in the first historical song, 20 times in the second historical song, 5 times in the third historical song, and 15 times in the fourth historical song, each song There are corresponding song types, the first historical song and the second historical song belong to the RAP type, the third historical song belongs to the rock music type, and the fourth song belongs to the jazz type, so that the third participle can be derived in the RAP. 30 times in the type, 5 times in the rock type, and 15 times in the jazz type.
  • the third participle may be a participle arbitrarily selected from a plurality of participles included in the historical lyrics, wherein the third participle is distinguished from the first participle and the second participle mentioned above for the name, and There are no other definitions such as order.
  • S404 Determine likelihood information that the third participle belongs to different song types according to the number of occurrences of the third participle in the word segmentation process and the number of occurrences of the third participle in the same song type.
  • Each historical song may include historical lyrics and song types, and the word segmentation included in each song corresponds to the song type of the song, and the same participle may be from different historical songs, that is, the same participle may correspond to different Song type.
  • the likelihood information may be information indicating the degree of association between the word segmentation and the song type, and may be stored in the form of a numerical value, wherein the numerical value may be a percentage.
  • the 15 historical songs to build a type library 10 of the 15 songs belong to the RAP type, and the remaining 5 songs belong to the rock music type.
  • the participle "I” appears 100 times in the 15 songs, among them, 70 times in the RAP type of historical songs, 30 times in the rock music type of historical songs, you can get the possibility that the participle "I” belongs to the RAP type is 70%, and the participle "I” belongs to the rock music type.
  • Sexual information is 30%.
  • S405 The type library is established according to possibility information that the word segmentation obtained in the historical lyrics belongs to different song types.
  • the third participle is one of the selected participles of all the participles included in the historical song.
  • the processing of the remaining participles is the same as the processing of the third participle.
  • S403 and S404 it is possible to determine the possibility that all the participles in the historical song belong to different song types.
  • Sexual information whereby a type library for storing possibility information that the word segmentation obtained in the historical lyrics belongs to different song types can be established. Once the type library is created, it can be saved in the soundtrack server.
  • the establishment of the type library enables the soundtrack server to directly determine the type of the song to which the music to be played belongs according to the possibility information that the word segmentation stored in the type library belongs to different song types before the preliminary screening is performed, which simplifies the determination of the song type.
  • the steps further improve the efficiency of the automatic soundtrack.
  • the to-be-matched lyrics described in the above embodiments may have two levels of word segmentation and text segmentation in terms of composition form or division granularity, and the soundtrack generation method provided by the embodiment of the present invention can be applied to such a to-be-sorted music.
  • the lyrics or the granularity of the division of the lyrics for the music can also be applied to the lyrics to be composed with other compositions, or to other divisional granularities of the lyrics to be dealt with.
  • FIG. 5 is a flowchart of a method for generating a soundtrack of a lyric according to an embodiment of the present invention, which is applied to a soundtrack server, where the soundtrack server includes a music library, and the soundtrack library includes a first correspondence between a segment and a soundtrack segment, wherein Any one of the segments has a set of soundtrack segments including at least one soundtrack segment.
  • the soundtrack server includes a music library
  • the soundtrack library includes a first correspondence between a segment and a soundtrack segment, wherein Any one of the segments has a set of soundtrack segments including at least one soundtrack segment.
  • the method includes:
  • S501 Acquire at least one segment obtained by performing word segmentation processing on the music lyrics.
  • At least one segment can be obtained through the to-be-matched lyrics, and the segments described herein are related to the division granularity, and may also be related to the composition form.
  • the segment when the partitioning granularity is fine, the segment may be in the form of a word segmentation, and when the partitioning granularity is coarse, the segment may be in the form of a text segment.
  • the segment when there are fewer punctuation in the lyrics to be scored, and there is no clear paragraph relationship, the segment may be in the form of a text segment.
  • the segment may be in the form of word segmentation.
  • the segmentation of the lyrics to be divided into genres may be adjusted according to the requirements of the scene, the calculation accuracy, and the like, which is not limited by the present invention.
  • the segment into which the lyrics to be classified is divided is also in the same form as the segment stored in the music library in the soundtrack server, so that the segment and the first correspondence can be matched to the soundtrack segment set.
  • the clip stored in the score library is in the form of a participle, and the lyrics to be scored need to be processed by word segmentation to obtain at least one participle.
  • the segment stored in the music library includes both the form of a word segment and the form of a text segment
  • the segment obtained by the word segmentation of the music score may be related to the above-mentioned division granularity or lyric composition form.
  • S502 Searching the music library according to the at least one segment, and determining, by the first correspondence, a soundtrack segment set corresponding to the at least one segment.
  • the divided segments are the same as the segments stored in the music library, it is possible to determine, by the first correspondence, the respective music score segment sets corresponding to each of the divided segments.
  • S503 Determine, according to the to-be-matched music lyrics, a soundtrack segment corresponding to each of the at least one segment from the set of soundtrack segments respectively corresponding to the at least one segment.
  • S504 Generate a soundtrack of the to-be-matched lyrics by stitching the determined soundtrack segments.
  • the music score segment corresponding to each segment can be determined from the soundtrack segment set by the composition relationship of each segment in the music score lyrics, and the specific determination manner and the stitching manner can be based on the segment.
  • the specific form varies.
  • the at least one segment can be utilized from the music library.
  • Determining, according to the first correspondence relationship, the set of the music scores corresponding to the at least one segment respectively, and according to the to-be-matched music lyrics by determining the soundtrack segment from the soundtrack segment set, the soundtrack corresponding to the to-be-matched music lyrics is visible, because The music score segment corresponding to the segment is pre-stored, so that the soundtrack segment corresponding to the segment can be matched after the word segmentation of the music score is processed, so that the soundtrack of the to-be-matched music lyrics can be automatically generated by using the matched soundtrack segment and the position of the segment in the to-be-matched music lyrics. Effectively improve the efficiency of the lyrics.
  • the fragment can be a participle:
  • the participle itself is roughly the length of a word or a phrase, it is unlikely that a single participle is used as the lyrics to be assigned, and generally at least a plurality of participles are required to constitute the lyrics to be played. Therefore, in the case where the segment is a word segmentation, a plurality of word segments obtained by performing word segmentation processing on the to-be-matched music lyrics can be acquired.
  • the splicing corresponding to the lyrics corresponding to the lyrics can be directly used to splicing the melody corresponding to the lyrics to be matched;
  • the sub-track of the text segment is first obtained, and then the soundtrack corresponding to the music to be composed is arranged by the sub-track according to the position of the text segment in the lyrics to be played.
  • the to-be-matched lyrics include at least one text segment, and one of the at least one text segments includes at least one word segment.
  • the child soundtrack corresponding to the at least one text segment may be determined according to the music score segment set corresponding to the plurality of word segments and the word segment included in the at least one text segment, wherein a text is respectively determined by a text
  • the soundtrack segment corresponding to the segmentation word segment respectively obtains the sub-track corresponding to the text segment.
  • the soundtrack of the to-be-matched lyrics may be generated by splicing the sub-tracks corresponding to the at least one text segment.
  • the fragment can be a text fragment:
  • the text segment can include a certain length of content, it is possible that a text segment can be used as a to-be-matched music lyric. In this case, when the music score segment corresponding to the text segment is found, the music score segment can be determined. The score segment directly serves as the soundtrack for the score to be determined.
  • the corresponding soundtrack segments may be first determined from the soundtrack segment corresponding to each of the text segments, and then determined according to the positions of the text segments in the to-be-matched music lyrics.
  • the soundtrack corresponding to the soundtrack lyrics may be first determined from the soundtrack segment corresponding to each of the text segments, and then determined according to the positions of the text segments in the to-be-matched music lyrics.
  • FIG. 6 is a flowchart of a method for acquiring a soundtrack of a lyric according to an embodiment of the present invention, where the method includes:
  • S601 Send the acquired to-be-sorted lyrics to the soundtrack server, the to-be-matched music lyrics include at least one text segment, one of the at least one text segment includes at least one word segment, and the soundtrack server includes a music library, the soundtrack
  • the library includes a first correspondence of word segmentation and score segments, wherein any one of the segments has a set of score segments including at least one of the score segments.
  • the interactive end can obtain the lyrics to be played.
  • the form of the lyrics to be played is not limited, and the lyrics to be played can be manually input or voice input into the interactive end.
  • the lyrics to be distributed can be directly sent to the soundtrack server.
  • the lyrics to be distributed are sent to the soundtrack server.
  • S602 Acquire a soundtrack corresponding to the to-be-matched music lyrics from the soundtrack server, where the soundtrack is obtained by stitching the soundtrack clips in the soundtrack clip set corresponding to the word segmentation in the to-be-matched music lyrics, the to-be-matched music lyrics
  • the set of music scores corresponding to the Chinese word segmentation is obtained by the soundtrack server searching for the music library according to the first correspondence relationship.
  • the interactive terminal may implement related processing of the soundtrack, for example, presenting the soundtrack to the user. Therefore, after the soundtrack server determines the soundtrack corresponding to the to-be-matched music lyrics, the soundtrack may be directly sent to the interactive end, or the soundtrack corresponding to the to-be-matched music lyrics is received when receiving the request for acquiring the soundtrack sent by the interactive terminal. Send to the interactive end.
  • the soundtrack corresponding to the to-be-matched music lyrics determined by the soundtrack server may have one or more soundtracks.
  • the interactive terminal may use the acquired soundtrack as a to-be-sorted soundtrack, and select from a plurality of to-be-determined soundtracks to meet the requirements.
  • the to-be-determined soundtrack is used as the soundtrack for the lyrics to be scored.
  • the plurality of to-be-sorted music corresponding to the to-be-matched music lyrics may be obtained from the soundtrack server, and the to-be-scheduled music piece carries the music-sort information; and according to the music information of the plurality of to-be-determined music points, the to-be-determined soundtrack that meets the demand is selected as The soundtrack of the lyrics to be played.
  • the music information may be information for identifying a music to be determined, and each of the pending music has its corresponding music information, and the music information may reflect the related features of the music to be determined to some extent.
  • the soundtrack information may include a score of a to-be-scheduled music piece carrying the soundtrack information and/or a soundtrack type.
  • the score can be used to indicate the degree to which the music to be determined conforms to the musical principle, and can be derived from the smoothness between the scores of the scores in the to-be-determined score. The higher the score, the more the music is to be matched with the music principle, that is, the tune of the to-be-determined soundtrack will be more pleasant and can give the user a better hearing enjoyment.
  • the soundtrack type can be used to indicate the type of song to which the music to be determined belongs, for example, the music to be determined belongs to rock music or jazz.
  • the first method may be selected based on the score included in the soundtrack information, or may be the highest scored to-be-determined soundtrack.
  • the soundtrack of the music lyrics is either a score with a higher score as the score of the lyrics to be scored, or a preset value can be set, and the score to be scored higher than the preset value is used as the score of the lyrics to be played.
  • the second method may be selected based on the type of the soundtrack included in the soundtrack information. For example, the user may select one type of pending soundtrack according to the soundtrack type of the plurality of to-be-determined soundtracks, and use the to-be-determined soundtrack as the to-be-soundtrack.
  • the soundtrack of the lyrics may be selected based on the type of the soundtrack included in the soundtrack information. For example, the user may select one type of pending soundtrack according to the soundtrack type of the plurality of to-be-determined soundtracks, and use the to-be-determined soundtrack as the to-be-soundtrack.
  • the soundtrack of the lyrics may be selected based on the type of the soundtrack included in the soundtrack information. For example, the user may select one type of pending soundtrack according to the soundtrack type of the plurality of to-be-determined soundtracks, and use the to-be-determined soundtrack as the to-be-soundtrack.
  • the soundtrack of the lyrics may be selected based on the type of the soundtrack included in the soundtrack information. For example, the user may select one type of pending soundtrack according to
  • the third mode may be selected based on the type of the score and the score included in the soundtrack information. For example, if there are multiple scores of the same score of the same score type, the highest score among the plurality of to-be-determined scores may be taken as the wait. If there is more than one soundtrack of the music lyrics, or if there are multiple scores of the same score, the user can select one type of pending soundtrack as the soundtrack of the music to be played according to the soundtrack type of the plurality of pending soundtracks.
  • the soundtrack server needs to be re-determined.
  • the interactive terminal may send feedback information to the soundtrack server, so that the server regenerates the soundtrack of the to-be-matched music lyrics according to the feedback information.
  • the feedback information may be information indicating that the soundtrack does not meet the demand.
  • information describing the demand may be included in the feedback information, for example, the user wants to obtain a rock-like soundtrack, and the demand is a rock-like soundtrack, and the feedback information may be Carry this information.
  • the soundtrack corresponding to the music lyrics determined by the soundtrack server can be more in line with the user's needs, and the accuracy of the soundtrack for the music to be played is further improved.
  • FIG. 7 is a structural diagram of a device for generating a soundtrack of a lyric according to an embodiment of the present invention, which is applied to a soundtrack server, where the soundtrack server includes a music library, and the soundtrack library includes a first correspondence between a word segmentation and a soundtrack segment, wherein Any one of the word segments has a set of soundtrack segments including at least one soundtrack segment, and the apparatus includes an obtaining unit 701, a searching unit 702, a determining unit 703, and a generating unit 704:
  • the acquiring unit 701 is configured to acquire a plurality of participles obtained by performing word segmentation processing on the music lyrics, wherein the to-be-matched music lyrics include at least one text segment, and one of the at least one text segment includes at least one word segment;
  • the searching unit 702 is configured to search the music library according to the plurality of word segments, and determine, by using the first correspondence relationship, a soundtrack segment set corresponding to each word segment;
  • the determining unit 703 is configured to determine, according to the music score segment set corresponding to the plurality of word segments and the word segment included in the at least one text segment, the sub-tracks corresponding to the at least one text segment respectively;
  • the generating unit 704 is configured to generate a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the first text segment is one of the at least one text segment
  • the determining unit is specifically configured to acquire a music score segment set corresponding to the word segment included in the first text segment; and according to the word segment included in the first text segment a neighboring relationship, determining a degree of smoothness between the selected soundtrack segments from the set of scores corresponding to the participles included in the first text segment; and a set of soundtrack segments satisfying the preset condition as the first text segment Sub-track.
  • the first participle and the second participle are adjacent participles in the first piece of text
  • the determining unit comprises a calculating subunit, obtaining a subunit, and as a subunit:
  • the calculating sub-unit is configured to calculate a splicing degree between the soundtrack segment in the soundtrack segment set corresponding to the first segment of the first segment and the score segment in the score segment corresponding to the second segment segment, to obtain the At least one stitching score between a participle and a second participle;
  • the obtaining subunit is configured to calculate, according to the splicing score of the adjacent word segment in the first text segment, at least one splicing total score of the first text segment;
  • the subunit is configured to select a group of music scores corresponding to a spliced total score from a spliced total score exceeding a preset threshold as a sub-track of the first text segment.
  • the calculating sub-unit is specifically configured to acquire one of the soundtrack segments in the set of music scores corresponding to the first word segment, and obtain one of the soundtrack segments in the set of music score segments corresponding to the second word segment; The degree of splicing between the score segments yields a stitch score determined by the two score segments between the first segment and the second segment.
  • the soundtrack server further includes a type library, wherein the type library includes possibility information that the word segment belongs to different song types, and the music library further includes a second correspondence between the soundtrack segment and the song type;
  • the device also includes determining a song type unit:
  • the determining a song type unit configured to search the type library according to the plurality of word segments, and determine a song type to which the to-be-matched music lyric belongs;
  • the lookup unit includes a screening subunit and a determining subunit:
  • the screening subunit is configured to filter, according to the second correspondence relationship, a soundtrack segment corresponding to a song type to which the to-be-matched music lyrics belongs according to the music library;
  • the determining subunit is configured to determine, according to the first correspondence, a set of music scores corresponding to the plurality of word segments from the soundtrack segments corresponding to the song type to which the music to be played.
  • the device further includes: acquiring a historical song unit, a word segmentation unit, a statistics unit, a determining information unit, and an establishing unit:
  • the acquiring a historical song unit for acquiring a historical song, the historical song including a historical lyric, a historical soundtrack, and a song type;
  • the word segmentation unit is configured to segment the historical lyrics
  • the statistical unit is configured to count the number of occurrences of the third participle in the word segmentation process, and the correspondence between the third participle and the song type of the belonging historical song, thereby determining the third participle in the same song type
  • the number of occurrences, the third participle is a participle obtained from the historical lyrics
  • the determining information unit is configured to determine, according to the number of occurrences of the third word segmentation in the word segmentation process and the number of occurrences of the third word segment in the same song type, the possibility that the third word segment belongs to different song types information;
  • the establishing unit is configured to establish the type library according to the possibility information that the word segmentation obtained in the historical lyrics belongs to different song types.
  • the device further includes: acquiring a historical song unit, a word segmentation unit, determining a soundtrack segment unit, and establishing the unit:
  • the acquiring a historical song unit for acquiring a historical song, the historical song including a historical lyric, a historical soundtrack, and a song type;
  • the word segmentation unit is configured to segment the historical lyrics
  • the establishing unit is configured to establish the music library according to the first correspondence between the obtained word segment and the music score in the historical soundtrack, and the second correspondence between the music score and the song type in the historical soundtrack.
  • the obtaining unit is specifically configured to perform word segmentation on the acquired lyrics to be scored to obtain a plurality of word segments.
  • the soundtrack server includes a soundtrack library including a first correspondence between the word segmentation and the soundtrack segment. After acquiring a plurality of word segments obtained by the word segmentation processing of the music scores, the plurality of word segments can be utilized from the soundtrack library. Determining, according to the first correspondence relationship, a set of soundtrack segments corresponding to the plurality of word segments, and then combining the text segment of the word segmentation in a text segment with a text segment of the word segmentation in a text segment, and combining the children corresponding to the segment of the text segment.
  • the soundtrack, and the sub-track of the text segment is spliced into the soundtrack of the lyrics to be matched.
  • the soundtrack segment corresponding to the word segmentation can be matched after the word segmentation of the music score is processed, so that the matching can be utilized.
  • the soundtrack segment and the position of the participle in the lyrics to be played automatically generate the soundtrack of the lyrics to be played, which effectively improves the efficiency of the lyrics.
  • FIG. 8 is a structural diagram of a device for generating a soundtrack of a lyric according to an embodiment of the present invention, which is applied to a soundtrack server, where the soundtrack server includes a music library, and the soundtrack library includes a first correspondence between a segment and a soundtrack segment, wherein Any one of the segments has a set of soundtrack segments including at least one soundtrack segment, and the apparatus includes an obtaining unit 801, a searching unit 802, a determining unit 803, and a splicing unit 804:
  • the obtaining unit 801 is configured to acquire at least one segment obtained by performing word segmentation processing on the music lyrics to be processed;
  • the searching unit 802 is configured to search the music library according to the at least one segment, and determine, by using the first correspondence, a soundtrack segment set respectively corresponding to the at least one segment;
  • the determining unit 803 is configured to determine, according to the to-be-matched music lyrics, a soundtrack segment corresponding to each of the at least one segment from the set of soundtrack segments respectively corresponding to the at least one segment;
  • the splicing unit 804 is configured to generate a soundtrack of the to-be-matched lyrics by splicing the determined soundtrack segments.
  • the acquiring unit is specifically configured to acquire a plurality of word segments obtained by performing word segmentation processing on the to-be-matched music lyrics.
  • the segment is a text segment
  • the to-be-matched music lyric includes at least one text segment
  • the determining unit is configured to determine, according to the music score segment corresponding to the plurality of word segments and the word segment included in the at least one text segment, a sub-track corresponding to the at least one text segment, where The sub-score corresponding to the text segment is obtained from the soundtrack segment corresponding to the word segment included in the text segment;
  • the splicing unit is specifically configured to generate a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively.
  • the at least one segment can be utilized from the music library.
  • Determining, according to the first correspondence relationship, the set of the music scores corresponding to the at least one segment respectively, and according to the to-be-matched music lyrics by determining the soundtrack segment from the soundtrack segment set, the soundtrack corresponding to the to-be-matched music lyrics is visible, because The music score segment corresponding to the segment is pre-stored, so that the soundtrack segment corresponding to the segment can be matched after the word segmentation of the music score is processed, so that the soundtrack of the to-be-matched music lyrics can be automatically generated by using the matched soundtrack segment and the position of the segment in the to-be-matched music lyrics. Effectively improve the efficiency of the lyrics.
  • FIG. 9 is a structural diagram of a device for acquiring a soundtrack of a lyric according to an embodiment of the present invention, which is applied to an interactive terminal, where the device includes a sending unit 901 and an obtaining unit 902:
  • the sending unit 901 is configured to send the acquired to-be-matched music lyrics to the soundtrack server, the to-be-matched music lyrics include at least one text segment, and one of the at least one text segment includes at least one word segment, and the soundtrack server includes a music library, the music library comprising a first correspondence of word segmentation and music scores, wherein any one of the word segments has a soundtrack segment set including at least one soundtrack segment;
  • the obtaining unit 902 is configured to acquire, from the soundtrack server, a soundtrack corresponding to the to-be-matched music lyrics, where the soundtrack is obtained by stitching the soundtrack segments in the soundtrack segment corresponding to the word segmentation in the to-be-matched music lyrics. And the set of the music scores corresponding to the word segmentation in the to-be-matched music lyrics is obtained by the soundtrack server searching the music library according to the first correspondence relationship.
  • the obtaining unit includes an obtaining subunit and a selecting subunit:
  • the obtaining subunit is configured to acquire, from the soundtrack server, a plurality of to-be-sorted music corresponding to the to-be-matched music lyrics, where the to-be-scheduled musictrack carries the music score information;
  • the selection sub-unit is configured to select, according to the music information of the plurality of to-be-determined music scores, a to-be-sorted music that meets the requirement as a soundtrack of the to-be-matched music lyrics.
  • the soundtrack information includes a score of a to-be-scheduled music piece carrying the music score information and/or a soundtrack type.
  • the device further includes a feedback unit:
  • the feedback unit is triggered, and the feedback unit is configured to send feedback information to the soundtrack server, so that the server regenerates the soundtrack of the to-be-matched music lyrics according to the feedback information.
  • the feedback information includes information describing the requirement.
  • the interaction end can realize the interaction with the soundtrack server and the interaction with the user, and the feedback of the interaction end can make the soundtrack corresponding to the music lyrics determined by the soundtrack server more in line with the user's needs, and further improve the lyrics for the music to be played.
  • the accuracy of the soundtrack can be realized.
  • FIG. 10 is a system structural diagram of a soundtrack system for a lyric according to an embodiment of the present invention.
  • the soundtrack system includes a soundtrack server 1001 and an interactive terminal 1002:
  • the soundtrack server 1001 includes a music library including a first correspondence of word segmentation and music scores, wherein any one of the word segments has a soundtrack segment set including at least one soundtrack segment, and the soundtrack server 1001 is configured to obtain a plurality of word segments obtained by word segmentation processing, the to-be-matched music lyrics including at least one text segment, one of the at least one text segment including at least one word segment; searching the music library according to the plurality of word segments Determining, by the first correspondence, a set of scores corresponding to each of the word segments; determining the at least one text according to the set of scores corresponding to the plurality of word segments and the participles included in the at least one text segment a sub-track corresponding to the segments; generating a soundtrack of the to-be-matched lyrics by splicing the sub-tracks corresponding to the at least one text segment respectively;
  • the interaction end 1002 is configured to send the acquired to-be-matched music lyrics to the soundtrack server; and acquire a soundtrack corresponding to the to-be-matched music lyrics from the soundtrack server.
  • the music editor can obtain the lyrics to be played, and can also edit the soundtrack corresponding to the lyrics to be played by the soundtrack server. .
  • the music editor may include an input interface having an lyric for inputting the to-be-matched music, and an editing interface for displaying the soundtrack.
  • the input interface is used to obtain the lyrics to be played, and the acquired lyrics to be played may be lyrics input in the input interface or lyrics obtained after recognizing the input speech.
  • the lyrics to be played can be manually input in the input interface, or imported from other text editing tools and texts to be distributed through the input interface.
  • the smart device configured with the music editor has a voice input recognition function
  • the user can play the audio content having the to-be-matched music lyrics or directly speak the to-be-matched music lyrics, and the smart device can identify the recorded audio content.
  • the to-be-matched music lyrics are recognized, and the input interface can acquire the recognized to-be-matched music lyrics. The rich way to get the lyrics to be matched improves the scope of the music editor.
  • the to-be-matched lyrics can be sent to the soundtrack server when needed.
  • the user can generate an output instruction through the function key set on the music editor to instruct the music editor to send the to-be-sorted lyrics to the soundtrack server.
  • the music editor is configured to send, to the soundtrack server, the to-be-matched lyrics acquired through the input interface, the soundtrack server includes a music library, and the music library includes a first correspondence between the word segmentation and the music score segment. Any one of the word segments has a set of soundtrack segments including at least one soundtrack segment.
  • the music editor is configured to display the soundtrack corresponding to the to-be-matched music lyrics in the editing interface, and the soundtrack corresponding to the to-be-matched music lyrics is used as the soundtrack corresponding to the to-be-matched music lyrics.
  • the soundtrack server is spliced according to the music scores in the set of music scores corresponding to the word segmentation in the music lyrics to be scored, and the soundtrack segment corresponding to the word segmentation in the music to be lyrics is used by the soundtrack server to search the music library according to the first correspondence relationship. owned.
  • the editing interface By displaying the soundtrack, the editing interface allows the user to observe the specific information of the soundtrack, such as notes, beats, timelines, etc., and can be used as a basis for editing the soundtrack. It should be noted that the editing interface can display other soundtracks in addition to the soundtrack sent by the soundtrack server, for example, by identifying the soundtrack of the recorded audio data.
  • the editing interface has an editing button for setting the displayed soundtrack to be editable when a trigger for the editing button is received.
  • the soundtrack can be deleted, added, modified and other editing operations to ultimately meet the user's needs.
  • the music editor can obtain the music to be played through the input interface, and can output the music to be played to the soundtrack server.
  • the soundtrack can be displayed in the editing interface. To make the soundtrack editable and improve the user experience.
  • the music player can acquire the music to be played and its corresponding soundtrack, and mix the two into a song for playing.
  • the music player has a hybrid interface and a playback interface.
  • the mixing interface is configured to mix the music to be played and the soundtrack corresponding to the music to be played to obtain a song
  • the soundtrack corresponding to the music to be played is a soundtrack server according to the music score segment corresponding to the word segmentation in the to-be-matched music lyrics.
  • the soundtrack segments are stitched together, and the soundtrack segment set corresponding to the word segmentation in the to-be-matched music lyrics is obtained by the soundtrack server searching for the music library according to the first correspondence relationship, and the soundtrack server includes a music library, and the music library includes word segmentation and soundtrack segments.
  • the first correspondence relationship wherein any one of the word segments has a set of soundtrack segments including at least one soundtrack segment.
  • the play interface is used to play the song.
  • the mixed interface can mix the pre-acquired lyrics to be matched with the corresponding soundtrack, and the manner of mixing is not limited in the present invention.
  • the lyrics are the lyrics of the to-be-matched music
  • the music is the song of the soundtrack.
  • the soundtrack can be played only, and the lyrics corresponding to the time axis of the played soundtrack (the lyrics corresponding to the time axis in the lyrics of the music to be played) can be displayed at the relevant position, or the sound can be simultaneously outputted while playing the soundtrack.
  • the analog sound of the soundtrack lyrics Therefore, the user can see or hear whether the corresponding lyrics match the soundtrack while playing the soundtrack, thereby improving the user experience.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Library & Information Science (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Machine Translation (AREA)
  • Management Or Editing Of Information On Record Carriers (AREA)

Abstract

一种歌词的配乐生成方法和相关装置,配乐服务器中包括分词(100)和配乐片段之间第一对应关系的配乐库(200),当获取通过对待配乐歌词进行分词处理所获取的多个分词(100)后,可以利用多个分词(100)从配乐库(200)中根据第一对应关系确定出多个分词(100)分别对应的配乐片段集合(300),之后以待配乐歌词的文字片段为单元,利用一个文字片段中分词(100)对应的配乐片段集合(300),组合出一个文字片段所对应的子配乐(400),并将文字片段的子配乐(400)拼接成该待配乐歌词的配乐,由于预存了分词(100)对应的配乐片段,故可以在对待配乐歌词分词后匹配到分词(100)对应的配乐片段,从而可以利用匹配到的配乐片段以及分词(100)在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。

Description

一种歌词的配乐生成方法和相关装置
本申请要求2016年12月30日递交的申请号为201611264888.5、发明名称为“一种歌词的配乐生成方法和相关装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及数据处理领域,特别是涉及一种歌词的配乐生成方法和相关装置。
背景技术
在音乐创作中,很多情况是先创作出歌词,再根据歌词进行谱曲,即为歌词完成相应的配乐,从而形成一首歌曲。
如何能够为歌词快速的完成配乐是一个亟需解决的问题,如果能够自动化的根据歌词的内容进行配乐将大大提高音乐创作的效率。
然而,目前并没有一种可以根据歌词的内容自动为该歌词进行配乐的方式。
发明内容
为了解决上述技术问题,本发明提供了一种歌词的配乐生成方法和相关装置,以利用匹配到的配乐片段以及分词在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
本发明实施例公开了如下技术方案:
第一方面,本发明提供了一种歌词的配乐生成方法,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述方法包括:
获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;
根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;
根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;
通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
可选的,第一文字片段为所述至少一个文字片段中的一个,所述根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,包括:
获取所述第一文字片段所包括分词对应的配乐片段集合;
根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度;
将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
可选的,第一分词与第二分词为所述第一文字片段中的相邻分词,所述根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度,包括:
计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分;
根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分;
所述将流畅程度满足预设阈值的一组配乐片段作为所述第一文字片段的子配乐,包括:
从超过预设阈值的拼接总分中选取一个拼接总分所对应的一组配乐片段作为所述第一文字片段的子配乐。
可选的,所述计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分,包括:
获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;
计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
可选的,所述配乐服务器还包括类型库,所述类型库中包括分词属于不同歌曲类型的可能性信息,所述配乐库还包括配乐片段与歌曲类型之间的第二对应关系;在所述根据所述多个分词查找所述配乐库,确定出所述多个分词对应的配乐片段集合之前,还包括:
根据所述多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型;
所述根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出所述多个分词对应的配乐片段集合,包括:
根据所述第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段;
根据所述第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
可选的,还包括:
获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
对所述历史歌词进行分词;
统计在分词过程中第三分词的出现次数,以及所述第三分词与所属历史歌曲的歌曲类型的对应关系,从而确定出在同一个歌曲类型中所述第三分词出现的次数,所述第三分词为从所述历史歌词中得到的一个分词;
根据所述第三分词在分词过程中的出现次数和在同一个歌曲类型中所述第三分词出现的次数确定出所述第三分词属于不同歌曲类型的可能性信息;
根据所述历史歌词中得到的分词属于不同歌曲类型的可能性信息建立所述类型库。
可选的,还包括:
获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
对所述历史歌词进行分词;
根据得到的分词确定在所述历史配乐中对应的配乐片段;
根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
可选的,所述获取通过对待配乐歌词进行分词处理所获取的多个分词,包括:
对获取的待配乐歌词进行分词处理,得到多个分词。
第二方面,本发明提供了一种歌词的配乐生成装置,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元、查找单元、确定单元和生成单元:
所述获取单元,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一 个分词;
所述查找单元,用于根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;
所述确定单元,用于根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;
所述生成单元,用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
可选的,第一文字片段为所述至少一个文字片段中的一个,所述确定单元具体用于获取所述第一文字片段所包括分词对应的配乐片段集合;根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度;将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
可选的,第一分词与第二分词为所述第一文字片段中的相邻分词,所述确定单元包括计算子单元、得到子单元和作为子单元:
所述计算子单元,用于计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分;
所述得到子单元,用于根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分;
所述作为子单元,用于从超过预设阈值的拼接总分中选取一个拼接总分所对应的一组配乐片段作为所述第一文字片段的子配乐。
可选的,所述计算子单元具体用于获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
可选的,所述配乐服务器还包括类型库,所述类型库中包括分词属于不同歌曲类型的可能性信息,所述配乐库还包括配乐片段与歌曲类型之间的第二对应关系;所述装置还包括确定歌曲类型单元:
所述确定歌曲类型单元,用于根据所述多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型;
所述查找单元包括筛选子单元和确定子单元:
所述筛选子单元,用于根据所述第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段;
所述确定子单元,用于根据所述第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
可选的,所述装置还包括获取历史歌曲单元、分词单元、统计单元、确定信息单元和建立单元:
所述获取历史歌曲单元,用于获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
所述分词单元,用于对所述历史歌词进行分词;
所述统计单元,用于统计在分词过程中第三分词的出现次数,以及所述第三分词与所属历史歌曲的歌曲类型的对应关系,从而确定出在同一个歌曲类型中所述第三分词出现的次数,所述第三分词为从所述历史歌词中得到的一个分词;
所述确定信息单元,用于根据所述第三分词在分词过程中的出现次数和在同一个歌曲类型中所述第三分词出现的次数确定出所述第三分词属于不同歌曲类型的可能性信息;
所述建立单元,用于根据所述历史歌词中得到的分词属于不同歌曲类型的可能性信息建立所述类型库。
可选的,所述装置还包括获取历史歌曲单元、分词单元、确定配乐片段单元和建立单元:
所述获取历史歌曲单元,用于获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
所述分词单元,用于对所述历史歌词进行分词;
所述确定配乐片段单元,用于根据得到的分词确定在所述历史配乐中对应的配乐片段;
所述建立单元,用于根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
可选的,所述获取单元具体用于对获取的待配乐歌词进行分词处理,得到多个分词。
第三方面,本发明提供了一种歌词的配乐生成方法,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片 段具有一个包括至少一个配乐片段的配乐片段集合,所述方法包括:
获取通过对待配乐歌词进行分词处理所得到的至少一个片段;
根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合;
根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段;
通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
可选的,所述片段为分词,所述获取通过对待配乐歌词进行分词处理所得到的至少一个片段,包括:
获取通过对所述待配乐歌词进行分词处理所得到的多个分词。
可选的,所述片段为文字片段,所述待配乐歌词包括至少一个文字片段。
可选的,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,则所述根据所述待配乐歌词,从所述至少一个分词分别对应的配乐片段集合中确定出所述至少一个分词中每个分词分别对应的配乐片段,包括:
根据所述多个分词分别对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐;
所述通过拼接确定出的配乐片段生成所述待配乐歌词的配乐,包括:
通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
第四方面,本发明提供了一种歌词的配乐生成装置,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片段具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元、查找单元、确定单元和拼接单元:
所述获取单元,用于获取通过对待配乐歌词进行分词处理所得到的至少一个片段;
所述查找单元,用于根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合;
所述确定单元,用于根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段;
所述拼接单元,用于通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
可选的,所述获取单元具体用于获取通过对所述待配乐歌词进行分词处理所得到的 多个分词。
可选的,所述片段为文字片段,所述待配乐歌词包括至少一个文字片段。
可选的,所述确定单元具体用于根据所述多个分词分别对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐;
所述拼接单元具体用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
第五方面,本发明提供了一种歌词的配乐获取方法,应用于交互端,所述方法包括:
向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
可选的,所述从所述配乐服务器获取所述待配乐歌词对应的配乐,包括:
从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;
根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
可选的,所述配乐信息包括携带所述配乐信息的待定配乐的评分和/或配乐类型。
可选的,在所述从所述配乐服务器获取所述待配乐歌词对应的配乐之后,还包括:
若获取的配乐不符合需求,向所述配乐服务器发送反馈信息,以使得所述服务器根据所述反馈信息重新生成所述待配乐歌词的配乐。
可选的,所述反馈信息中包括描述所述需求的信息。
第六方面,本发明提供了一种歌词的配乐获取装置,应用于交互端,所述装置包括发送单元和获取单元:
所述发送单元,用于向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括 至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
所述获取单元,用于从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
可选的,所述获取单元包括获取子单元和选择子单元:
所述获取子单元,用于从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;
所述选择子单元,用于根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
可选的,所述配乐信息包括携带所述配乐信息的待定配乐的评分和/或配乐类型。
可选的,所述装置还包括反馈单元:
若获取的配乐不符合需求,触发所述反馈单元,所述反馈单元,用于向所述配乐服务器发送反馈信息,以使得所述服务器根据所述反馈信息重新生成所述待配乐歌词的配乐。
可选的,所述反馈信息中包括描述所述需求的信息。
第七方面,本发明提供了一种歌词的配乐系统,所述配乐系统包括配乐服务器和交互端:
所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述配乐服务器,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐;
所述交互端,用于向所述配乐服务器发送所获取的所述待配乐歌词;从所述配乐服务器获取所述待配乐歌词对应的配乐。
第八方面,本发明提供了一种音乐编辑器,所述音乐编辑器具有用于获取所输入待配乐歌词的输入界面,以及用于展示配乐的编辑界面;
所述输入界面用于获取待配乐歌词,所获取的待配乐歌词为在所述输入界面中输入的歌词,或者为对输入的语音进行识别后得到的歌词;
当获取输出指令时,所述音乐编辑器用于向配乐服务器发送通过所述输入界面获取的待配乐歌词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
当从所述配乐服务器获取所述待配乐歌词对应的配乐时,所述音乐编辑器用于将所述待配乐歌词对应的配乐展示在所述编辑界面中,所述待配乐歌词对应的配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找所述配乐库得到的。
可选的,所述编辑界面具有编辑按钮,当接收到对所述编辑按钮的触发时,所述编辑界面用于将所展示的配乐设置为可编辑的;
所述编辑界面所展示的配乐还包括通过对所录制音频数据识别得到的配乐。
第九方面,本发明提供了一种音乐播放器,所述音乐播放器具有混合界面和播放界面;
所述混合界面用于将待配乐歌词以及所述待配乐歌词对应的配乐进行混合得到歌曲,所述待配乐歌词对应的配乐为配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找配乐库得到的,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
所述播放界面用于播放所述歌曲。
由上述技术方案可以看出,配乐服务器中预设有包括分词和配乐片段之间第一对应关系的配乐库,当获取通过对待配乐歌词进行分词处理所获取的多个分词后,可以利用该多个分词从配乐库中根据第一对应关系确定出该多个分词分别对应的配乐片段集合,之后以待配乐歌词的文字片段为单元,利用一个文字片段中分词对应的配乐片段集合,组合出一个文字片段所对应的子配乐,并将文字片段的子配乐拼接成该待配乐歌词的配乐,可见,由于预存了分词对应的配乐片段,故可以在对待配乐歌词分词后匹配到分词 对应的配乐片段,从而可以利用匹配到的配乐片段以及分词在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本发明的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1为本发明实施例提供的一种歌词的配乐生成的系统示意图;
图2为本发明实施例提供的一种歌词的配乐生成方法的方法流程图;
图2a为本发明实施例提供的一种确定第一文字片段对应的子配乐的方法的流程图。
图3为本发明实施例提供的一种配乐库建立过程的流程示意图;
图4为本发明实施例提供的一种类型库建立过程的流程示意图;
图5为本发明实施例提供的另一种歌词的配乐生成方法的方法流程图;
图6为本发明实施例提供的一种歌词的配乐获取方法的方法流程图;
图7为本发明实施例提供的一种歌词的配乐生成装置的装置结构图;
图8为本发明实施例提供的另一种歌词的配乐生成装置的装置结构图;
图9为本发明实施例提供的一种歌词的配乐获取装置的装置结构图;
图10为本发明实施例提供的一种歌词的配乐系统的系统结构图。
具体实施方式
为使本发明实施例的目的、技术方案和优点更加清楚,下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚地描述,显然,所描述的实施例是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
在音乐创作的过程中,一首歌曲的完成一般是先词后曲,即先创作出歌词,再根据歌词进行谱曲(配乐),最终形成一首歌曲。而对歌词进行配乐往往需要耗费大量的时间。随着深度学习在语音领域的发展,越来越多的任务能够通过机器自动化地完成,如果能够自动化的根据歌词的内容进行配乐将极大提高音乐创作的效率。
为此,本发明实施例提供一种歌词的配乐生成方法,配乐服务器获取通过对待配乐 歌词进行分词处理所获取的多个分词,依据预先保存的分词和配乐片段的对应关系,匹配到分词对应的配乐片段,从而可以利用匹配到的配乐片段以及分词在待配乐歌词中的位置自动生成出待配乐歌词的配乐。
基于上述思想,本发明实施例可以通过配乐服务器来实现。可以参见图1所示,配乐服务器中包括配乐库200,其中配乐库200中保存有分词和配乐片段的对应关系。以第一文字片段100为例,配乐服务器对第一文字片段100进行分词处理后,可以得到第一文字片段100包括的分词,如第一分词、第二分词和第三分词。由配乐库200中保存的分词与配乐片段的对应关系,可以确定出分词对应的配乐片段集合300。如图1所示,第一分词与配乐片段集合A相对应,第二分词与配乐片段集合B相对应,第三分词与配乐片段集合C相对应。每个配乐片段集合中包括至少一个配乐片段,如配乐片段集合A可以包括配乐片段a1、a2和a3,配乐片段集合B可以包括配乐片段b1和b2,配乐片段集合C可以包括配乐片段c1、c2和c3。配乐服务器从每个分词对应的配乐片段集合中选取合适的一个配乐片段,作为第一文字片段的子配乐400,如可以选取配乐片段a1-b2-c3作为该文字片段的子配乐400。
在本发明实施例中,一首歌词可以理解为用于形成一首完整歌曲的歌词,一首歌词中可以包括至少一个文字片段,一个文字片段可以是词语结构范畴的一个段落,也可以是以标点符合作为划分的一句文字或者一段文字。一个文字片段中可以由至少一个分词组成。第一文字片段即该至少一个文字片段中的一个文字片段。本发明并不限定对歌词进行分词的具体实现方式,可以通过词义或词语结构等从歌词中划分出具有独立词义或词语结构的分词,所划分出的分词可以是一个字,也可以是一个词。例如一个文字片段“我爱北京天安门”可以划分出的分词包括“我”、“爱”、“北京”和“天安门”。
配乐片段可以理解为是为一个分词所确定的配乐片段,一般一个分词是处于一个歌词中的,这个歌词具有已经配好的配乐,可以先确定这个分词在这个歌词中的位置,该位置可以通过该歌词的时间线上该分词所对应的时间范围来标识。再根据该位置从该已经配好的配乐中截取该位置所对应一部分配乐,例如该配乐处于该时间范围中的一部分配乐,这一部分配乐可以是该分词所对应的配乐片段。
相应的,一首歌曲可以包括歌词和该歌词对应的配乐,该歌词中的一个文字片段具有这个文字片段对应的子配乐,该子配乐为这个文字片段在该配乐中所对应的一部分配乐。这个文字片段对应的子配乐可以由这个文字片段中分词对应的配乐片段组成,例如 一个未配乐的文字片段“我爱北京天安门”可以划分出的分词包括“我”、“爱”、“北京”和“天安门”。分词“我”具有对应的配乐片段集合A中包括配乐片段a1、a2和a3,分词“爱”具有对应的配乐片段集合B中包括配乐片段b1和b2,分词“北京”具有对应的配乐片段集合C中包括配乐片段c1、c2和c3,分词“天安门”具有对应的配乐片段集合D中包括配乐片段d1,那么这个文字片段“我爱北京天安门”的子配乐可以是分别由配乐片段集合A、B、C和D中的一个配乐片段所组成,例如a1-b2-c3-d1,或者a2-b1-c3-d1等。
可以将需要进行配乐的歌词称为待配乐歌词,通过上述介绍可知待配乐歌词,从组成形式或划分粒度上看,可以具有分词,文字片段这两个层级,而本发明实施例所提供的配乐生成方法可以应用于这类待配乐歌词或这种对待配乐歌词的划分粒度,还可以应用于具有其他组成形式的待配乐歌词,或应用于对待配乐歌词的其他划分粒度的情况。
接下来将以从分词对应的配乐片段,确定出文字片段对应的子配乐,从而得出待配乐歌词的配乐的方式为例,详细介绍本发明实施例所提供的歌词的配乐生成方法。图2为本发明实施例提供的一种歌词的配乐生成方法的方法流程图,该方法包括:
S201:获取通过对待配乐歌词进行分词处理所获取的多个分词。
当需要为待配乐歌词配乐时,可以将歌词手动输入或语音输入到相关的设备中。
在本发明实施例中可以通过配乐服务器对歌词进行配乐。配乐服务器中包括有配乐库,在该配乐库中预先存储有分词与配乐片段的第一对应关系,为了便于配乐服务器对待配乐歌词进行自动配乐,首先需要对待配乐歌词进行分词处理。
本发明实施例对待配乐歌词进行分词处理的具体设备不作限定,可以是由配乐服务器对待配乐歌词进行分词处理,得到多个分词。也可以是由其他设备对待配乐歌词进行分词处理,得到多个分词,配乐服务器可以从该设备获取到进行分词处理后获取的多个分词。
进行分词处理时,具体的可以采用递进的形式对待配乐歌词进行分词处理,即可以先将待配乐歌词划分为文字片段,再对文字片段进行分词处理。其中,可以以标点符号作为划分文字片段的依据,或者以段落作为划分文字片段的依据。对文字片段进行分词处理,具体的可以通过词义或词语结构等从文字片段中划分出具有独立词义或词语结构的分词。也可以直接对待配乐歌词进行分词处理,再确定分词所属的文字片段。
例如,待配乐歌词“我爱北京天安门,天安门上太阳升”,若以标点符号作为划分文字片段的依据,该歌词可以分为“我爱北京天安门”和“天安门上太阳升”这两个文 字片段。通过词义或词语结构等从歌词中划分出具有独立词义或词语结构的分词,其中对文字片段“我爱北京天安门”进行分词处理,可以得到“我”、“爱”、“北京”、“天安门”这四个分词,对文字片段“天安门上太阳升”进行分词处理,可以得到“天安门”、“上”、“太阳”、“升”这四个分词。
对于大篇幅的待配乐歌词,采用上述递进形式进行分词处理,可以使得到的分词更加准确,降低了出错的概率。
S202:根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合。
配乐服务器可以预先建立好配乐库,在该配乐库中保存有分词和配乐片段的对应关系,可以将该对应关系称为第一对应关系。在该配乐库中,一个分词可以具有对应的至少一个配乐片段。
一首待配乐歌词可以划分为多个分词,由于配乐库中一个分词可以具有对应的有至少一个配乐片段,故在通过划分出的分词匹配配乐库时,一个分词可以匹配出一个或者多个配乐片段,可以将针对一个划分出的分词所匹配到的配乐片段作为这个分词对应的配乐片段集合,也即一个分词具有一个对应的配乐片段集合。例如,在一首歌曲中分词“我”匹配到三个配乐片段,分别为a1、a2和a3,可以将这三个配乐片段作为分词“我”的配乐片段集合。
以第一分词,即该多个分词中的一个分词为例,确定所述第一分词对应的配乐片段集合具体包括:根据所述第一分词查找所述配乐库,确定出所述第一分词对应的一个包括至少一个配乐片段的配乐片段集合。
对于其它分词确定对应的配乐片段集合的方法与第一分词确定对应的配乐片段集合的方法相同,在此不再赘述。
S203:根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐。
待配乐歌词包括至少一个文字片段,以第一文字片段,即该至少一个文字片段中的一个文字片段为例展开介绍。若第一文字片段中只包括一个分词,那么可以直接从这个分词中确定出一个配乐片段作为第一文字片段的子配乐即可。若第一文字片段中包括多个分词,确定所述第一文字片段对应的子配乐的方法可以如图2a所示,具体包括S2031-S2033。
S2031:获取所述第一文字片段所包括分词对应的配乐片段集合。
通过S201的分词处理可以获取到第一文字片段包括的至少一个分词,通过S202可以确定出第一文字片段包括的至少一个分词分别对应的配乐片段集合。
S2032:根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度。
当获取到第一文字片段所包括的至少一个分词分别对应的配乐片段集合后,需要从每个分词对应的配乐片段集合中选取一个合适的配乐片段作为该分词合适的配乐片段,以便将选取出的配乐片段拼接成第一文字片段的子配乐。
从每个分词对应的一个配乐片段集合中选取出一个配乐片段,第一文字片段包括有几个分词,相应的可以选取出几个配乐片段,可以将该配乐片段作为一组配乐片段,例如,第一文字片段包括有三个分词,分词1、分词2和分词3,分词1对应有一个配乐片段集合,在该配乐片段集合中包括有三个配乐片段;分词2对应有一个配乐片段集合,在该配乐片段集合中包括有一个配乐片段;分词3对应有一个配乐片段集合,在该配乐片段集合中包括有两个配乐片段,从每个分词对应的配乐片段集合中选取出一个配乐片段,可以作为一组配乐片段,对于该第一文字片段可以选取出六组配乐片段。
以一组配乐片段为例,该组配乐片段是否合适,可以依据于配乐片段之间的流畅程度判别。流畅程度可以反映出配乐片段之间的舒适度,流畅程度越高,说明配乐片段之间的舒适度越好,也即生成的曲调越优美,可以给用户带来更好的听觉享受。
在判断配乐片段之间流畅程度时,指的是具有相邻关系的多个(至少两个)配乐片段之间的流畅程度。其中,配乐片段之间的相邻关系可以依据于第一文字片段中分词之间的相邻关系确定出,例如,第一文字片段中的两个相邻分词,分词1和分词2,则分词1对应的配乐片段,与分词2对应的配乐片段具有相邻关系。
对于流畅程度的高低,具体可以通过配乐片段之间的拼接程度来表示。
在本发明实施例中,可以通过音乐原理作为评判两个配乐片段之间拼接程度的依据。拼接程度可以理解为两个相邻分词的配乐片段的衔接度,衔接度越高则表示两个分词之间的旋律越优美。例如,歌曲和噪音,这两者最本质的区别在于,歌曲具有优美的旋律,可以给人带来听觉上的享受,而噪音的旋律较差会使人觉得刺耳。故歌曲中配乐片段的拼接程度是影响歌曲旋律的主要因素之一,两个配乐片段的拼接程度越好,这两个配乐片段拼接在一起的音乐越符合音乐原理,旋律越优美,给人带来听觉效果越好。
为了便于直观的了解相邻分词的配乐片段之间的拼接程度,可以用拼接得分表示。两个配乐片段的拼接程度越好,则对应的拼接得分越高。
配乐片段之间的拼接程度可以依据于音乐的构造进行计算,例如旋律、节奏、和声、对位、复调、曲式、乐器法、配器法等因素。计算拼接程度需要考虑的因素众多,而神经网络具有大规模并行、分布式存储和处理、自组织、自适应和自学能力,特别适合处理需要同时考虑许多因素和条件的、不精确和模糊的信息处理问题,因此在本发明实施例中可以通过循环神经网络模型进行配乐片段之间拼接程度的计算,从而可以高效准确的得出拼接程度对应的拼接得分。此外,还可以用配乐和歌词建立以歌词为主键的倒排索引,利用倒排索引来训练循环神经网络模型,以提升模型的性能。
以第一文字片段中任意两个相邻的分词为例,可以将这两个分词分别称为第一分词和第二分词,计算这两个分词之间的拼接得分的方法如下:
Sa:计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分。
一个分词可以对应至少一个配乐片段,相邻两个分词的配乐片段可以构成多种组合,每个组合对应有一个拼接得分,有多少种组合对应的就可以得出多少个拼接得分。例如,第一文字片段中两个相邻分词“我”和“爱”,“我”具有两个配乐片段,“爱”具有三个配乐片段,这两个相邻分词的配乐片段有六种组合方式,对应的可以得出六个拼接得分。
以一个拼接得分为例,该拼接得分的具体计算方法如下:
获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;
计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
依据上述方法可以计算出相邻两个分词对应的所有拼接得分。例如,分词“我”和分词“爱”为两个相邻的分词,分词“我”具有对应的配乐片段集合A中包括配乐片段a1、a2和a3,分词“爱”具有对应的配乐片段集合B中包括配乐片段b1和b2,这两个分词的配乐片段有六种组合:a1b1、a1b2、a2b1、a2b2、a3b1和a3b2,通过循环神经网络模型可以得出这六种组合对应的拼接得分分别为:3、5、8、4、7和2。
上述第一分词和第二分词为第一文字片段中任意选取的相邻的两个分词,第一文字片段可以包括多个分词,每个分词都有与其相邻的分词,参照上述第一分词和第二分词的拼接得分的计算方法,可以计算出第一文字片段中其它相邻分词的拼接得分,从而得 出该文字片段中所有相邻分词对应的拼接得分。
例如,待配乐歌词中的一个文字片段“我爱北京天安门”可以划分为“我”、“爱”、“北京”和“天安门”这四个分词,具有三组相邻的分词:“我”和“爱”、“爱”和“北京”、“北京”和“天安门”,依据上述计算拼接得分的方法,可以分别计算出“我”和“爱”对应的至少一个拼接得分,“爱”和“北京”对应的至少一个拼接得分,“北京”和“天安门”对应的至少一个拼接得分。
Sb:根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分。
由拼接得分计算出的拼接总分,可以用于表示一个文字片段由每一个分词所对应的配乐片段之间的拼接程度,拼接程度越好则相应的拼接总分越高。
其中,拼接总分的计算方法可以有多种,可以通过拼接得分相加计算出拼接总分,也可以通过拼接得分相乘计算出拼接总分,还可以将拼接得分按照一定的权重比例相加或相乘计算出拼接总分。计算方法可以依据歌词配乐的具体要求进行选择,在本发明实施例中,采用拼接得分相加的方法计算出拼接总分。
例如,一个文字片段“我爱北京天安门”划分为“我”、“爱”、“北京”、“天安门”这四个分词,分词“我”具有对应的配乐片段集合A中包括配乐片段a1、a2和a3,分词“爱”具有对应的配乐片段集合B中包括配乐片段b1和b2,分词“北京”具有对应的配乐片段集合C中包括配乐片段c1、c2和c3,分词“天安门”具有对应的配乐片段集合D中包括配乐片段d1。“我”和“爱”为两个相邻的分词,这两个分词的配乐片段组合有六种:a1b1、a1b2、a2b1、a2b2、a3b1和a3b2,对应的拼接得分分别为6、4、3、7、2、5;“爱”和“北京”为两个相邻的分词,这两个分词的配乐片段组合有六种:b1c1、b2c1、b1c2、b2c2、b1c3和b2c3,对应的拼接得分分别为6、2、3、1、4、5;“北京”和“天安门”为两个相邻的分词,这两个分词的配乐片段组合有三种:c1d1、c2d1、c3d1,对应的拼接得分分别为4、2、5。该文字片段中,所有分词对应的配乐片段可以构成18种组合,对应的可以计算出18个拼接总分。当一组配乐片段为a1-b1-c3-d1,其中a1b1的拼接得分为6分,b1c3的拼接得分为4分,c3d1的拼接得分为5分,若以相加的方式确定拼接总分的话,则该组配乐片段的拼接总分为15分。
通过上述方法可以计算出该文字片段所有配乐片段组合对应的拼接总分。拼接总分越高,则说明该配乐片段的组合越符合音乐原理、所形成的曲调可以给听众带来更好的听觉效果。
S2033:将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
预设条件可以是用于评判流畅程度高低的条件,预设条件可以预先设置,具体可以是根据音乐原理来衡量流畅程度时的相关条件。当一组配乐片段中配乐片段之间的流畅程度满足预设条件时,说明流畅程度已经足够高,高到可以将该组配乐片段作为第一文字片段的子配乐。
针对上述以拼接程度来表示流畅程度的情况,可以计算得到第一文字片段的至少一个拼接总分,与此相对应,该预设条件可以是一个预设阈值。具体的,可以从超过预设阈值的拼接总分中选取一个拼接总分所对应的一组配乐片段作为所述第一文字片段的子配乐。
一个文字片段可以包括多个分词,每个分词可以对应多个配乐片段,从每个分词中选取一个配乐片段可以构成一组配乐片段,
一组配乐片段对应一个拼接总分,一个文字片段中可以有多组配乐片段,即一个文字片段中可以包括多个拼接总分。要完成对该文字片段的配乐,需要从多组配乐片段中选取出一组配乐片段,可以将选取出的一组配乐片段作为该文字片段对应的子配乐。
一个文字片段的子配乐可以依据于拼接总分进行选取。由于一个文字片段一般包括多个拼接总分,因此可以采用设置预设阈值的方式进行选取。
具体的,从该文字片段对应的所有拼接总分中,选取出超过预设阈值的拼接总分,再从超过预设阈值的拼接总分中选取合适的一个拼接总分,该拼接总分对应的一组配乐片段即可作为该文字片段的子配乐。设置预设阈值,一方面可以缩小选取的范围,另一方面也可以将不符合要求的配乐片段组合排除。
其中,预设阈值可以是一个固定数值,选取出超过预设阈值的拼接总分即选取拼接总分的分值超过该固定数值的拼接总分。例如,一个文字片段的拼接总分分别为1、3、2、9、7、5、4、8、6,预设阈值为5,则超过预设阈值的拼接总分分别为6、7、8、9。可以从这四组中任意选择一组配乐片段用于拼接出这个文字片段的子配乐,也可以从这四组中选择拼接总分最高的一组配乐片段用于拼接出这个文字片段的子配乐。
设置预设阈值的时机可以有多种选择,例如可以在为歌词配乐之前就预先设置,也可以在计算出拼接总分后,在选取拼接总分之前设置。
上述第一文字片段是所述多个文字片段中任意选取出的一个文字片段,由此可知,其余文字片段的子配乐的生成方法与上述第一文字片段的生成方法相同,在此不再赘述。
S204:通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配 乐。
由S203可以计算出待配乐歌词中包括的所有文字片段对应的子配乐,每一个文字片段在该歌词中都有其固定的位置,因此,可以将该歌词中所有的文字片段对应的子配乐按照文字片段所在的位置进行拼接,从而生成一首完整的配乐,即待配乐歌词的配乐。
上述内容详细介绍了歌词的配乐生成方法,由于配乐库中预存了分词对应的配乐片段,故可以在对待配乐歌词分词后匹配到分词对应的配乐片段,从而可以利用匹配到的配乐片段以及分词在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
接下来,将对配乐库的建立过程进行详细介绍,建立配乐库的依据可以是历史音乐,也可以是通过音乐原理预先设置好一个分词可以具有的配乐可能。可选的,本发明将对通过历史音乐建立配乐库的方式进行说明。
如图3所示,为本发明实施例提供的一种配乐库建立过程的流程示意图,该建立过程包括:
S301:获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型。
历史歌曲可以是目前已经存在的包括完整歌词和配乐的歌曲。该历史歌曲可以存储在云端或服务器中,当需要使用这些历史歌曲时,可以从云端或服务器获取。为了便于对历史歌曲进行后续处理,该历史歌曲中可以包括历史歌词、历史配乐和歌曲类型。
歌曲类型可以根据歌曲的旋律、歌词的内容、配乐的节奏等进行划分。例如依据歌曲的发展历程,歌曲类型大致可以分为十类:节奏怨曲(R&B)、说唱乐(RAP)、摇滚乐、爵士乐、乡村音乐、新世纪音乐、古典流行乐、民歌、迪斯科、英伦摇滚(Trip-Hop)。或者,根据歌曲的旋律可以划分出的歌曲类型包括抒情、伤感等。
S302:对所述历史歌词进行分词。
对歌词进行分词可以与S201中得到多个分词的方法相同,在此不再赘述。
S303:根据得到的分词确定在所述历史配乐中对应的配乐片段。
具体的,可以先确定这个分词在歌词中的位置,该位置可以通过该歌词的时间线上该分词所对应的时间范围来标识,再根据该位置从该歌词的配乐中截取该位置所对应一部分配乐,作为分词对应的配乐片段。
例如,在一首歌曲中,依照时间顺序,分词“天安门”处于歌曲播放的第50秒,该分词所对应的时间范围可以是45秒-55秒,可以在配乐中截取45秒-55秒对应的配乐, 作为该分词的配乐片段。
一个分词在一首歌曲中可能会多次出现,每次出现都可以截取到对应的一个配乐片段,因此,在一首歌曲中,一个分词可以对应有多个配乐片段。此外,不同的歌曲中,也可能会出现相同的分词。
由此可知,每个分词对应的配乐片段可能不仅来自于一首歌曲,还可能来自于其它歌曲。因此,获取的历史歌曲越多,得到的分词就越多,并且分词对应的配乐片段也越广泛,从而使分词与对应的配乐片段更加完善。
S304:根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
由S303可以得到分词对应的配乐片段,可以将得到的分词与所述历史配乐中配乐片段的对应关系称为第一对应关系。
由于配乐片段都是从历史歌曲中划分得到,故每个配乐片段都有其所属的历史歌曲,依据一个历史歌曲中包括的歌曲类型,可以确定出从这个历史歌曲中划分得到的配乐片段与这个历史歌曲的歌曲类型之间的对应关系,区别于上述第一对应关系,可以将所述历史配乐中配乐片段与歌曲类型的对应关系称为第二对应关系。
根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系可以建立出配乐库。配乐库建立完成后,可以保存在配乐服务器中。
在建立配乐库的过程中,为了使配乐库更加完善,往往会对大量的历史歌曲进行分析,从而使每一个分词对应的配乐片段尽可能的多,但是在为歌词配乐时,配乐片段越多,选取每个分词合适的配乐片段的计算难度就越大。因此,在为分词选取合适的配乐片段时,可以先对配乐片段进行初步筛选,再从筛选后的配乐片段中选取合适的配乐片段。
考虑到歌词与歌曲类型往往存在紧密的联系,因此可以依据歌词包括的分词与歌曲类型的关联程度进行配乐片段的初步筛选。
具体的,可以先确定所述待配乐歌词所属的歌曲类型,再根据配乐库中保存的第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段。
经过初步筛选后,再根据配乐库中保存的第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
待配乐歌词的歌曲类型可以是预先设置的,也可以是根据待配乐歌词的内容确定的,也可以依据于配乐服务器中建立的类型库来确定,该类型库中可以包括分词属于不同歌曲类型的可能性信息。具体的,可以根据待配乐歌词包括的多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型,即依据类型库中的可能性信息,确定出待配乐歌词所属的歌曲类型。
确定出的歌曲类型的是否准确直接影响后续生成的配乐质量的好坏,歌曲类型越准确,则后续生成的配乐质量越好,从而使得最终生成的歌曲可以给用户带来更好的听觉效果。朴素贝叶斯模型是目前使用最广泛的分类模型之一,其发源于古典数学理论,有着坚实的数学基础,以及稳定的分类效率。因此,在本发明实施例中该类型库具体可以通过朴素贝叶斯模型来确定待配乐歌词所属的歌曲类型。此外,还可以通过分词与分词对应的歌曲类型训练朴素贝叶斯模型,提升其对分词所属歌曲类型判断的准确性。
若确定出待配乐歌词属于一个歌曲类型,那么可以通过这个歌曲类型的配乐片段为该待配乐分词生成配乐。属于这个歌曲类型的配乐片段更能够体现这个歌曲类型的特点,故以此生成的配乐也更能够符合该待配乐歌词所想体现的风格。进一步的,通过歌曲类型对配乐片段的筛选,可以有效减少用于为该待配乐歌词进行配乐的配乐片段,从而减少了自动配乐的计算难度。
进行上述初步筛选之前可以依据于类型库先确定出歌曲类型。
接下来,将对类型库的建立过程进行详细介绍,如图4所示,为本发明实施例提供的一种类型库建立过程的流程示意图,该建立过程包括:
S401:获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型。
S402:对所述历史歌词进行分词。
S401-S402与图3中S301-S302执行过程相同,在此不再赘述。
S403:统计在分词过程中第三分词的出现次数,以及所述第三分词与所属历史歌曲的歌曲类型的对应关系,从而确定出在同一个歌曲类型中所述第三分词出现的次数,所述第三分词为从所述历史歌词中得到的一个分词。
对历史歌曲进行分词,是以每首歌曲为单位对历史歌词进行分词处理,因此可以得出每首歌曲中分词出现的次数。以第三分词为例,对历史歌词分词后,可以统计出第三分词在历史歌曲中出现的次数,并且由历史歌曲中包括的歌曲类型,可以统计出第三分词与歌曲类型的对应关系,从而可以确定出在同一个歌曲类型中所述第三分词出现的次 数。例如,第三分词在第一首历史歌曲中出现10次,在第二首历史歌曲中出现20次,第三首历史歌曲中出现5次,第四首历史歌曲中出现15次,每首歌曲都有其对应的歌曲类型,第一首历史歌曲和第二首历史歌曲同属于RAP类型,第三首历史歌曲属于摇滚乐型,第四首歌曲属于爵士乐类型,从而可以得出第三分词在RAP类型中出现30次,在摇滚乐型中出现5次,在爵士乐类型中出现15次。
需要说明的是,上述第三分词可以是从历史歌词包括的多个分词中任意选取的一个分词,其中,第三分词是为了名称与上述提到的第一分词和第二分词相区分,并无先后顺序等其它意义上的限定。
S404:根据所述第三分词在分词过程中的出现次数和在同一个歌曲类型中所述第三分词出现的次数确定出所述第三分词属于不同歌曲类型的可能性信息。
每首历史歌曲中可以包括历史歌词和歌曲类型,每首歌曲中包括的分词与该歌曲的歌曲类型相对应,相同的一个分词可以来自于不同的历史歌曲,即相同的一个分词可以对应不同的歌曲类型。
可能性信息可以是表示分词与歌曲类型的关联程度的信息,可以采用数值的形式进行存储,其中数值可以采用百分数。
例如,根据15首历史歌曲建立类型库,这15首歌曲中有10首历史歌曲属于RAP类型,剩余5首属于摇滚乐型,分词“我”在这15首歌曲中一共出现了100次,其中,在RAP类型的历史歌曲中出现70次,在摇滚乐型的历史歌曲中出现30次,则可以得出分词“我”属于RAP类型的可能性信息为70%,分词“我”属于摇滚乐型的可能性信息为30%。
S405:根据所述历史歌词中得到的分词属于不同歌曲类型的可能性信息建立所述类型库。
第三分词是历史歌曲包括的所有分词中任选出的一个分词,其余分词的处理过程与第三分词的处理过程相同,通过S403和S404可以确定出历史歌曲中所有分词属于不同歌曲类型的可能性信息,由此可以建立出保存所述历史歌词中得到的分词属于不同歌曲类型的可能性信息的类型库。类型库建立完成后,可以保存在配乐服务器中。
类型库的建立,使得在进行上述初步筛选之前,配乐服务器可以依据于类型库中保存的分词属于不同歌曲类型的可能性信息,直接确定出待配乐歌词所属的歌曲类型,简化了确定歌曲类型的步骤,进一步提升了自动配乐的效率。
以上实施例中所说明的待配乐歌词,从组成形式或划分粒度上看,可以具有分词,文字片段这两个层级,而本发明实施例所提供的配乐生成方法除了可以应用于这类待配乐歌词或这种对待配乐歌词的划分粒度,还可以应用于具有其他组成形式的待配乐歌词,或应用于对待配乐歌词的其他划分粒度的情况。接下来,将针对本发明实施例如何应用于可能组成形式的待配乐歌词或者可能的划分粒度进行说明。
图5为本发明实施例提供的一种歌词的配乐生成方法的方法流程图,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片段具有一个包括至少一个配乐片段的配乐片段集合,这里关于配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
所述方法包括:
S501:获取通过对待配乐歌词进行分词处理所得到的至少一个片段。
根据分词处理的划分粒度,或者待配乐歌词的组成形式,可以通过待配乐歌词得到至少一个片段,这里所述的片段与划分粒度相关,也可以与组成形式相关。例如,在划分粒度较细时,该片段可以是分词的形式,在划分粒度较粗时,该片段可以是文字片段的形式。例如当待配乐歌词中具有较少标点,且没有明确的段落关系时,该片段可以是文字片段的形式,当待配乐歌词中标点较多,且段落划分清晰时,该片段可以是分词的形式。具体将待配乐歌词划分为何种形式的片段,可以根据场景需求,计算精度等相应调整,本发明对此并不限定。
需要注意的是,将待配乐歌词划分为何种形式的片段还与配乐服务器中配乐库内所存储的片段的形式相同,由此才能通过划分出的片段以及第一对应关系匹配到配乐片段集合。例如配乐库中所存储的片段为分词的形式,那待配乐歌词需要通过分词处理得到至少一个分词。若配乐库中存储的片段既包括分词的形式,也包括文字片段的形式,那对待配乐歌词分词得到的片段可以与上述划分粒度或歌词组成形式相关。
S502:根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合。
由于划分出的片段与配乐库中所保存的片段相同,故可以实现通过第一对应关系确定每个所划分片段各自对应的配乐片段集合。
S503:根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段。
S504:通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
由于配乐片段集合中包括至少一个配乐片段,故可以通过待配乐歌词中各片段的组成关系,从配乐片段集合中确定出每个片段分别对应的配乐片段,具体的确定方式和拼接方式可以根据片段的具体形式的不同而有所不同。
可见,通过在配乐服务器中预设有包括分词和配乐片段之间第一对应关系的配乐库,当对获取的待配乐歌词进行分词以得到至少一个片段后,可以利用该至少一个片段从配乐库中根据第一对应关系确定出该至少一个片段分别对应的配乐片段集合,并依据该待配乐歌词,将通过从配乐片段集合中确定出配乐片段拼接出该待配乐歌词对应的配乐,可见,由于预存了片段对应的配乐片段,故可以在对待配乐歌词分词后匹配到片段对应的配乐片段,从而可以利用匹配到的配乐片段以及片段在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
接下来将通过片段的可能形式,进一步对本发明实施例进行说明。
第一种可能形式,片段可以是分词:
这种情况下,由于分词本身大概为一个词或是一个短语的长度,故单独一个分词作为待配乐歌词的可能性不大,一般需要至少多个分词构成待配乐歌词。故在片段为分词的情况下,可以获取通过对所述待配乐歌词进行分词处理所得到的多个分词。
在拼接该待配乐歌词的配乐过程中,可以根据分词在待配乐歌词中的位置,直接使用已经确定的分词对应配乐片段进行拼接,以拼接出待配乐歌词对应的配乐;也可以在待配乐歌词还具有文字片段这个层级的情况下,先得出文字片段的子配乐后,再根据文字片段在待配乐歌词中的位置,由子配乐拼接出待配乐歌词对应的配乐。
针对待配乐歌词还具有文字片段这个层级的情况,该待配乐歌词包括至少一个文字片段,该至少一个文字片段中一个文字片段包括至少一个分词。
在确定分词对应的配乐片段时,可以根据该多个分词分别对应的配乐片段集合以及该至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐。具体的确定方式可以参见图2所对应实施例中的相关描述,这里不再赘述。
在得到文字片段对应的子配乐后,可以通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
具体的拼接方式可以参见图2所对应实施例中的相关描述,这里不再赘述。
第二种可能形式,片段可以是文字片段:
由于文字片段可以包括一定篇幅的内容,故有可能一个文字片段可以作为一个待配 乐歌词,这种情况下,当查找到这个文字片段对应的配乐片段集合时,可以将配乐片段集合中确定出的配乐片段直接作为该待确定配乐的配乐。
当待配乐歌词划分出多个文字片段时,可以先从每个文字片段对应的配乐片段集合中确定出对应的配乐片段,再根据这些文字片段在待配乐歌词中的位置,通过拼接确定出待配乐歌词对应的配乐。
接下来从交互端详细介绍本发明实施例所提供的歌词的配乐获取方法,该交互端可以是与歌词配乐相关的智能设备,可以实现与配乐服务器的交互,以及与用户之间的交互。图6为本发明实施例提供的一种歌词的配乐获取方法的方法流程图,该方法包括:
S601:向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合。
交互端可以获取到待配乐歌词,在本发明实施例中对获取待配乐歌词的形式不做限定,可以是将待配乐歌词手动输入或语音输入到该交互端中。
交互端获取到待配乐歌词后,可以直接将该待配乐歌词发送给配乐服务器。也可以是在接收到配乐服务器发送的用于获取待配乐歌词的请求时,将该待配乐歌词发送给配乐服务器。
S602:从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
图6所对应实施例中特征的说明可以参见图1,图2所对应实施例的相关描述,这里不再一一赘述。
在本发明实施例中,交互端可以实现对配乐的相关处理,例如,向用户呈现该配乐。因此在配乐服务器确定出该待配乐歌词对应的配乐后,可以直接将该配乐发送给交互端,或者是在接收到交互端发送的用于获取配乐的请求时,将该待配乐歌词对应的配乐发送给交互端。
由配乐服务器确定出的该待配乐歌词对应的配乐可以有一个或者多个,对于具有多个配乐的情况,交互端可以将获取的该配乐作为待定配乐,从多个待定配乐中选择出符 合要求的待定配乐作为该待配乐歌词的配乐。可选的,可以从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
配乐信息可以是用于标识待定配乐的信息,每个待定配乐都有其对应的配乐信息,该配乐信息可以从一定程度上体现出该待定配乐的相关特征。具体的,配乐信息可以包括携带所述配乐信息的待定配乐的评分和/或配乐类型。
评分可以用于表示待定配乐符合音乐原理的程度,可以依据于待定配乐中配乐片段之间的流畅程度得出。评分越高,说明该待定配乐越符合音乐原理,即该待定配乐的曲调会比较悦耳,可以带给用户更好的听觉享受。
配乐类型可以用于表示待定配乐所属的歌曲类型,例如,待定配乐属于摇滚乐或者爵士乐等。
在选择符合需求的待定配乐作为所述待配乐歌词的配乐时,可以有多种方式,第一种方式可以以配乐信息中包括的评分为依据进行选择,可以是将评分最高的待定配乐作为待配乐歌词的配乐,或者是将评分较高的待定配乐作为待配乐歌词的配乐,又或者是可以设置预设数值,将评分高于预设数值的待定配乐作为待配乐歌词的配乐。
第二种方式可以以配乐信息中包括的配乐类型为依据进行选择,例如,用户可以根据该多个待定配乐的配乐类型,选择出喜欢的一种类型的待定配乐,将该待定配乐作为待配乐歌词的配乐。
第三种方式可以以配乐信息中包括的配乐类型和评分为依据进行选择,例如,对于同一配乐类型的待定配乐有多个时,则可以将该多个待定配乐中评分最高的作为所述待配乐歌词的配乐,或者是对于相同评分的待定配乐有多个时,则用户可以根据该多个待定配乐的配乐类型,选择喜欢的一种类型的待定配乐作为待配乐歌词的配乐。
考虑到可能会出现由配乐服务器确定出的待配乐歌词对应的配乐不符合需求的情况,例如,用户想要获取摇滚类型的配乐,但是配乐服务器确定出的配乐的配乐类型中没有摇滚类的配乐。对于这种情况,说明配乐服务器确定出的配乐存在问题,需要配乐服务器重新进行确定。
因此,对于获取的配乐不符合需求的情况,交互端可以向所述配乐服务器发送反馈信息,以使得所述服务器根据所述反馈信息重新生成所述待配乐歌词的配乐。
反馈信息可以是用于表示配乐不符合需求的信息。为了使得重新生成的配乐能够更好的符合需求,可以在该反馈信息中包括描述所述需求的信息,例如,用户想要获取摇 滚类的配乐,需求即为摇滚类配乐,则可以在反馈信息中携带该信息。
通过该反馈操作可以使得配乐服务器确定出的待配乐歌词对应的配乐更加符合用户的需求,进一步提升了为待配乐歌词配乐的准确性。
图7为本发明实施例提供的一种歌词的配乐生成装置的装置结构图,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元701、查找单元702、确定单元703和生成单元704:
所述获取单元701,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;
所述查找单元702,用于根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;
所述确定单元703,用于根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;
所述生成单元704,用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
可选的,第一文字片段为所述至少一个文字片段中的一个,所述确定单元具体用于获取所述第一文字片段所包括分词对应的配乐片段集合;根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度;将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
可选的,第一分词与第二分词为所述第一文字片段中的相邻分词,所述确定单元包括计算子单元、得到子单元和作为子单元:
所述计算子单元,用于计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分;
所述得到子单元,用于根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分;
所述作为子单元,用于从超过预设阈值的拼接总分中选取一个拼接总分所对应的一 组配乐片段作为所述第一文字片段的子配乐。
可选的,所述计算子单元具体用于获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
可选的,所述配乐服务器还包括类型库,所述类型库中包括分词属于不同歌曲类型的可能性信息,所述配乐库还包括配乐片段与歌曲类型之间的第二对应关系;所述装置还包括确定歌曲类型单元:
所述确定歌曲类型单元,用于根据所述多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型;
所述查找单元包括筛选子单元和确定子单元:
所述筛选子单元,用于根据所述第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段;
所述确定子单元,用于根据所述第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
可选的,所述装置还包括获取历史歌曲单元、分词单元、统计单元、确定信息单元和建立单元:
所述获取历史歌曲单元,用于获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
所述分词单元,用于对所述历史歌词进行分词;
所述统计单元,用于统计在分词过程中第三分词的出现次数,以及所述第三分词与所属历史歌曲的歌曲类型的对应关系,从而确定出在同一个歌曲类型中所述第三分词出现的次数,所述第三分词为从所述历史歌词中得到的一个分词;
所述确定信息单元,用于根据所述第三分词在分词过程中的出现次数和在同一个歌曲类型中所述第三分词出现的次数确定出所述第三分词属于不同歌曲类型的可能性信息;
所述建立单元,用于根据所述历史歌词中得到的分词属于不同歌曲类型的可能性信息建立所述类型库。
可选的,所述装置还包括获取历史歌曲单元、分词单元、确定配乐片段单元和建立单元:
所述获取历史歌曲单元,用于获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
所述分词单元,用于对所述历史歌词进行分词;
所述确定配乐片段单元,用于根据得到的分词确定在所述历史配乐中对应的配乐片段;
所述建立单元,用于根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
可选的,所述获取单元具体用于对获取的待配乐歌词进行分词处理,得到多个分词。
上述关于配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
可见,配乐服务器中预设有包括分词和配乐片段之间第一对应关系的配乐库,当获取通过对待配乐歌词进行分词处理所获取的多个分词后,可以利用该多个分词从配乐库中根据第一对应关系确定出该多个分词分别对应的配乐片段集合,之后以待配乐歌词的文字片段为单元,利用一个文字片段中分词对应的配乐片段集合,组合出一个文字片段所对应的子配乐,并将文字片段的子配乐拼接成该待配乐歌词的配乐,可见,由于预存了分词对应的配乐片段,故可以在对待配乐歌词分词后匹配到分词对应的配乐片段,从而可以利用匹配到的配乐片段以及分词在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
图8为本发明实施例提供的一种歌词的配乐生成装置的装置结构图,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片段具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元801、查找单元802、确定单元803和拼接单元804:
所述获取单元801,用于获取通过对待配乐歌词进行分词处理所得到的至少一个片段;
所述查找单元802,用于根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合;
所述确定单元803,用于根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段;
所述拼接单元804,用于通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
可选的,所述获取单元具体用于获取通过对所述待配乐歌词进行分词处理所得到的多个分词。
可选的,所述片段为文字片段,所述待配乐歌词包括至少一个文字片段。
可选的,所述确定单元具体用于根据所述多个分词分别对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐;
所述拼接单元具体用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
上述关于配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
可见,通过在配乐服务器中预设有包括分词和配乐片段之间第一对应关系的配乐库,当对获取的待配乐歌词进行分词以得到至少一个片段后,可以利用该至少一个片段从配乐库中根据第一对应关系确定出该至少一个片段分别对应的配乐片段集合,并依据该待配乐歌词,将通过从配乐片段集合中确定出配乐片段拼接出该待配乐歌词对应的配乐,可见,由于预存了片段对应的配乐片段,故可以在对待配乐歌词分词后匹配到片段对应的配乐片段,从而可以利用匹配到的配乐片段以及片段在待配乐歌词中的位置自动生成出待配乐歌词的配乐,有效提高了为歌词配乐的效率。
图9为本发明实施例提供的一种歌词的配乐获取装置的装置结构图,应用于交互端,所述装置包括发送单元901和获取单元902:
所述发送单元901,用于向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
所述获取单元902,用于从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
可选的,所述获取单元包括获取子单元和选择子单元:
所述获取子单元,用于从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;
所述选择子单元,用于根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
可选的,所述配乐信息包括携带所述配乐信息的待定配乐的评分和/或配乐类型。
可选的,所述装置还包括反馈单元:
若获取的配乐不符合需求,触发所述反馈单元,所述反馈单元,用于向所述配乐服务器发送反馈信息,以使得所述服务器根据所述反馈信息重新生成所述待配乐歌词的配乐。
可选的,所述反馈信息中包括描述所述需求的信息。
上述关于配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
可见,交互端可以实现与配乐服务器的交互,以及与用户之间的交互,通过交互端的反馈可以使得配乐服务器确定出的待配乐歌词对应的配乐更加符合用户的需求,进一步提升了为待配乐歌词配乐的准确性。
图10为本发明实施例提供的一种歌词的配乐系统的系统结构图,所述配乐系统包括配乐服务器1001和交互端1002:
所述配乐服务器1001包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述配乐服务器1001,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐;
所述交互端1002,用于向所述配乐服务器发送所获取的所述待配乐歌词;从所述配乐服务器获取所述待配乐歌词对应的配乐。
接下来,介绍一种与实施本发明实施例所提供的歌词配乐生成方式相关的音乐编辑 器,该音乐编辑器可以获取待配乐歌词,也可以对配乐服务器提供的待配乐歌词对应的配乐进行编辑。
该音乐编辑器可以包括具有用于获取所输入待配乐歌词的输入界面,以及用于展示配乐的编辑界面。
该输入界面用于获取待配乐歌词,所获取的待配乐歌词可以为在所述输入界面中输入的歌词,或者为对输入的语音进行识别后得到的歌词。
也就是说,用户在使用该音乐编辑器时,可以在该输入界面中手动输入待配乐歌词,或者从其他文字编辑工具、文本中具有的待配乐歌词通过该输入界面进行导入。当配置有该音乐编辑器的智能设备具有语音录入识别功能时,用户可以通过播放具有该待配乐歌词的音频文件或者直接说出该待配乐歌词,智能设备可以对录入的音频内容进行识别,从中识别出该待配乐歌词,而该输入界面可以获取识别出的待配乐歌词。丰富的获取待配乐歌词的方式提高了该音乐编辑器的适用范围。
输入界面在获取到待配乐歌词后,由于该音乐编辑器可以与配乐服务器相连,故可以在需要时将该待配乐歌词向配乐服务器发送。用户可以通过该音乐编辑器上设置的功能键生成输出指令,以指示该音乐编辑器向配乐服务器发送待配乐歌词。
当获取输出指令时,所述音乐编辑器用于向配乐服务器发送通过所述输入界面获取的待配乐歌词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合。
该待配乐歌词、配乐和配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
当从所述配乐服务器获取所述待配乐歌词对应的配乐时,所述音乐编辑器用于将所述待配乐歌词对应的配乐展示在所述编辑界面中,所述待配乐歌词对应的配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找所述配乐库得到的。
该编辑界面通过展示配乐,可以使得用户能够观察到该配乐的具体信息,例如音符、节拍、时间轴等,并可以以此作为编辑该配乐的基础。需要注意的是,该编辑界面除了可以展示的由配乐服务器发送的配乐,还可以展示其他配乐,例如通过对所录制音频数据识别得到的配乐。
为了能够实现对配乐的编辑,可选的,所述编辑界面具有编辑按钮,当接收到对所 述编辑按钮的触发时,所述编辑界面用于将所展示的配乐设置为可编辑的。在可编辑状态下,该配乐可以被删除、增加、修改等编辑操作,以最终达到用户的需求。
可见,通过音乐编辑器可以通过输入界面获取待配乐歌词,并可以将待配乐歌词输出至配乐服务器,当接收配乐服务器返回的针对该待配乐歌词的配乐时,可以通过在编辑界面中展示该配乐,让配乐具有可编辑的可能,提高了用户体验。
接下来,介绍一种与实施本发明实施例所提供的歌词配乐生成方式相关的音乐播放器,该音乐播放器可以获取待配乐歌词和其对应的配乐,将两者混合为歌曲进行播放。
该音乐播放器具有混合界面和播放界面。
所述混合界面用于将待配乐歌词以及所述待配乐歌词对应的配乐进行混合得到歌曲,所述待配乐歌词对应的配乐为配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找配乐库得到的,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合。
该待配乐歌词、配乐和配乐服务器的相关特征的说明可以参见图1,图2所对应实施例中的相关描述,这里不再赘述。
所述播放界面用于播放所述歌曲。
可见,该混合界面可以将预先获取的待配乐歌词和对应的配乐进行混合,混合的方式本发明并不限定,混合后得到了歌词为该待配乐歌词,音乐为该配乐的歌曲,在播放该歌曲时,可以只播放该配乐,并在相关位置展示与所播放配乐时间轴对应的歌词(待配乐歌词中与时间轴对应位置的歌词),也可以通过模拟声音,在播放配乐时同时输出待配乐歌词的模拟声音。由此,用户可以在播放配乐的同时看到或听到对应歌词与配乐是否配合,提高了用户体验。
本领域普通技术人员可以理解:实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成,前述程序可以存储于一计算机可读取存储介质中,该程序在执行时,执行包括上述方法实施例的步骤;而前述的存储介质可以是下述介质中的至少一种:只读存储器(英文:read-only memory,缩写:ROM)、RAM、磁碟或者光盘等各种可以存储程序代码的介质。
需要说明的是,本说明书中的各个实施例均采用递进的方式描述,各个实施例之间 相同相似的部分互相参见即可,每个实施例重点说明的都是与其他实施例的不同之处。尤其,对于设备及系统实施例而言,由于其基本相似于方法实施例,所以描述得比较简单,相关之处参见方法实施例的部分说明即可。以上所描述的设备及系统实施例仅仅是示意性的,其中作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性劳动的情况下,即可以理解并实施。
以上所述,仅为本发明较佳的具体实施方式,但本发明的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本发明揭露的技术范围内,可轻易想到的变化或替换,都应涵盖在本发明的保护范围之内。因此,本发明的保护范围应该以权利要求的保护范围为准。

Claims (31)

  1. 一种歌词的配乐生成方法,其特征在于,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述方法包括:
    获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;
    根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;
    根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;
    通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
  2. 根据权利要求1所述的方法,其特征在于,第一文字片段为所述至少一个文字片段中的一个,所述根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,包括:
    获取所述第一文字片段所包括分词对应的配乐片段集合;
    根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度;
    将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
  3. 根据权利要求2所述的方法,其特征在于,第一分词与第二分词为所述第一文字片段中的相邻分词,所述根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度,包括:
    计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分;
    根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分;
    所述将流畅程度满足预设阈值的一组配乐片段作为所述第一文字片段的子配乐,包括:
    从超过预设阈值的拼接总分中选取一个拼接总分所对应的一组配乐片段作为所述 第一文字片段的子配乐。
  4. 根据权利要求3所述的方法,其特征在于,所述计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分,包括:
    获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;
    计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
  5. 根据权利要求1至4任意一项所述的方法,其特征在于,所述配乐服务器还包括类型库,所述类型库中包括分词属于不同歌曲类型的可能性信息,所述配乐库还包括配乐片段与歌曲类型之间的第二对应关系;在所述根据所述多个分词查找所述配乐库,确定出所述多个分词对应的配乐片段集合之前,还包括:
    根据所述多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型;
    所述根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出所述多个分词对应的配乐片段集合,包括:
    根据所述第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段;
    根据所述第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
  6. 根据权利要求5所述的方法,其特征在于,还包括:
    获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
    对所述历史歌词进行分词;
    统计在分词过程中第三分词的出现次数,以及所述第三分词与所属历史歌曲的歌曲类型的对应关系,从而确定出在同一个歌曲类型中所述第三分词出现的次数,所述第三分词为从所述历史歌词中得到的一个分词;
    根据所述第三分词在分词过程中的出现次数和在同一个歌曲类型中所述第三分词出现的次数确定出所述第三分词属于不同歌曲类型的可能性信息;
    根据所述历史歌词中得到的分词属于不同歌曲类型的可能性信息建立所述类型库。
  7. 根据权利要求5所述的方法,其特征在于,还包括:
    获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
    对所述历史歌词进行分词;
    根据得到的分词确定在所述历史配乐中对应的配乐片段;
    根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
  8. 根据权利要求1所述的方法,其特征在于,所述获取通过对待配乐歌词进行分词处理所获取的多个分词,包括:
    对获取的待配乐歌词进行分词处理,得到多个分词。
  9. 一种歌词的配乐生成装置,其特征在于,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元、查找单元、确定单元和生成单元:
    所述获取单元,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;
    所述查找单元,用于根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;
    所述确定单元,用于根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;
    所述生成单元,用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
  10. 根据权利要求9所述的装置,其特征在于,第一文字片段为所述至少一个文字片段中的一个,所述确定单元具体用于获取所述第一文字片段所包括分词对应的配乐片段集合;根据所述第一文字片段中所包括分词的相邻关系,确定从所述第一文字片段所包括分词对应的配乐片段集合中所选取配乐片段之间的流畅程度;将流畅程度满足预设条件的一组配乐片段作为所述第一文字片段的子配乐。
  11. 根据权利要求10所述的装置,其特征在于,第一分词与第二分词为所述第一文字片段中的相邻分词,所述确定单元包括计算子单元、得到子单元和作为子单元:
    所述计算子单元,用于计算所述第一文字片段中第一分词所对应配乐片段集合中的配乐片段与第二分词所对应配乐片段集合中的配乐片段之间的拼接程度,得到所述第一分词和第二分词之间的至少一个拼接得分;
    所述得到子单元,用于根据所述第一文字片段中相邻分词的拼接得分计算得到所述第一文字片段的至少一个拼接总分;
    所述作为子单元,用于从超过预设阈值的拼接总分中选取一个拼接总分所对应的一组配乐片段作为所述第一文字片段的子配乐。
  12. 根据权利要求11所述的装置,其特征在于,所述计算子单元具体用于获取所述第一分词所对应配乐片段集合中的一个配乐片段,获取所述第二分词所对应配乐片段集合中的一个配乐片段;计算这两个配乐片段之间的拼接程度,得到所述第一分词和第二分词之间由这两个配乐片段确定的一个拼接得分。
  13. 根据权利要求9-12任意一项所述的装置,其特征在于,所述配乐服务器还包括类型库,所述类型库中包括分词属于不同歌曲类型的可能性信息,所述配乐库还包括配乐片段与歌曲类型之间的第二对应关系;所述装置还包括确定歌曲类型单元:
    所述确定歌曲类型单元,用于根据所述多个分词查找所述类型库,确定所述待配乐歌词所属的歌曲类型;
    所述查找单元包括筛选子单元和确定子单元:
    所述筛选子单元,用于根据所述第二对应关系,从所述配乐库中筛选出与所述待配乐歌词所属的歌曲类型对应的配乐片段;
    所述确定子单元,用于根据所述第一对应关系,从与所述待配乐歌词所属的歌曲类型对应的配乐片段中确定出所述多个分词对应的配乐片段集合。
  14. 根据权利要求13所述的装置,其特征在于,所述装置还包括获取历史歌曲单元、分词单元、确定配乐片段单元和建立单元:
    所述获取历史歌曲单元,用于获取历史歌曲,所述历史歌曲包括历史歌词、历史配乐和歌曲类型;
    所述分词单元,用于对所述历史歌词进行分词;
    所述确定配乐片段单元,用于根据得到的分词确定在所述历史配乐中对应的配乐片段;
    所述建立单元,用于根据得到的分词与所述历史配乐中配乐片段的第一对应关系,以及所述历史配乐中配乐片段与歌曲类型的第二对应关系建立所述配乐库。
  15. 一种歌词的配乐生成方法,其特征在于,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片段具有一个包括至少一个配乐片段的配乐片段集合,所述方法包括:
    获取通过对待配乐歌词进行分词处理所得到的至少一个片段;
    根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合;
    根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段;
    通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
  16. 根据权利要求15所述的方法,其特征在于,所述片段为分词,所述获取通过对待配乐歌词进行分词处理所得到的至少一个片段,包括:
    获取通过对所述待配乐歌词进行分词处理所得到的多个分词。
  17. 根据权利要求15所述的方法,其特征在于,所述片段为文字片段,所述待配乐歌词包括至少一个文字片段。
  18. 根据权利要求16所述的方法,其特征在于,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,则所述根据所述待配乐歌词,从所述至少一个分词分别对应的配乐片段集合中确定出所述至少一个分词中每个分词分别对应的配乐片段,包括:
    根据所述多个分词分别对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐;
    所述通过拼接确定出的配乐片段生成所述待配乐歌词的配乐,包括:
    通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
  19. 一种歌词的配乐生成装置,其特征在于,应用于配乐服务器,所述配乐服务器包括配乐库,所述配乐库包括片段和配乐片段的第一对应关系,其中任意一个片段具有一个包括至少一个配乐片段的配乐片段集合,所述装置包括获取单元、查找单元、确定单元和拼接单元:
    所述获取单元,用于获取通过对待配乐歌词进行分词处理所得到的至少一个片段;
    所述查找单元,用于根据所述至少一个片段查找所述配乐库,通过所述第一对应关系确定出所述至少一个片段分别对应的配乐片段集合;
    所述确定单元,用于根据所述待配乐歌词,从所述至少一个片段分别对应的配乐片段集合中确定出所述至少一个片段中每个片段分别对应的配乐片段;
    所述拼接单元,用于通过拼接确定出的配乐片段生成所述待配乐歌词的配乐。
  20. 根据权利要求19所述的装置,其特征在于,所述确定单元具体用于根据多个分词分别对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐,其中由一个文字片段所包括分词分别对应的配乐片段得到这个文字片段所对应的子配乐;
    所述拼接单元具体用于通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐。
  21. 一种歌词的配乐获取方法,其特征在于,应用于交互端,所述方法包括:
    向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
    从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
  22. 根据权利要求21所述的方法,其特征在于,所述从所述配乐服务器获取所述待配乐歌词对应的配乐,包括:
    从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;
    根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
  23. 根据权利要求22所述的方法,其特征在于,所述配乐信息包括携带所述配乐信息的待定配乐的评分和/或配乐类型。
  24. 根据权利要求21所述的方法,其特征在于,在所述从所述配乐服务器获取所述待配乐歌词对应的配乐之后,还包括:
    若获取的配乐不符合需求,向所述配乐服务器发送反馈信息,以使得所述服务器根据所述反馈信息重新生成所述待配乐歌词的配乐。
  25. 根据权利要求24所述的方法,其特征在于,所述反馈信息中包括描述所述需求的信息。
  26. 一种歌词的配乐获取装置,其特征在于,应用于交互端,所述装置包括发送单元 和获取单元:
    所述发送单元,用于向配乐服务器发送所获取的待配乐歌词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
    所述获取单元,用于从所述配乐服务器获取所述待配乐歌词对应的配乐,所述配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据所述第一对应关系查找所述配乐库得到的。
  27. 根据权利要求26所述的装置,其特征在于,所述获取单元包括获取子单元和选择子单元:
    所述获取子单元,用于从所述配乐服务器获取所述待配乐歌词对应的多个待定配乐,所述待定配乐携带有配乐信息;
    所述选择子单元,用于根据所述多个待定配乐的配乐信息,选择符合需求的待定配乐作为所述待配乐歌词的配乐。
  28. 一种歌词的配乐系统,其特征在于,所述配乐系统包括配乐服务器和交互端:
    所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合,所述配乐服务器,用于获取通过对待配乐歌词进行分词处理所获取的多个分词,所述待配乐歌词包括至少一个文字片段,所述至少一个文字片段中一个文字片段包括至少一个分词;根据所述多个分词查找所述配乐库,利用所述第一对应关系确定出每个分词分别对应的配乐片段集合;根据所述多个分词对应的配乐片段集合以及所述至少一个文字片段所包括的分词,确定出所述至少一个文字片段分别对应的子配乐;通过拼接所述至少一个文字片段分别对应的子配乐生成所述待配乐歌词的配乐;
    所述交互端,用于向所述配乐服务器发送所获取的所述待配乐歌词;从所述配乐服务器获取所述待配乐歌词对应的配乐。
  29. 一种音乐编辑器,其特征在于,所述音乐编辑器具有用于获取所输入待配乐歌词的输入界面,以及用于展示配乐的编辑界面;
    所述输入界面用于获取待配乐歌词,所获取的待配乐歌词为在所述输入界面中输入的歌词,或者为对输入的语音进行识别后得到的歌词;
    当获取输出指令时,所述音乐编辑器用于向配乐服务器发送通过所述输入界面获取的待配乐歌词,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
    当从所述配乐服务器获取所述待配乐歌词对应的配乐时,所述音乐编辑器用于将所述待配乐歌词对应的配乐展示在所述编辑界面中,所述待配乐歌词对应的配乐为所述配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找所述配乐库得到的。
  30. 根据权利要求29所述的音乐编辑器,其特征在于,所述编辑界面具有编辑按钮,当接收到对所述编辑按钮的触发时,所述编辑界面用于将所展示的配乐设置为可编辑的;
    所述编辑界面所展示的配乐还包括通过对所录制音频数据识别得到的配乐。
  31. 一种音乐播放器,其特征在于,所述音乐播放器具有混合界面和播放界面;
    所述混合界面用于将待配乐歌词以及所述待配乐歌词对应的配乐进行混合得到歌曲,所述待配乐歌词对应的配乐为配乐服务器根据所述待配乐歌词中分词所对应配乐片段集合中的配乐片段拼接得到,所述待配乐歌词中分词所对应配乐片段集合为所述配乐服务器根据第一对应关系查找配乐库得到的,所述配乐服务器包括配乐库,所述配乐库包括分词和配乐片段的第一对应关系,其中任意一个分词具有一个包括至少一个配乐片段的配乐片段集合;
    所述播放界面用于播放所述歌曲。
PCT/CN2017/117358 2016-12-30 2017-12-20 一种歌词的配乐生成方法和相关装置 WO2018121368A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201611264888.5 2016-12-30
CN201611264888.5A CN108268530B (zh) 2016-12-30 2016-12-30 一种歌词的配乐生成方法和相关装置

Publications (1)

Publication Number Publication Date
WO2018121368A1 true WO2018121368A1 (zh) 2018-07-05

Family

ID=62711084

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/117358 WO2018121368A1 (zh) 2016-12-30 2017-12-20 一种歌词的配乐生成方法和相关装置

Country Status (3)

Country Link
CN (1) CN108268530B (zh)
TW (1) TW201824249A (zh)
WO (1) WO2018121368A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109086408A (zh) * 2018-08-02 2018-12-25 腾讯科技(深圳)有限公司 文本生成方法、装置、电子设备及计算机可读介质
CN110807124A (zh) * 2019-11-05 2020-02-18 广州酷狗计算机科技有限公司 歌曲搜索方法、装置、设备和计算机可读存储介质
CN112669849A (zh) * 2020-12-18 2021-04-16 百度国际科技(深圳)有限公司 用于输出信息的方法、装置、设备以及存储介质
CN113377992A (zh) * 2021-06-21 2021-09-10 腾讯音乐娱乐科技(深圳)有限公司 歌曲分段方法、设备和存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109166564B (zh) * 2018-07-19 2023-06-06 平安科技(深圳)有限公司 为歌词文本生成乐曲的方法、装置及计算机可读存储介质
CN109309863B (zh) * 2018-08-01 2019-09-13 磐安鬼谷子文化策划有限公司 电影内容匹配机构
TWI713958B (zh) * 2018-12-22 2020-12-21 淇譽電子科技股份有限公司 自動詞曲創作系統及其方法
CN111339352B (zh) * 2020-01-22 2024-04-26 花瓣云科技有限公司 一种音频生成方法、装置和存储介质
TWI784434B (zh) * 2021-03-10 2022-11-21 國立清華大學 使用對抗生成網路與逆增強式學習法的自動作曲系統及方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103839559A (zh) * 2012-11-20 2014-06-04 华为技术有限公司 音频文件制作方法及终端设备
CN104391980A (zh) * 2014-12-08 2015-03-04 百度在线网络技术(北京)有限公司 生成歌曲的方法和装置
CN105070283A (zh) * 2015-08-27 2015-11-18 百度在线网络技术(北京)有限公司 为歌声语音配乐的方法和装置
CN105931625A (zh) * 2016-04-22 2016-09-07 成都涂鸦科技有限公司 基于文字输入的说唱音乐自动生成方法

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2206741A1 (en) * 1997-06-02 1998-12-02 Mitac Inc. Method and apparatus for generating musical accompaniment signals, and method and device for generating a video output in a musical accompaniment apparatus
CN101271457B (zh) * 2007-03-21 2010-09-29 中国科学院自动化研究所 一种基于旋律的音乐检索方法及装置
GB201109731D0 (en) * 2011-06-10 2011-07-27 System Ltd X Method and system for analysing audio tracks
US9459828B2 (en) * 2012-07-16 2016-10-04 Brian K. ALES Musically contextual audio advertisements
US20140018947A1 (en) * 2012-07-16 2014-01-16 SongFlutter, Inc. System and Method for Combining Two or More Songs in a Queue
CN104078035B (zh) * 2013-07-02 2016-10-05 深圳市腾讯计算机系统有限公司 一种演奏乐曲的方法及装置
CN105788589B (zh) * 2016-05-04 2021-07-06 腾讯科技(深圳)有限公司 一种音频数据的处理方法及装置
CN106057208B (zh) * 2016-06-14 2019-11-15 科大讯飞股份有限公司 一种音频修正方法及装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103839559A (zh) * 2012-11-20 2014-06-04 华为技术有限公司 音频文件制作方法及终端设备
CN104391980A (zh) * 2014-12-08 2015-03-04 百度在线网络技术(北京)有限公司 生成歌曲的方法和装置
CN105070283A (zh) * 2015-08-27 2015-11-18 百度在线网络技术(北京)有限公司 为歌声语音配乐的方法和装置
CN105931625A (zh) * 2016-04-22 2016-09-07 成都涂鸦科技有限公司 基于文字输入的说唱音乐自动生成方法

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109086408A (zh) * 2018-08-02 2018-12-25 腾讯科技(深圳)有限公司 文本生成方法、装置、电子设备及计算机可读介质
CN109086408B (zh) * 2018-08-02 2022-10-28 腾讯科技(深圳)有限公司 文本生成方法、装置、电子设备及计算机可读介质
CN110807124A (zh) * 2019-11-05 2020-02-18 广州酷狗计算机科技有限公司 歌曲搜索方法、装置、设备和计算机可读存储介质
CN112669849A (zh) * 2020-12-18 2021-04-16 百度国际科技(深圳)有限公司 用于输出信息的方法、装置、设备以及存储介质
CN113377992A (zh) * 2021-06-21 2021-09-10 腾讯音乐娱乐科技(深圳)有限公司 歌曲分段方法、设备和存储介质

Also Published As

Publication number Publication date
TW201824249A (zh) 2018-07-01
CN108268530A (zh) 2018-07-10
CN108268530B (zh) 2022-04-29

Similar Documents

Publication Publication Date Title
WO2018121368A1 (zh) 一种歌词的配乐生成方法和相关装置
US10229669B2 (en) Apparatus, process, and program for combining speech and audio data
CN108806656B (zh) 歌曲的自动生成
CN108806655B (zh) 歌曲的自动生成
US10776422B2 (en) Dual sound source audio data processing method and apparatus
US9532136B2 (en) Semantic audio track mixer
US8712776B2 (en) Systems and methods for selective text to speech synthesis
US11710474B2 (en) Text-to-speech from media content item snippets
EP3759706B1 (en) Method, computer program and system for combining audio signals
JP2009210790A (ja) 選曲歌手分析推薦装置、その方法及びプログラム
TW202025078A (zh) 自動詞曲創作系統及其方法
Lin et al. Audio musical dice game: A user-preference-aware medley generating system
JP2006178104A (ja) 楽曲生成方法,その装置,そのシステム
JP2014013340A (ja) 作曲支援装置、作曲支援方法、作曲支援プログラム、作曲支援プログラムを格納した記録媒体およびメロディ検索装置
KR101807754B1 (ko) 자작곡 생성 서버 및 그 방법
Jani et al. Experimental investigation of transitions for mixed speech and music playlist generation
Cano et al. Sesquialtera in the Colombian Bambuco: Perception and Estimation of Beat and Meter-Extended version.
TWM578439U (zh) 自動詞曲創作系統
KR20140054810A (ko) 반주음악 제작 서비스 시스템 및 그 방법, 그리고 이에 적용되는 장치
Antoine et al. Computer Generated Orchestration: Towards Using Musical Timbre in Composition
Aspillaga et al. Mixme: A recommendation system for DJs
Liu Research on Aesthetic Emotion Expression of Piano Music Based on Deep Learning
JP2007072316A (ja) 唱歌録音作品の検索方法、唱歌録音作品データベースの構造

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17887733

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17887733

Country of ref document: EP

Kind code of ref document: A1