WO2021235708A1 - 음성 인식 선곡 서비스 제공 방법 및 음성 인식 선곡 장치 - Google Patents
음성 인식 선곡 서비스 제공 방법 및 음성 인식 선곡 장치 Download PDFInfo
- Publication number
- WO2021235708A1 WO2021235708A1 PCT/KR2021/004822 KR2021004822W WO2021235708A1 WO 2021235708 A1 WO2021235708 A1 WO 2021235708A1 KR 2021004822 W KR2021004822 W KR 2021004822W WO 2021235708 A1 WO2021235708 A1 WO 2021235708A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- search
- utterance
- tokens
- song
- speech
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 27
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 description 16
- 230000008569 process Effects 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000013473 artificial intelligence Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 230000003321 amplification Effects 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000000844 transformation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/686—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title or artist information, time, location or usage information, user ratings
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/242—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2457—Query processing with adaptation to user needs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/40—Information retrieval; Database structures therefor; File system structures therefor of multimedia data, e.g. slideshows comprising image and additional audio data
- G06F16/43—Querying
- G06F16/432—Query formulation
- G06F16/433—Query formulation using audio data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/63—Querying
- G06F16/632—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/63—Querying
- G06F16/638—Presentation of query results
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/903—Querying
- G06F16/9032—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/268—Morphological analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/18—Speech classification or search using natural language modelling
- G10L15/183—Speech classification or search using natural language modelling using context dependencies, e.g. language models
- G10L15/19—Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10—TECHNICAL SUBJECTS COVERED BY FORMER USPC
- Y10S—TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y10S715/00—Data processing: presentation processing of document, operator interface processing, and screen saver display processing
- Y10S715/978—Audio interaction as part of an operator interface
Definitions
- the present invention relates to a method for providing a voice recognition music selection service and a voice recognition music selection apparatus.
- this method has a problem in that it takes a lot of time and effort to search for a song as well as a lot of time.
- Patent Document 1 Republic of Korea Utility Model Publication No. 20-0202916 (Notice on Nov. 15, 2000)
- An object of the present invention is to provide a method of providing a voice recognition music selection service and a voice recognition music selection apparatus capable of more easily and accurately searching for a song.
- Step of receiving search utterances uttered for search in text form generating a plurality of search tokens by dividing the search utterances based on spacing, part-of-speech, and stems and endings, and converting the plurality of search tokens into a plurality of reference tokens
- a method for providing a voice recognition selection service comprising the steps of comparing with , calculating a similarity score of a search utterance with respect to a reference utterance according to the comparison result, and providing song information to a user based on the similarity score.
- the standard utterance is first divided based on spacing, the first divided standard utterance is divided second based on the part-of-speech, and verbs and adjectives are divided into stems and endings among the second divided standard speech sentences. It can be generated by tertiary division based on .
- an extension token configured by sequentially adding one letter to the first letter and first letter of each of the plurality of reference tokens may be further provided.
- the step of receiving the search utterance in text form may include receiving the search utterance uttered by the user to search for a song by voice, and converting the search utterance into text form.
- the step of receiving the search utterance in text form may further include correcting errors in the search utterance by comparing the search utterance with a song name and a singer name after converting the search utterance into a text form.
- the search utterances are first divided based on spacing, the first divided search utterances are divided second based on the part-of-speech, and verbs and adjectives among the secondly divided search utterances It is possible to generate a plurality of search tokens by tertiary division based on the stem and the ending.
- the similarity score when the plurality of search tokens are all included in the plurality of reference tokens, the similarity score may be calculated.
- a unit score may be calculated for each of the search tokens matching the reference token, and the similarity score may be calculated by summing the unit scores.
- the unit score may be calculated higher as the number of characters of the search token matching the reference token increases.
- a reference token providing unit that receives a plurality of reference tokens generated by dividing a reference utterance including a song name and a singer name based on spacing, part-of-speech, and stem and ending, and a user utters a utterance to search for a song
- An input unit that receives one search utterance in text form, a search token generator that generates a plurality of search tokens by dividing the search utterances based on spacing, part-of-speech, stem and ending, and a plurality of search tokens with a plurality of reference tokens
- a speech recognition selection device comprising a token comparison unit to compare, a score calculation unit for calculating a similarity score of a search utterance with respect to a reference utterance according to a comparison result, and a result providing unit for providing song information to a user based on the similarity score
- FIG. 1 is a diagram illustrating a system for providing a voice recognition music selection service according to an embodiment of the present invention.
- FIG. 2 is a flowchart illustrating a method for providing a voice recognition music selection service according to an embodiment of the present invention.
- FIG. 3 is a flowchart illustrating a step of receiving a search utterance in text form according to an embodiment of the present invention
- FIG. 4 is a diagram illustrating a process of determining whether to calculate a similarity score in the step of calculating a similarity score according to an embodiment of the present invention
- FIG. 5 is a flowchart illustrating a similarity score calculation process in the step of calculating a similarity score according to an embodiment of the present invention.
- FIG. 6 is a diagram showing the configuration of a voice recognition music selection apparatus according to an embodiment of the present invention.
- an input device for receiving a search utterance uttered by the user 10 for song search, a conversion server for converting the search utterance into text form, and text
- a generation server that generates a plurality of search tokens by dividing search utterances converted into forms based on spacing, part-of-speech, stem and ending, and division based on spacing, part-of-speech, and stem and ending stores a plurality of reference tokens generated through
- An analysis server for generating song information to be provided to the user 10, an accompaniment device and input device for reserving and playing a song selected by the user 10 from among the song information provided to the user 10, a generation server, and analysis
- a system for providing a voice recognition music selection service including a transmission/reception device for transmitting and receiving data between a server and an accompaniment device.
- the voice data called the search utterance uttered by the user 10 for song search is sent to the transceiver through the input device, and the corresponding voice data is transmitted to the conversion server by the transceiver and is converted into text data to become text data.
- the corresponding text data is again received by the transceiver and transmitted to the generating server, and at this time, it is divided based on spacing, part-of-speech, and stem and ending to become a plurality of search tokens, and the plurality of search tokens are sent to the analysis server through the transceiver. can be sent to
- the plurality of search tokens are compared with a plurality of reference tokens stored in the analysis server and generated by dividing the reference utterance including the song name and the singer name based on spacing, part-of-speech, and stem and ending, and the analysis server receives the comparison result Accordingly, it is possible to calculate the similarity score, generate song information to be provided to the user 10 based on the similarity score, and then transmit it to the transceiver device.
- the corresponding information is transmitted to the accompaniment device to finally start accompaniment of the corresponding song.
- a plurality of reference tokens generated by dividing a standard utterance including a song name and a singer name based on spacing, part-of-speech, and stem and ending are provided.
- a method for providing a voice recognition music selection service including a step ( S160 ) of providing song information to the user 10 .
- the search is performed only by uttering or saying the content of the song to be searched, the user 10 can search for the song more easily, and the search uttered by the user 10 is performed.
- a more accurate song search is possible because a search result is derived by comparing a plurality of search tokens generated by dividing a utterance with a plurality of reference tokens generated by dividing a reference utterance including a song name and a singer name.
- a plurality of reference tokens generated by dividing a reference utterance including a song name and a singer name based on spacing, part-of-speech, and stem and ending may be provided.
- data called a plurality of reference tokens having all of the song information-related contents such as the song name and the singer name are formed, and the data are accumulated to form one data field for comparison with the contents of the user's search request.
- 'token' is a group of one letter or a plurality of letters generated by dividing an utterance based on spacing, part-of-speech, and stem and ending.
- 'spoken text' can be understood as literally meaning 'a sentence in which words spoken out loud are written in text'.
- the standard utterance is first divided based on spacing, the first divided standard utterance is divided second based on the part-of-speech, and verbs and adjectives are divided into stems and endings among the second divided standard speech sentences. It can be generated by tertiary division based on . More specifically, the tertiary division based on the stem and the ending can be done at the character level.
- a plurality of reference tokens may be prepared by dividing a plurality of words constituting a reference utterance based on meaning, etc.
- a song search is compared with a plurality of search tokens to be described later. In doing so, more effective results can be derived.
- 'Lee Eun-mi's lover has a lover and find it' is first divided into 'Lee Eun-mi's', 'lover', 'have', and 'find me' based on the spacing, which is again ), 'of' (subject), 'have' (verb), and 'find me' (verb), among which the verbs 'have' and 'find me' '(Stem), 'Ah' (Stem), and 'Find' (Stem), 'A' (Mother), and 'Give' (Stem + End) can be tertiary, respectively.
- reference tokens such as 'Eunmi Lee', 'of', 'lover', 'have,' 'here', 'find', 'ah', and 'give me' can be generated.
- an expansion token configured by sequentially adding one letter to the first letter and first letter of each of the plurality of reference tokens may be further provided.
- the first letter of each of the plurality of reference tokens and the data of the expansion token are added to the data field, so that a more detailed and accurate comparison with a search token to be described later and a song search result can be derived.
- the first letter of 'Lee' and extension tokens such as 'Eun Lee', 'Eun Lee', and 'Eunmi Lee' may be further provided.
- the duplicate token 'Lee Eun-mi' can be selectively deleted.
- the search utterance uttered by the user 10 for song search may be input in text form.
- the user 10 can search for a song by simply speaking, so that the user 10 can more easily search for a song.
- the step of receiving the search utterance in text form is, as shown in FIG. 3 , the step of receiving the search utterance uttered by the user 10 by voice (S122), and It may include converting the search utterance into a text form (S124).
- the user 10 may utter a search utterance including at least one of a song name and a singer's name, and preferably, a search utterance including a song name and a singer's name (for example, 'Lee Eun-mi') for more accurate song search. You can request a search by uttering 'I have a lover.
- source data for generating a plurality of search tokens may be prepared through the step of converting the search utterance into a text form ( S124 ).
- song information data including a song name and a singer name may be utilized in order to increase the conversion accuracy of the search utterance into a text form, and more specifically, the above-mentioned song
- the conversion accuracy may be improved by performing conversion by setting information data as a reference value of conversion.
- AI artificial intelligence
- the search utterance is compared with the song name and singer name to obtain the search utterance. It may further include the step of correcting the error (S126).
- the search utterance 'Lee Eun-mi' is compared with the singer's name 'Lee Eun-mi' and the song name 'I have a lover'. You can prepare the text by correcting the above error with 'Find me a lover of'.
- a score may be calculated by comparing the search utterance with the song name and the singer name, and the comparison score may exceed a preset reference value. You can edit search utterances.
- the search utterance when the search utterance is corrected, the possibility that the search utterance will be corrected to a text suitable for the intention of the user 10 may be further improved.
- artificial intelligence technology for example, deep learning can be applied, and accordingly, more accurate correction of sentences containing the words can be made through learning about frequently searched words. .
- step S122 of receiving the search utterance by voice the search utterance may be input through a receiver 122 capable of voice recognition.
- the receiver 122 may perform at least one of noise filtering and voice amplification in order to increase the recognition rate of voice in a noise situation.
- a place such as a karaoke room where the present invention can be mainly implemented is a space in which a lot of ambient noise such as a song sound is scattered, it is important to accurately recognize the voice of the user 10 .
- the receiver 122 may increase the voice recognition rate of the user 10 by performing at least one of filtering the ambient noise and amplifying the user 10's voice, thereby further improving the accuracy of the song search.
- a plurality of search tokens may be generated by dividing search utterances based on spacing, parts of speech, and stems and endings. Accordingly, by providing a plurality of search tokens corresponding to the plurality of reference tokens, the effectiveness in comparing the two tokens may be improved.
- the search utterance is first divided based on spacing, the first divided search utterance is secondarily divided based on the part-of-speech, and among the secondly divided search utterances,
- a plurality of search tokens can be generated by dividing verbs and adjectives into thirds based on stems and endings.
- a plurality of search tokens may be prepared by dividing a plurality of words constituting a search utterance based on the meaning, etc. In this case, in performing a song search by comparing the above-mentioned reference token based on the meaning of the word, etc. It is possible to derive more effective search results.
- the plurality of search tokens may be compared with the plurality of reference tokens. Accordingly, a reference data value for calculating a similarity score, which will be described later, and providing a result thereof may be prepared.
- the step of comparing the plurality of search tokens with the plurality of reference tokens may be performed through a separately provided comparison server.
- the similarity score of the search utterance with respect to the reference utterance may be calculated according to the above-described comparison result.
- song information can be selected from among the search results and provided to the user 10 , and accordingly, the user 10 can be provided with a more effective search result.
- a similarity score may be calculated.
- a more accurate song search result can be provided to the user 10 by preventing unnecessary song information from being included in the search result by calculating the similarity score.
- the user 10 may utter again except for letters or words that are not expected to be included in the plurality of reference tokens. be able to
- the search token 'I am' (mother) is not included in the plurality of criterion tokens for 'Find Eun-mi Lee's lover. Find it,' so the user You can ask (10) to re-speak the search utterance.
- step of calculating the similarity score S150
- information on a search token not included in the plurality of reference tokens is provided while requesting the user 10 to utter the search utterance again.
- a unit score may be calculated for each of the search tokens matching the reference token, and the similarity score may be calculated by summing the unit scores.
- a unit score is given for each search token matching the reference token, and a similarity score as a final score may be calculated by adding up all of the above-described unit scores.
- the unit score may be calculated to be higher as the number of characters of the search token matching the reference token increases.
- the matching of the search token with a large number of characters to the reference token means that the user 10 is highly likely to be a keyword constituting the desired song information. Differentiality may be assigned to the calculated similarity score.
- the search token 'lover' which has more characters than the search token 'have', can be viewed as a keyword constituting the song title when viewed as a whole song title. By giving it, it is possible to set the 'lover' to contribute more to the calculation of the similarity score.
- the song information may be provided to the user 10 based on the similarity score.
- song information including a song name and a singer name having a similarity score equal to or greater than a preset value may be provided to the user 10 .
- the accuracy and appropriateness of the song search result can be further improved.
- the song information may be provided to the user 10 in the order of the highest similarity score.
- the user 10 can more quickly find a desired song by looking at the song information with a high similarity score from the song information including the song name and singer name of the provided score.
- a voice recognition music selection apparatus 100 according to an embodiment of the present invention will be described.
- a reference token providing unit 110 that receives a plurality of reference tokens generated by dividing a reference utterance including a song name and a singer name based on spacing, part-of-speech, and stem and ending.
- the input unit 120 for receiving the search utterance uttered by the user 10 to search for a song in text form, dividing the search utterance based on spacing, part-of-speech, and stem and ending to generate a plurality of search tokens
- the user 10 can more easily search for a song and obtain a more accurate song search result.
- the reference token providing unit 110 may receive a plurality of reference tokens generated by dividing a reference utterance including a song name and a singer name based on spacing, parts of speech, and stems and endings.
- a plurality of reference tokens may be provided in the form of download from a server, etc., and a search token to be described later is transmitted to the server while the plurality of reference tokens are stored in the server, etc.
- a reference token can also be used.
- the standard utterance is first divided based on spacing, the first divided standard utterance is divided second based on the part-of-speech, and verbs and adjectives are divided into stems and endings among the second divided standard speech sentences. It can be generated by tertiary division based on .
- the reference token providing unit 110 may further receive an extension token configured by sequentially adding one letter to the first letter and the first letter of each of the plurality of reference tokens.
- the input unit 120 may receive a search utterance uttered by the user 10 for song search in the form of text.
- the input unit 120 may include a voice input unit 120 that receives the search utterance uttered by the user 10 for song search by voice, and a text converter that converts the search utterance into a text form. .
- the conversion of the search utterance into the text form may be performed through a separate server, and for this purpose, the search utterance received by voice may be transmitted to the server.
- the text conversion unit may utilize song information data including a song name and a singer name in order to increase the conversion accuracy of the search utterance into a text form, and more specifically, set the above-mentioned song information data as a reference value for conversion to perform conversion. By doing so, the conversion accuracy can be improved.
- artificial intelligence technology can be utilized, for example, by improving the sensitivity to frequently exposed words through deep learning, the conversion accuracy can be further improved.
- the input unit 120 may further include an error correction unit for correcting errors in the search utterance by comparing the search utterance with the song name and the singer's name.
- the error correction unit may calculate a score by comparing the search utterance with the song name and the singer's name, and may correct the search utterance based on the song information of the highest score among the song information for which the comparison score exceeds a preset reference value.
- artificial intelligence technology for example, deep learning may be applied, and accordingly, more accurate correction of sentences including the corresponding words may be made through learning of frequently searched words.
- the voice input unit 120 may receive a search utterance through the receiver 122 capable of voice recognition.
- the receiver 122 may include any input device capable of receiving and recognizing a voice, for example, a microphone provided in the accompaniment or remote control, a microphone provided in the device, a dedicated receiver provided separately, or the device and A microphone provided in the terminal of the interlocked user 10 may be included in the receiver 122 .
- the receiver 122 may perform at least one of noise filtering and voice amplification in order to increase the recognition rate of voice in a noise situation.
- the search token generating unit 130 may generate a plurality of search tokens by dividing search utterances based on spacing, parts of speech, and stems and endings.
- the plurality of search tokens may be generated in the server by transmitting the search utterance converted into text form to a separately provided server.
- the search token generating unit 130 first divides the search utterances based on spacing, divides the firstly divided search utterances secondarily based on the part-of-speech, and divides verbs and adjectives among the secondly divided search utterances.
- a plurality of search tokens can be generated by tertiary division based on the stem and the ending.
- the token comparison unit 140 may compare a plurality of search tokens with a plurality of reference tokens.
- a plurality of search tokens and a plurality of reference tokens may be compared with each other in a separately provided server, and for this purpose, a plurality of search tokens may be transmitted to the server.
- the score calculator 150 may calculate a similarity score of the search utterance with respect to the reference utterance according to the above-described comparison result.
- the score calculator 150 may calculate a similarity score when a plurality of search tokens are all included in a plurality of reference tokens.
- the score calculator 150 may request the user 10 to re-utter the search utterance.
- the score calculator 150 may provide information on a search token not included in the plurality of reference tokens while requesting the user 10 to re-utter the search utterance.
- the score calculator 150 may calculate a unit score for each of the search tokens matching the reference token, and may calculate a similarity score by summing the unit scores.
- the unit score may be calculated to be higher as the number of characters of the search token matching the reference token increases.
- the result providing unit 160 may provide song information to the user 10 based on the similarity score.
- the result providing unit 160 may provide the user 10 with song information including the name of the song and the singer whose similarity score is equal to or greater than a preset value.
- the result providing unit 160 may provide the user 10 with song information including the song name and the singer name in the order of the highest similarity score.
- the above-described similarity score calculation may be performed on a separate server, and a display for requesting re-sentence of a search utterance, providing information on a search token not included in a plurality of reference tokens, and providing song information as a search result Additional may be provided.
- each component may be identified as a respective process.
- the process of the above-described embodiment can be easily understood from the point of view of the components of the apparatus.
- the technical contents described above may be implemented in the form of program instructions that can be executed through various computer means and recorded in a computer-readable medium.
- the computer-readable medium may include program instructions, data files, data structures, etc. alone or in combination.
- the program instructions recorded on the medium may be specially designed and configured for the embodiments or may be known and available to those skilled in the art of computer software.
- Examples of the computer-readable recording medium include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic such as floppy disks.
- - includes magneto-optical media, and hardware devices specially configured to store and carry out program instructions, such as ROM, RAM, flash memory, and the like.
- Examples of program instructions include not only machine language codes such as those generated by a compiler, but also high-level language codes that can be executed by a computer using an interpreter or the like.
- a hardware device may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Library & Information Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Description
Claims (10)
- 곡명과 가수명을 포함한 기준 발화문(發話文)을 띄어쓰기, 품사 및 어간과 어미를 기준으로 분할하여 생성된 복수의 기준 토큰(token)을 제공받는 단계;사용자가 곡 검색을 위해 발화한 검색 발화문을 텍스트 형태로 입력받는 단계;상기 검색 발화문을 띄어쓰기, 품사 및 어간과 어미를 기준으로 분할하여 복수의 검색 토큰을 생성하는 단계;상기 복수의 검색 토큰을 상기 복수의 기준 토큰과 비교하는 단계;상기 비교 결과에 따라 상기 기준 발화문에 대한 상기 검색 발화문의 유사도 점수를 산출하는 단계; 및상기 유사도 점수를 기준으로 상기 사용자에게 곡 정보를 제공하는 단계를 포함하는 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 복수의 기준 토큰은,상기 기준 발화문을 상기 띄어쓰기를 기준으로 1차 분할하고,상기 1차 분할된 상기 기준 발화문을 상기 품사를 기준으로 2차 분할하며,상기 2차 분할된 상기 기준 발화문 중 동사와 형용사를 상기 어간과 어미를 기준으로 3차 분할하여 생성되는, 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 복수의 기준 토큰을 제공받는 단계에서,상기 복수의 기준 토큰 각각의 첫 글자 및 상기 첫 글자에 한 글자씩 순차적으로 추가하여 구성되는 확장 토큰을 더 제공받는, 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 검색 발화문을 텍스트 형태로 입력받는 단계는,상기 사용자가 상기 곡 검색을 위해 발화한 상기 검색 발화문을 음성으로 입력받는 단계; 및상기 검색 발화문을 상기 텍스트 형태로 변환하는 단계를 포함하는, 음성 인식 선곡 서비스 제공 방법.
- 제4항에 있어서,상기 검색 발화문을 텍스트 형태로 입력받는 단계는,상기 검색 발화문을 텍스트 형태로 변환하는 단계 이후에,상기 검색 발화문을 상기 곡명 및 상기 가수명과 비교하여 상기 검색 발화문의 오류를 수정하는 단계를 더 포함하는, 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 복수의 검색 토큰을 생성하는 단계에서,상기 검색 발화문을 상기 띄어쓰기를 기준으로 1차 분할하고,상기 1차 분할된 상기 검색 발화문을 상기 품사를 기준으로 2차 분할하며,상기 2차 분할된 상기 검색 발화문 중 동사와 형용사를 상기 어간과 어미를 기준으로 3차 분할하여 상기 복수의 검색 토큰을 생성하는, 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 유사도 점수를 산출하는 단계에서,상기 복수의 검색 토큰이 상기 복수의 기준 토큰에 모두 포함되는 경우 상기 유사도 점수를 산출하는, 음성 인식 선곡 서비스 제공 방법.
- 제1항에 있어서,상기 유사도 점수를 산출하는 단계에서,상기 기준 토큰에 매칭되는 상기 검색 토큰 각각에 대하여 단위 점수를 산출하고 상기 단위 점수를 합산하여 상기 유사도 점수를 산출하는, 음성 인식 선곡 서비스 제공 방법.
- 제8항에 있어서,상기 단위 점수는 상기 기준 토큰에 매칭되는 상기 검색 토큰의 글자수가 많을수록 높게 산출되는, 음성 인식 선곡 서비스 제공 방법.
- 곡명과 가수명을 포함한 기준 발화문을 띄어쓰기, 품사 및 어간과 어미를 기준으로 분할하여 생성된 복수의 기준 토큰을 제공받는 기준 토큰 제공부;사용자가 곡 검색을 위해 발화한 검색 발화문을 텍스트 형태로 입력받는 입력부;상기 검색 발화문을 띄어쓰기, 품사 및 어간과 어미를 기준으로 분할하여 복수의 검색 토큰을 생성하는 검색 토큰 생성부;상기 복수의 검색 토큰을 상기 복수의 기준 토큰과 비교하는 토큰 비교부;상기 비교 결과에 따라 상기 기준 발화문에 대한 상기 검색 발화문의 유사도 점수를 산출하는 점수 산출부; 및상기 유사도 점수를 기준으로 상기 사용자에게 곡 정보를 제공하는 결과 제공부를 포함하는 음성 인식 선곡 장치.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2022558058A JP2023524374A (ja) | 2020-05-18 | 2021-04-16 | 音声認識選曲サービスの提供方法及び音声認識選曲装置 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020200059286A KR102362815B1 (ko) | 2020-05-18 | 2020-05-18 | 음성 인식 선곡 서비스 제공 방법 및 음성 인식 선곡 장치 |
KR10-2020-0059286 | 2020-05-18 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021235708A1 true WO2021235708A1 (ko) | 2021-11-25 |
Family
ID=78707971
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2021/004822 WO2021235708A1 (ko) | 2020-05-18 | 2021-04-16 | 음성 인식 선곡 서비스 제공 방법 및 음성 인식 선곡 장치 |
Country Status (3)
Country | Link |
---|---|
JP (1) | JP2023524374A (ko) |
KR (2) | KR102362815B1 (ko) |
WO (1) | WO2021235708A1 (ko) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20090041923A (ko) * | 2007-10-25 | 2009-04-29 | 한국전자통신연구원 | 음성 인식 방법 |
KR20130040054A (ko) * | 2011-10-13 | 2013-04-23 | 현대자동차주식회사 | 음원정보 관리 서비스 시스템 |
JP2014197117A (ja) * | 2013-03-29 | 2014-10-16 | 富士通株式会社 | 音声合成装置及び言語辞書登録方法 |
US20150154958A1 (en) * | 2012-08-24 | 2015-06-04 | Tencent Technology (Shenzhen) Company Limited | Multimedia information retrieval method and electronic device |
KR101944303B1 (ko) * | 2017-08-29 | 2019-02-07 | 쌍용자동차 주식회사 | 음성인식을 이용한 자동 음원 선곡이 가능한 자동차 오디오 시스템 제어방법 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR200202916Y1 (ko) | 2000-06-15 | 2000-11-15 | 최요환 | 노래목록검색기능을 갖는 노래반주기용 리모콘 |
-
2020
- 2020-05-18 KR KR1020200059286A patent/KR102362815B1/ko active IP Right Grant
-
2021
- 2021-04-16 WO PCT/KR2021/004822 patent/WO2021235708A1/ko active Application Filing
- 2021-04-16 JP JP2022558058A patent/JP2023524374A/ja active Pending
-
2022
- 2022-02-09 KR KR1020220016963A patent/KR20220020878A/ko active IP Right Grant
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20090041923A (ko) * | 2007-10-25 | 2009-04-29 | 한국전자통신연구원 | 음성 인식 방법 |
KR20130040054A (ko) * | 2011-10-13 | 2013-04-23 | 현대자동차주식회사 | 음원정보 관리 서비스 시스템 |
US20150154958A1 (en) * | 2012-08-24 | 2015-06-04 | Tencent Technology (Shenzhen) Company Limited | Multimedia information retrieval method and electronic device |
JP2014197117A (ja) * | 2013-03-29 | 2014-10-16 | 富士通株式会社 | 音声合成装置及び言語辞書登録方法 |
KR101944303B1 (ko) * | 2017-08-29 | 2019-02-07 | 쌍용자동차 주식회사 | 음성인식을 이용한 자동 음원 선곡이 가능한 자동차 오디오 시스템 제어방법 |
Also Published As
Publication number | Publication date |
---|---|
KR102362815B1 (ko) | 2022-02-14 |
KR20220020878A (ko) | 2022-02-21 |
KR20210142446A (ko) | 2021-11-25 |
JP2023524374A (ja) | 2023-06-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020145439A1 (ko) | 감정 정보 기반의 음성 합성 방법 및 장치 | |
WO2020263034A1 (en) | Device for recognizing speech input from user and operating method thereof | |
WO2020246702A1 (en) | Electronic device and method for controlling the electronic device thereof | |
WO2019208860A1 (ko) | 음성 인식 기술을 이용한 다자간 대화 기록/출력 방법 및 이를 위한 장치 | |
WO2015005679A1 (ko) | 음성 인식 방법, 장치 및 시스템 | |
WO2015068947A1 (ko) | 녹취된 음성 데이터에 대한 핵심어 추출 기반 발화 내용 파악 시스템과, 이 시스템을 이용한 인덱싱 방법 및 발화 내용 파악 방법 | |
WO2015178600A1 (en) | Speech recognition method and apparatus using device information | |
WO2019112342A1 (en) | Voice recognition apparatus and operation method thereof cross-reference to related application | |
WO2020050509A1 (en) | Voice synthesis device | |
WO2013077589A1 (ko) | 음성인식 부가 서비스 제공 방법 및 이에 적용되는 장치 | |
WO2020027619A1 (ko) | 순차적 운율 특징을 기초로 기계학습을 이용한 텍스트-음성 합성 방법, 장치 및 컴퓨터 판독가능한 저장매체 | |
WO2020139058A1 (en) | Cross-device voiceprint recognition | |
WO2021002584A1 (ko) | 음성을 통한 전자문서 제공 방법, 음성을 통한 전자문서 작성 방법 및 장치 | |
WO2019139428A1 (ko) | 다중 언어 텍스트-음성 합성 방법 | |
JP3799280B2 (ja) | 対話システムおよびその制御方法 | |
WO2019139301A1 (ko) | 전자 장치 및 그 자막 표현 방법 | |
WO2021029643A1 (en) | System and method for modifying speech recognition result | |
WO2022080774A1 (ko) | 말 장애 평가 장치, 방법 및 프로그램 | |
WO2021251539A1 (ko) | 인공신경망을 이용한 대화형 메시지 구현 방법 및 그 장치 | |
WO2014115952A1 (ko) | 유머 발화를 이용하는 음성 대화 시스템 및 그 방법 | |
WO2020138662A1 (ko) | 전자 장치 및 그의 제어 방법 | |
WO2018056779A1 (en) | Method of translating speech signal and electronic device employing the same | |
WO2022260432A1 (ko) | 자연어로 표현된 스타일 태그를 이용한 합성 음성 생성 방법 및 시스템 | |
WO2014163231A1 (ko) | 복수의 음원이 출력되는 환경하에서 음성 인식에 이용될 음성 신호의 추출 방법 및 음성 신호의 추출 장치 | |
WO2020204256A1 (ko) | 음성합성엔진을 이용한 멀티미디어 음성인식 자동 평가시스템 |
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: 21808085 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2022558058 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21808085 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 12/05/2023) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21808085 Country of ref document: EP Kind code of ref document: A1 |