FR3038440A1 - METHOD OF EXTRACTING AND ASSEMBLING SONGS FROM MUSICAL RECORDINGS - Google Patents

Abstract

A method of extracting a musical segment from a musical recording file, the method comprising the steps of: identifying in a musical recording (AF), beat times and first and second successive periods of absence of main voice (MVC), identifying a first beat moment in the first period, and a second beat moment in the second period, extracting in a file (AFE) a musical segment (SG) of the musical recording between the first beat moment (TSS) of the segment, and the second beat instant (TEE), and forming in the segment a start transition (TS) extending from the beginning of the segment to a third beat (TSE), during which a decreasing attenuation is applied to the sound volume of the segment, and an end transition (TE) extending from a fourth beat (TES) to the end of the segment, during which an att increasing nuation is applied to the volume of the segment.

Description

METHOD OF EXTRACTING AND ASSEMBLING SONGS FROM MUSICAL RECORDINGS

The present invention relates to a method and a device for editing musical recordings, in particular for extracting parts from these recordings and combining them. Generally, a musical work includes a melody, a harmony, a dynamic, and a rhythm. The musical rhythm corresponds to the organization of the musical work in time. The rhythm is usually composed of different pulsations occurring at different time scales and at different amplitudes. The most remarkable pulsation is the beat felt corresponding to the rhythm felt leading the listener to stomp. Another pulse is at the scale of the musical measure. Generally, the musical measure concerns the rhythm of harmonic variation or the duration of the rhythmic pattern of the musical work. In musical writing, the musical measures are separated by vertical bars. Classically, a measurement comprises at least two felt beats.

Many musical works, especially those related to pop music, have a remarkable structure of different parts that can be repeated. Thus, such a musical work comprises several parts such as an introduction, one or more verses, a chorus, a bridge and a conclusion (outro).

Since the appearance of musical recordings in digital form, stored on optical disc or in memories, it is possible to listen to musical recordings in an order different from that in which they appear on the recording medium. When the recordings are stored in a memory, it is possible to set up playlists ("playlists") to continuously chain, listening to musical recordings from different recording media.

However, when a playlist is played, each of the musical recordings is played from the beginning to the end of the recording. In addition, a period of silence is usually inserted between two successive readings of musical recordings from the playlist.

On the other hand, it is difficult to extract parts of different musical recordings, without causing parasitic sounds to appear, and to string together such excerpts so as to form a single musical recording without discontinuity between the assembled excerpts. It is even more difficult to assemble excerpts of musical recordings without showing rhythm breaks or beat shifts. This is why the assembly of musical excerpts is done by professionals. Generally, such an assembly is performed by providing a transition period during which playback of a next extract is started before the end of the reading of the current extract. In order for this transition period to occur smoothly, the rhythms of the two extracts must match and be synchronized.

It is therefore desirable to provide a method for extracting from musical recordings, segments that can be easily assembled with segments extracted from other musical recordings. It is also desirable to assemble two segments with a transition between the two segments that is as smooth as possible, with synchronized beats. It is also desirable that these operations can be performed without using a professional or professional software. It may further be desirable that the assembly of two extracted segments can be performed by a mobile terminal such as a smartphone.

Embodiments are directed to a method of extracting a musical segment from a musical recording file, the method comprising the steps of: identifying beat times in a musical recording, identifying in the musical recording first and second successive periods of absence of main voice, identifying in the first period a first beat moment, identifying in the second period a second beat time, extracting in a file a musical segment of the musical recording between a moment the beginning of the musical segment corresponding to the first beat time identified, and an end time of the musical segment corresponding to the second beat time identified, and forming in the extracted musical segment a start transition extending from the beginning of the musical segment until 'to a third beat, during which a decreasing attenuation is applied to the sound volume at the beginning of the musical segment, and an end transition extending from a fourth beat to the end of the segment, during which increasing attenuation is applied to the sound volume at the end of the musical segment.

According to one embodiment, the third and second beat times coincide with first musical measurement times of the musical recording.

According to one embodiment, the decreasing attenuation applied during the start transition of the musical segment has a maximum value at the start time of the musical segment, and a zero value at the third beat time.

According to one embodiment, the decreasing attenuation applied during the end transition of the musical segment has a zero value at the fourth beat time, and a maximum value at the end time of the musical segment.

According to one embodiment, the extraction method comprises steps of recording in a synchronization file associated with the segment of at least one of the following information: a list of positions of beat instants of the musical segment, a sound volume at the beginning of the music segment, a sound volume at the end of the music segment, a musical segment position indicator at the beginning of the music recording file, a musical segment position indicator at the end of the music recording file, a mean number of beats per unit of time in the musical segment, an average number of beats per unit of time at the beginning of the musical segment, an average number of beats per unit of time at the end of the musical segment, an average number of beats per unit of time before the beginning of the musical segment, an average number of beats per unit of time after the end of the musical segment, a position of the first beat moment of the seg in the musical recording from which the musical segment is extracted, a position of the second beat moment of the musical segment in the musical recording from which the musical segment is extracted, a position of the third beat moment in the musical recording from which is extracted the musical segment, a position of the fourth beat moment in the musical recording from which the musical segment is extracted, a presence indicator of the beginning of a main voice sequence during the start transition of the musical segment, an indicator of coincidence between the third beat moment with a first musical measurement time of the musical recording, and an indicator of coincidence between the second beat moment with a first musical measure time of the musical recording .

According to one embodiment, the extraction method comprises generating steps for the musical segment of at least one of the following recording files: a recording file of the extracted musical segment in which the end-of-end transitions of the musical segment are short, a recording file of the extracted musical segment in which the start transition of the musical segment is short, and the end transition of the musical segment is long, a recording file of the extracted musical segment in which the transition from the beginning of the musical segment is long, and the end transition of the musical segment is short, and a recording file of the extracted musical segment in which the beginning transitions of the end of the musical segment are long, the start transition or end of the musical segment being long when it extends over a whole number of times the beat period of the musical segment, and being short l when it spans less than one beat period of the musical segment.

According to one embodiment, the musical recording file comprises a musical recording in digitized form, in an uncompressed format such as WAV, AIFF or OGG, or in a compressed format such as MP3, WMA, FLAC, ALAC, AC3. , AAC.

Embodiments may also include a method of concatenating musical segments obtained by the previously-defined extraction method, the method of sequencing comprising steps of: selecting a first and a second musical segment to be chained from among a set of musical segments extracted from musical recordings, determining a start time of the second musical segment by matching the end time of the first musical segment with the end time of the beginning transition of the second musical segment, and associating the second segment the first musical segment overlapping the period between the start time of the second musical segment and the end time of the first musical segment.

According to one embodiment, the linking method comprises steps of reading the first musical segment and triggering the reading of the second musical segment at the start time of the second musical segment determined so as to make the end instant coincide. the first musical segment with the end time of the beginning transition of the second musical segment.

According to one embodiment, the linking method comprises a step of generating a musical recording file in which the recordings of the first musical segment and the second musical segment are merged by combining audio data of the first musical segment and the first musical segment. second musical segment during a period of overlap between the start time of the second musical segment and the end time of the first musical segment coinciding with the end time of the beginning transition of the second musical segment.

According to one embodiment, the linking method comprises a step of generating a musical recording file in which the records of several musical segments are concatenated, with overlapping periods, each overlapping period extending between the start time of a next musical segment and the end time of a preceding musical segment, and comprising audio data of the preceding music segment merged with audio data of the next musical segment, the end time of the musical segment preceding coinciding with the end time of the beginning transition of the next musical segment.

According to one embodiment, the linking method comprises steps of: progressively varying the beat period at the end of the first musical segment being played, to reach a modified value approximating the beat period at the end of start time of the second musical segment, and varying the beat period at the beginning of the second segment, starting from the modified value of the beat period at the end of the first segment being played, so as to reach a value initial of the beat period at the beginning of the second segment.

According to one embodiment, the linking method comprises a step of merging information stored in a synchronization file associated with the first musical segment with information stored in a synchronization file associated with the second musical segment.

According to one embodiment, the second musical segment is selected according to at least one of the following criteria: a ratio of average numbers of beats per unit of time in the first and second musical segments, between 0.9xN and 1, 1 xN, or between 0.9 / N and 1.1 / N, where N is set at 1, 2 or 4, a start transition time of the second segment close to an end transition duration of the first segment, the second musical segment belongs to a musical genre of the first musical segment, and the second segment appeared at the beginning of a musical recording.

Embodiments may also relate to a system configured to implement the extraction or linking method defined above.

Embodiments may also relate to a terminal configured to implement the chaining method defined above.

Exemplary embodiments of the invention will be described in the following, without limitation in connection with the appended figures, in which: FIG. 1 schematically represents a system for extracting musical segments from musical recordings, and assembling such musical segments, according to one embodiment, FIGS. 2 and 3 schematically represent displayed images, during the extraction of segments of a musical recording, according to one embodiment, FIG. 4 represents a form displayed on a display screen during the extraction of segments of a musical recording, according to one embodiment, FIGS. 5A to 5F schematically represent the ends of an extracted musical segment, in different situations, FIG. terminal configured to assemble musical segments, according to one embodiment, FIG. At the beginning end of a first musical segment in relation to an end end of a second musical segment with which the first segment is assembled, FIG. 8 represents steps of a method of assembling musical segments.

FIG. 1 represents a system comprising an SRV server and FT, MT user terminals, which can connect to the SRV server via an NT network such as the Internet network. The SRV server is connected to a DB database and GT operator terminals configured to extract AFE music segments from AF music recordings, according to one embodiment. The SRV server is configured to distribute musical recordings and musical segments thus generated to the FT, MT user terminals. User terminals may include FT personal computers and MT mobile terminals such as "smart" phones (smartphones) and digital tablets.

The system may also include other servers each associated with a database containing a portion of the data stored in the DB database, including AF files, AFE recordings and music segments. The URLs of these servers or the music files they store can be stored in the DB database.

According to one embodiment, at least some of the music recording files AF are each associated with an FM synchronization file. The GT operator terminals are configured to access the DB database via the SRV server, to select and obtain a music recording file AF, to play and playback the selected music recording file, and to generate an MF sync file for the selected music recording file.

Fig. 2 shows an image displayed by one of the GT terminals when the operator has selected a music recording file AF. The image of FIG. 2 shows, in relation to a horizontal time axis, an AWF waveform representative of two stereophonic channels contained in the selected musical recording file. The display area of the AWF waveform has a vertical bar CPT displayed during the sound reproduction of the music recording file. The CPT bar is a cursor moving from left to right, which indicates the time of the AF music recording being played in the displayed AWF waveform. The music recording file includes a musical recording in digitized form, in uncompressed format such as WAV, AIFF or OGG, or in a compressed format such as MP3, WMA, FLAC, ALAC, AC3, AAC, etc. The image of Figure 2 also includes a BTB pace bar, a ZSB zoom bar, an MKB marker bar, and ESB selection bars of segments to extract. Each of these bars indicates particular times in relation to the time axis and the scale of the displayed AWF waveform. The BTB rhythm bar indicates by vertical lines the beat times of the AF recording. The marker bar MKB includes markers MK defining notable parts P1 to P6 of the musical recording, such as an introduction, one or more couplets, a chorus, a bridge and a conclusion. The ZSB zoom bar has two ZSM markers that the operator can move in the ZSB bar to select a time window of the AWF waveform to be displayed specifically on the display screen. When the zoom function is enabled, the display scale of the AWF waveform is changed, and the marker positions in the BTB, ZSB, MKB, and ESB bars are adjusted accordingly. Each of the ESB bars makes it possible to define, in relation to the AWF waveform, time windows of segments to be extracted from the musical recording corresponding to the AWF waveform. Thus, FIG. 2 shows a segment SG1 corresponding to a first occurrence of the part P2 of the musical recording, a segment SG2 extending over the second half of a first occurrence of the part P3 and the whole of the part P4, an SG3 segment extending over the second half of a second occurrence of the P3 portion and all of the P5 portion, and an SG4 segment extending over the third occurrence of the P2 portion and the set of the part P6.

FIG. 3 represents an image displayed by the terminal GT when the operator has selected an area to be displayed using the markers ZSM in the zoom bar ZSB, for example from the image of FIG. 3 shows the AWF waveform of the selected area corresponding to the P1 part and the first occurrence of the P2 part of the musical recording. Only segment SG1 appears in the ESB selection bars of segments to extract.

In FIGS. 2 and 3, the lines delimiting the beginning and the end of each segment SG1-SG4 in the ESB bars are vertical or oblique lines. The oblique lines at the beginning or at the end of the segment indicate a transition of non-zero duration, corresponding to the length of the projection of the line on the horizontal time axis, taking into account the time scale at which the shape is displayed. wave. The vertical lines indicating beat times of the AF recording in the BTB beat bar, may correspond to the first beat of the AF recording's "downbeat".

According to one embodiment, the start transition of a segment is performed applying a maximum sound volume attenuation at the beginning of the transition, and gradually decreasing this attenuation to reach zero attenuation at the end of the transition. Similarly, the end transition of a segment is performed starting with an attenuation of the zero sound volume at the beginning of the transition to reach a maximum attenuation of the sound volume at the end of the transition. Such start and end transitions prevent sound click effects appearing in compressed music clips when they are cut off. The beginning and the end of the melted musical segments appear much more natural and pleasing to the human ear than a cut musical excerpt.

The start and end times of the start and end transitions of the SG1-SG4 segments coincide with a beat, as indicated in the BTB pace bar. Note that the beginning and end of segment transitions can each have a minimum duration corresponding to the period of the beats at the time of the transition in the musical recording. The position and inclination of the boundary lines of segments SG1-SG4 can be modified by the operator directly in the ESB bars.

According to one embodiment, a music recording file AF is associated with an FM synchronization file. The MF file can specify the beat times indicated in the BTB pace bar, for example in milliseconds or tens of milliseconds from the start time of the AF music recording. These instants may for example be specified by the operator by beating the rhythm while listening to the recording, or extracted from the recording by an appropriate acoustic analysis tool.

An example of an FM synchronization file is given in Appendix I. This file is written in a standard language such as JSON (JavaScript Object Notation). Conversely, such an MF file can be read and graphically represented in the form as shown in Figure 2 or 3.

The FM sync file introduces by a keyword 'AudioCase' the designation of a music recording file and rhythm data corresponding to the music recording file AF The MF file includes an 'AudioFile' keyword allowing to introduce a file name of the associated musical recording. The synchronization file MF then comprises: a list of markers introduced by the keyword "listMarkers", a list of labels introduced by the keyword "listTags". a list of zones introduced by the keyword "listZones", a list of beats introduced by the keyword "listBeats", and a list of synchronization files associated with musical segments extracted from the associated musical recording. , this list being introduced by the keyword "NstSubCases".

The list of markers corresponds to the different markers MK introduced in the marking bar MKB. Each marker MK is associated in the list at a position defined for example in milliseconds since the beginning of the musical file. The zone list corresponds to the parts delimited by the MK markers. Each zone of the list is associated with a number 1, 2, 3, 4, 5, corresponding to a type of zone (introduction, verse, chorus, solo, ...). These markers can be displayed by the GT terminal on several lines, each line being assigned to a category of marker, for example: - line 1: markers delimiting the verse / chorus structure of the piece, - line 2: markers identifying repetitive patterns inside the choruses / couplets, - line 3: specific markers introduced by the operator, - line 4: other markers.

The list list "tagsTags" introduces values of different time parameters of the AF music recording, such as beats per minute, average "BPM", minimum "BPM_min", and maximum "BPM_max", a total duration the "total_duration" record. The list of labels includes labels introducing information about the musical recording such as title, artist and album names, musical genre, musical theme, and so on.

The list of beats "listBeats" can specify the respective positions of beats, introduced by the keyword "position", for example in milliseconds since the beginning of the associated musical recording AF.

The selection and activation of one of the segments SG1-SG4 presented in the ESB selection bars triggers the display by the GT terminal of a form such as that shown in FIG. 4. The form of FIG. fields of the beat numbers corresponding to the start times "TSS" and end "TSE" of the start transition of the segment, and beat numbers corresponding to the start times "TSE" and end TEE of the end transition of the segment. The form can also indicate the number of beats per minute "BPM-S" at the beginning of the segment, and "BPM-E" at the end of the segment, and if necessary, a number of beats per minute "BPM-P" d a part preceding the segment in the musical recording, and a number of beats per minute "BPM-N" of a part following the segment. The form can also present "FAD-S" and "FAD-E" indicators respectively indicating whether the start transition and the end transition have a zero duration, that is to say if the "TSS" and "TSS" instants TSE "or" TES "and" TEE "correspond to the same beat number. The different values of the fields "TSS", "TSE", "TES", "TEE", "BPM-P", "BPM-S", "BPM-E", "BPM-N", "FAD-E" , "FAD-S" presented in the form can be determined by the GT terminal as a result of positioning the start and end transitions of a segment SG1-SG4 in the ESB selection bars. These values can then be modified by the operator. From Figure 4, the form may also include boxes for entering the "ODB-S", "ODB-E" flag values specifying whether the "TSE" and "TEE" end of transition times match. or not to a beat of beginning of musical measure, the value of an indicator "ANA-S", specifying if a main voice of the musical recording starts between the instants "TSS" and "TSE" (presence of an anacrouse input). An anacrouse is a note or a sequence of notes preceding a strong beat (for example a beat of beginning of a measure) of a musical sequence. An anacrouse can be part of the melody or accompaniment such as arpeggio setting of a first chord or a short drum roll giving a starting impulse of a musical sequence. Here and in what follows, "main voice" refers to a musical sequence produced by one or several instruments or singers (human voice or instrument with a melodic rather than rhythmic vocation, generally a vector of the main theme of the audio recording), this sequence appearing alone or in the foreground of a musical recording.

The form can also include boxes, for example in the form of checkboxes, for entering the values of different labels appearing in the list list "tagsTags". These tags allow the operator to specify in particular different qualitative information relating to the musical content of the musical recording from which the segment is extracted. Some of these boxes can be filled according to the label values of the "listTags" tag list in the MF file associated with the AF music recording file. These boxes make it possible to specify for example: - a mood (aggressive / sweet-bitter / happy / in love / festive / relaxing / sensual / spiritual), - a tempo (sleep / calm / moderate / fast / very fast), - a theme which can be specified by the decade to which the date of creation of the musical recording belongs, or by a keyword such as "sport", "marriage", "birth", "vacation", "work", "school" "," love "," friendship ", etc. - a genre (classical, rock, disco, reggae, jazz, blues, country, electro, dance, grunge, hip-hop, ...), - a field language if the musical recording includes a song, - a country of origin of the artist, - etc.

The validation by the operator of the form of Figure 4 triggers the generation of the segment by the GT terminal. According to one embodiment, the generation of a segment SG of a musical recording comprises the generation of an AFE record file of the segment, by extracting the part corresponding to the segment in the file AF containing the musical recording, and generating an associated synchronization file SF, from the FM synchronization file of the musical recording. This generation operation is based on the start and end times of the segment "TSS", "TEE" of the segment and end times "TSE" and start "TES" of the start and end transitions.

An example of a synchronization file SF associated with an SG segment extracted from a musical recording AF is given in Appendix II. This file is also written in a standard language such as JSON. A derived SF synchronization file is generated from a main synchronization file MF, taking up the list of tags "listTags" of the synchronization file MF, modifying the numbers of the first and last beats to take those of the extracted segment , resuming the list of beats "listBeats" by keeping only the positions of the beats in the extracted segment. Additional tags, some of which come from the form in Figure 4, can be added to the "listTags" list of the SF file, such as: - "isStart" indicating whether the extracted segment includes the beginning of the musical recording from where it is extracted, - "isEnd" indicating whether the segment includes the end of the musical recording from which it is extracted, - "StartLevel" indicating a volume level at the beginning of the segment (typically during the first 100 to 300 ms after the start of the segment transition), - "EndLevel" indicating a volume level at the end of the segment (typically during the last 100 to 300 ms of the segment before the end-of-segment transition), - "BPM" indicating an average number beats per minute in the segment, - "BPM-S" indicating a number of beats per minute at the beginning of the segment (typically during the first 100 to 300 ms), - "BPM-E" indicating a number of beats per minute at the end of the segment (typically the last 100 to 300 ms), - "BPM-P" indicating a number of beats per minute (typically during the last 100 to 300 ms) before the beginning of the segment, - "BPM-N" indicating a number of beats per minute minute (typically during the first 100 to 300 ms) after the end of the segment, - "TSS" indicating the number, in the musical recording from which the segment is extracted, of the first beat of the segment, - "TSE" indicating the number, in the musical recording from which the segment is extracted, of the ending beat of the transition from the beginning of the segment, - "TES" indicating the number, in the musical recording from which the segment is extracted, of the beat beginning of the end-of-segment transition, - "TEE" indicating the number in the musical record from which the segment is extracted from the last beat of the segment, - "ANA-S" indicating whether a main voice sequence of the segment begins during the segment start transition, - "ODB-S" indicating s at the instant TSE of the segment coincides or not with a start beat of musical measurement of the musical recording, - "ODB-E" indicating whether the TEE instant of the segment coincides or not with a start beat of musical measure.

FIGS. 5A, 5B, 5C, 5D, 5E and 5F show different start and end cases of SG segments extracted from a musical recording. Figs. 5A to 5F also show the MVC waveform of a main voice of the musical recording. Figures 5A and 5B correspond to correct segment splitting cases. Figure 5A illustrates the most common case of a main voice sequence MVC of the musical recording beginning after the end time TSE of the start transition TS of the segment SG. In the case of FIG. 5A, a main voice ends before the end time TEE of the end transition TE of the segment SG. The times TSE and TEE correspond to beats of beginning of musical measure (ODB-S and ODB-E = "true"). In the particular case of FIG. 5A, the start times TSS and end TSE of the start transition TS of the segment SG are substantially identical (TSS = TSE, that is to say FAD-S = "false") and the start times FES and end TEE of the end transition TE of the segment SG are different (TES ^ TEE, that is, FAD-E = "true"). In practice, the start and end transitions have a minimum duration of between 20 and 40 ms. This arrangement makes it possible to avoid the effects of unwanted noise (sound click) appearing in case of abrupt interruption of a sound recording.

FIG. 5B illustrates the case where the end time TSE of the start transition TS of the segment SG corresponds to a start beat of musical measurement (ODB-S = "true") and where a main voice sequence MVC of the Music recording starts between the TSS start and TSE end times of the start transition TS of the SG segment (input anacrouse). In the case of Figure 5B, a main voice ends before the TEE end time of the end transition TE of the SG segment.

FIGS. 5C, 5D, 5E and 5F illustrate cases of segment splitting to be avoided. In the case of FIGS. 5C to 5E, a PTC portion of a main voice MVC sequence of the musical recording is cut off. In the case of FIG. 5C, the main voice MVC sequence of the musical recording ends after the start time TSS of the start transition TS of the segment SG. In the case of FIG. 5D, the main voice MVC sequence of the musical recording begins before the start time TSS of the start transition TS of the segment SG. In the case of FIG. 5E, the main voice MVC sequence of the musical recording ends after the end time TEE of the end transition TE of the segment SG. In the case illustrated in FIG. 5F, a main voice sequence MVC1 terminates after the start of a main voice sequence MVC2. If the TSS start of the SG segment is placed just before the start of the MVC2 main voice sequence, the MVC1 sequence is cut off, which is undesirable.

It follows from FIGS. 5A to 5F that the procedure to be applied to perform a correct extraction of a segment consists in locating pause periods between main voice sequences, in a musical recording, and thus in locating start times and end of these sequences. This procedure then consists of locating beat times in the vicinity of these start and end times of the sequence during the pause periods. Then simply place the start and end times of the TS start and TE end transitions of the segment on beat times, as follows. The start time TSS of the start transition TS of the segment (i.e., the start time of the segment) may be placed before the start time of the identified main voice sequence, but after end of a preceding previous main voice sequence. The end time TSE of the start transition TS can be placed either at the instant TSS or after the beginning of the main voice sequence identified, for example at the moment of a strong beat (first time of a musical measure). The start time TES of the end transition TE of the segment can be placed before the end of a marked main voice sequence, and the end time TEE of the end transition TE (that is to say the end time of the segment) can be placed after the end of the sequence, but before the beginning of a next main voice sequence.

The segments thus obtained can be assembled together or their listening can be chained, with smooth transitions between the segments. Thus, playback of two AFE music recording segment files can be chained to one of the FT, MT, GT terminals. For this purpose, the terminal can perform a first step of asking the user to choose a first musical segment, for example in a list of available segments meeting different selection criteria previously introduced by the user. Following the selection of a first segment by the user, the terminal sends a request to the SRV server to obtain a list of one or more segments for example meeting the selection criteria previously specified by the user. This request includes for example an identifier of the first selected segment, or the information as specified for the first segment in the associated synchronization file SF, which are useful for selecting a second segment to be read after the first segment. Upon receipt of this request, the SRV server searches the DB database segments that can be chained smoothly following the first segment selected. For this purpose, the server SRV can apply some of the following rules: - the ratio between the number of beats per unit of time "BPM-E" at the end of the first segment and "BPM-S" at the beginning of the second segment, is for example between 0.9xN and 1.1xN, or between 0.9 / N and 1.1 / N, N being typically set at 1.2 or 4, - the ratio of the volume "StartLevel" at the beginning of the second segment, and the volume "EndLevel" at the end of the first segment is for example between 0.85 and 1.15.

The criterion on the number of beats can alternatively be applied to the number of beats "BPM-N" of the first segment, compared to the number of beats "BPM-S" of the second segment, or the number of beats "BPM-E" the first segment compared to the number of beats "BPM-P" of the second segment, or the number of beats "BPM-N" of the first segment, compared to the number of beats "BPM-P" of the second segment.

It is further preferable that the end transition TE of the first segment SG1 has a duration at least close to the start transition of the second segment SG2. Thus, a segment can be selected as the second segment SG2 if the duration DTS2 of its start transition TS2 is for example in the range: [DTE1 -100ms-10xE ((DTE1 / 100ms) -1), DTE1 + 100ms + 10xE ((DTE1 / 100ms) -1)] where DTE1 is the duration in ms of the end transition of the first segment, and E (x) is the integer part of the number x.

In addition, the SRV server can order the list of segments that can be assembled with the first selected segment, in order of numbers of satisfied criteria among the criteria defined above, as well as other criteria such as: the musical genre, the mood , the tempo, the theme, the sung language, etc. The SRV server can also take into account a criterion relating to the values of the labels isStart "and" isEnd "Thus, musical segments associated with labels" isStart "and" isEnd "respectively with the values" true "and" false ", are preferred to appear at the top of a list of segments to be chained in. Music segments with "isStart" and "isEnd" tags set to "false" are preferred for inclusion in this list between a header segment and a end segment of the list, and musical segments associated with labels "isStart" and "isEnd" respectively to the values "false" and "true", are preferred to appear at the end of this list of segments to be chained.

When the user selects a second segment SG2, then other possible segments, the terminal GT, MT, FT constitutes an ordered list of segments to be chained. For the selection of an additional segment, the last segment of the list is considered as the first segment in the steps previously described. In order to be able to read and restore the segments of such an ordered list of segments to be linked, the terminal can be configured as shown in FIG. 6. In FIG. 6, the terminal simultaneously executes two PLY1, PLY2 instances of a reading program. of musical recordings, as well as a sequencer SEQ. The sequencer SEQ is configured to sequence the playing of several musical segments, specified in the ordered list, by matching the end time TEE1 of the segment SG1 being read with the end time of transition TSE2 of start TS2 of the segment following SG2, as shown in Figure 7. Thus, during the period TSS2-TEE1 (equal to the period TSS2-TSE2) the two segments SG1 and SG2 are read at the same time.

Thus, the sequencer SEQ is configured to successively execute steps S11 to S19 shown in FIG. 8. In step S11, the sequencer SEQ initializes an index n to one. In step S12, the sequencer SEQ opens an LSSG list of segments to be chained and reads an identifier of a first segment to be chained. In step S13, the sequencer SEQ opens the synchronization file SF1 associated with the first segment SG1 identified in the list LSSG. In step S14, the sequencer SEQ activates the reading by the first instance PLY of the reading program, of the file AFE1 containing the recording of the first segment of the list LSSG. In step S15, the sequencer SEQ increments the index n by one. In step S16, the sequencer SEQ reads the list LSSG to determine if it comprises a segment to be chained next, corresponding to the index n. If the end of the LSSG list is reached, the procedure executed by the sequencer SEG terminates. In the opposite case, the sequencer executes step S17 where it opens the synchronization file SF2 corresponding to the next segment (corresponding to the index n) identified in the LSSG list. In step S18, the sequencer SEQ determines, as illustrated in FIG. 7, the instant TSS of the beginning of the next segment SG2 in the time scale of the segment SG1 currently being read, with respect to the instant TEE of end of this segment, so as to synchronize the instant TEE of the segment being read SG1 with the instant TSE of the next segment SG2. For this purpose, the sequencer SEQ reads the respective positions of the TEE1 instants of the segment SG1 and TSS2 and TSE2 of the next segment SG2 in the SF1 and SF2 files associated with the segment being read SG1 and the next segment to link SG2. In step S19, the sequencer SEQ waits for the instant TSS2 of the segment SG2, thus determined in the time scale of the reading of the segment SG1. At time TSS2, the sequencer executes steps S14 to S19 again. In particular, in step S14, it triggers the reading of the file AFE2 containing the recording of the segment SG2 by an open instance of the reading program, that is to say the instance PLY2 if the instance PLY2 is in train the SG1 segment between times TSS2 segment SG1 and TEE2 segment SG1.

When two successive segments have, in the vicinity of their overlapping TE, TS transitions, different beat periods, the sequencer SEQ can be configured to progressively vary the beat period at the end of the segment being played, to get closer of the beat period at the beginning of the next segment, and to vary the beat period at the beginning of the next segment, starting from the last modified value of the beat period at the end of the segment being played, so as to reach the initial value of the beat period at the beginning of the next segment. Segment beat periods can be modified, for example, by deleting or duplicating samples of the audio signal in the AFE files for recording the musical segments.

The GT, FT, MT terminal can also be configured to assemble the AFE recording files of the segments designated in the LSSG list into a single music recording file by concatenating the audio data outside the overlapping periods TSS2-TEE1 ( = TSS2-TSE2) and merging the audio data during the overlapping periods. The synchronization files SF1 and SF2 of segments SG1, SG2 thus assembled can also be merged to form a single synchronization file associated with the merged musical recording file. For this purpose, the information used as criteria for sequencing a segment before the first merged segment and for sequencing a segment after the last merged segment are kept in the new synchronization file associated with the registration file. resulting from the merger.

In a simplified embodiment, the GT terminal can assemble a plurality of music record segment files regardless of the label values "isEnd", "StartLevel", "EndLevel", and "BPM".

According to another embodiment, when the synchronization file SF comprises parameters "isSTART" and "isEND" to "false", one to four pairs of SF and musical synchronization files AFE can be generated: 1) a first couple with the parameters "isSTART" and "isEND" to "false", 2) a second pair with the parameter "isSTART" to "true" and the parameter "isEND" to "false", the corresponding music recording file AFE being modified by forming a long start transition, 3) a third pair with the parameter "isSTART" set to "false" and the parameter "isEND" to "true", the corresponding musical record file being modified by forming a transition therefrom. long end, 4) a fourth pair with the parameters "isSTART" and "isEND" to "true", for which the corresponding musical recording file has been modified by forming long start and end transitions.

The beginning or end transition of the musical segment being long when it extends over a whole number of times the beat period of the musical segment, and being short when it extends over less than one beat period of the musical segment. musical segment. In this way, the fourth pair segment (having parameters "isSTART" and "isEND" to "true") can be read separately (without being joined to other segments). The segment of the second pair can be inserted at the head of a list of segments to be linked together, and the segment of the third pair can be inserted at the end of a list of segments to be chained.

Of course, other forms of fade such as increasing or decreasing linear attenuation can be applied to the musical segments. For example, sinusoidal attenuation can be applied to musical segments during their start transition, especially when the duration of the transition is less than the beat period. Such a fade keeps the audio energy and therefore the sound volume. ANNEX I (forming part of the description)

Content snippet of an MF sync file {MAudioCase ": {MAudioFile": "06 xxxx.aac", "barlnfo": {"listTags": "masterframe_id": 0, // sync file identifier "BPM_moy" : 118, // average number of beats per minute "BPM_min": 7, // number bpm minimum "BPM_max": 444, // number bpm maximum "total_duration": 191, // total duration of the corresponding musical recording " bit rate ": 128, // bit rate in kb / s of the corresponding musical record" version_fg ":" V401 ", // version of the file generation software" song_title ":" ttt "," major ":" mmm " , "album": "bbb", "artist": "aaa", // references of // the corresponding musical record "categories": // definition of categories to which {"hit_of_the_moment_end": "ddd", // belongs the recording "best_year": null, // corresponding musical "mood": ["mood_party", "mood_bittersweet"], "pace": ["pace_fast_rough"], "release_year": yyy, "countries": cc, " sung ": dd," gem ": false," deluxe ": true," cult ": false," instrumental " : false, "genre": ["genre_techno", "genre_dance"], "theme": ["theme_2010"], "hit_of_the_moment_start": "ddd", "comment_3": "", "comment_4": "", " comment_2 ":" "," comment_1 ":" "} // fields reserved for inserting comments {" NstMarkers ": [{" marker ": 277}, {" marker ": 2756}, {" marker ": 31915}, {"marker": 37910}, {"marker": 43075}, {"marker": 48807}, {"marker": 53820}, {"marker": 59954}, {"marker": 65551} , {"marker": 71181}, {"marker": 76093}, {"marker": 82395}, {"marker": 88009}, {"marker": 93637}, {"marker": 98453}, { "marker": 210852}, {"marker": 233575}, {"marker": 255965}, {"marker": 258799}, {"marker": 260257}, {"marker": 261726}, {"marker" ": 264374}, {" marker ": 265770}, {" marker ": 267167}, {" marker ": 270011}, {" marker ": 271376}, {" marker ": 272824}, {" marker ": 275606}, {"marker": 277001}, {"marker": 278370}, {"marker": 279044}, {"marker": 300436}], "listZones": [{"zone": 2}, { "zone": 1}, {"zone": 0}, {"zone": 1}, {"zone": 3}, {"zone": 3}, {"zone": 5}, {"zone" ": 2}, {" zone ": 3}, {" zone ": 3}, {" zone ": 5}, {" zone ": 4}, {" zone ": 2}, {" zone ": 4}, { "zone" 2}, { "zo ne ": 4}, {" zone ": 3}, {" zone ": 3}, {" zone ": 3}, {" zone ": 3}, {" zone ": 3}, {" zone " : 3}, {"zone": 3}, {"zone": 3}, {"zone": 6}, {"zone": 6}, {"zone": 5}, {"zone": 5 }, {"zone": 4}, {"zone": 4}, {"zone": 2}, {"zone": 2}, {"zone": 0}, {"zone": 6}] , "listBeats": [{"position": 286}, {"position": 2709}, {"position": 4544}, {"position": 6400}, {"position": 7316}, {"position" : 9262}, {"position": 10188}, {"position": 12000}, {"position": 12963}, {"position": 14764}, {"position": 17589}, {"position": 19541 }, {"position": 324278}, {"position": 324673}, {"position": 325079}, {"position": 325486}],}} ANNEX II (forming part of the description)

Extracted content of a derived synchronization file {"SubFrameJSON": {"listTags": "subframe_id": 20150317154219001, // identifier of the file "masterframe_id": 20150317154219000, // identifier of the original file "sframe_label": " intro_1_part_1 ", // tag of the extract of // the corresponding musical record" subframe_previous_id ": null, // identifier of the previous file // extract from the original main file" subframe_next_id ": null, // file identifier next // extract from the original main file "duration": 38, // duration of the segment "total_duration": 191, // duration of the musical recording // corresponding to the original file "debit": 128, / / bit rate in kb / s of the corresponding "BPM" music recording: 118, "BPM-P": 118, "BPM_S": 118, "BPM_E": 118, "BPM_N": 118, "TSS": 12000, "TSE": 12001, "TES": 12962, "TEE": 12963, "ANA-S": false, "ANA-E": false, "ODB-S": true, "ODB-E": false, "isEnd": false, "isStart": true, "startLeve l ": 6," endLevel ": 17," version_fg ":" V401 "," artist ":" aaa "," categories ": {" hit_of_the_moment_end ":" ddd "," best_year ": null," mood ": ["mood_party", "mood_bittersweet", "mood_sad"], "pace": ["pace_fast_rough", "pace_cool"], "release_year": yyy, "countries": ["gb", "fr"], "sung ": [" uk "]," gem ": false," deluxe ": true," cult ": false," instrumental ": false," genre ": [" genre_techno "," genre_dance "," genre_pop "," genre_techno "]," theme ": [" theme_2010 "]," hit_of_the_moment_start ":" ddd "}," album ":" aaa "," song_title ":" ttt "," major ":" mmm "," comment_3 " : "", "comment_2": "", "comment_1": "", "comment_4": "", "stereo": true} "listBeats": {"position": 12000}, ... {"position" : 12963}]}}

Claims (16)

A method of extracting a musical segment from a musical recording file, the method comprising the steps of: identifying beat times in a musical recording (AF), identifying in the musical recording of first and second successive periods of absence of main voice (MVC), identifying in the first period a first beat moment, identifying in the second period a second beat moment, extracting in a file (AFE) a musical segment (SG) of the musical recording between a start time (TSS) of the musical segment corresponding to the first beat time identified, and an end time (TEE) of the musical segment corresponding to the second beat time identified, and to form in the extracted musical segment a start transition (TS) extending from the beginning of the musical segment to a third beat (TSE), during which a decay issante is applied to the sound volume at the beginning of the musical segment, and an end transition (TE) extending from a fourth beat (TES) to the end of the segment, during which increasing attenuation is applied to the sound volume at the end of the segment. end of the musical segment. 2. Extraction method according to claim 1, wherein the third (TSE) and second beat times (TEE) coincide with first musical measurement times of the musical recording (AF). 3. Extraction method according to claim 1 or 2, wherein the decreasing attenuation applied during the start transition (TS) of the musical segment (SG) has a maximum value at the start time (TSS) of the musical segment. (SG), and a zero value at the third beat (TSE). 4. Extraction method according to one of claims 1 to 3, wherein the increasing attenuation applied during the end transition (TE) of the musical segment (SG) has a zero value at the fourth beat (TES) of beat, and a maximum value at the end time (TEE) of the musical segment. 5. Extraction method according to one of claims 1 to 4, comprising steps of recording in a synchronization file (SF) associated with the segment (SG) of at least one of the following information: a list of positions of Moments of beatings of the musical segment, a sound volume at the beginning of the musical segment, a sound volume at the end of the musical segment, a position indicator of the musical segment at the beginning of the music recording file, a position indicator of the music segment at the end of the musical recording file, an average number of beats per unit of time in the musical segment, an average number of beats per unit of time at the beginning of the musical segment, an average number of beats per unit of time at the end of the musical segment , an average number of beats per unit of time before the beginning of the musical segment, an average number of beats per unit of time after the end of the musical segment, a position of the first moment of beat of the musical segment in the musical recording from which the musical segment is extracted, a position of the second moment of beat of the musical segment in the musical recording from which the musical segment is extracted, a position of the third moment of beat in the musical recording from which the musical segment is extracted, a position of the fourth beat moment in the musical recording from which the musical segment is extracted, a presence indicator of the beginning of a main voice sequence during the transition from beginning of the musical segment, a coincidence indicator between the third beat moment with a first musical measure time of the musical recording, and an indicator of coincidence between the second beat moment with a first musical measure time of the musical recording. 6. Extraction method according to one of claims 1 to 5, comprising generating steps for the music segment (SG) of at least one of the following recording files (AFE): a segment recording file musical score in which the start-of-end transitions (TS, TE) of the musical segment are short, a recording file of the extracted musical segment in which the start transition of the musical segment is short, and the end-of-segment transition musical is long, a recording file of the extracted musical segment in which the beginning transition of the musical segment is long, and the end transition of the musical segment is short, and a recording file of the extracted musical segment in which the transitions the beginning of the end of the musical segment are long, the beginning or end transition of the musical segment being long when it extends over a whole number of times the period of beat of the segment musical, and being short when it spans less than one beat period of the musical segment. 7. Extraction method according to one of claims 1 to 6, wherein the music recording file (AF) comprises a musical recording in digitized form, in an uncompressed format such as WAV, AIFF or OGG, or in a compressed format such as MP3, WMA, FLAC, ALAC, AC3, AAC. 8. Process for linking musical segments obtained by the extraction method according to one of claims 1 to 7, the linking method comprising the steps of: selecting a first and a second musical segment (SG1, SG2) to sequence among a set of musical segments extracted from musical recordings, determining a start time (TSS2) of the second musical segment by making the end time (TEE1) of the first musical segment coincide with the end time (TSE2) the start transition (TS1) of the second musical segment, and associating the second musical segment with the first musical segment by overlapping the period between the start time of the second musical segment and the end time of the first musical segment. The method of sequencing according to claim 8, comprising steps of playing the first musical segment (SG1) and initiating playback of the second musical segment (SG2) at the start time of the second musical segment (TSS2) determined from in order to match the end time (TEE1) of the first musical segment with the end time (TSE2) of the start transition (TS1) of the second musical segment. 10. A method of linking according to one of claims 8 and 9, comprising a step of generating a music recording file (AFE) in which the records of the first musical segment (SG1) and the second musical segment (SG2 ) are merged by combining audio data from the first musical segment and the second musical segment during an overlap period (TSS2-TEE1) between the start time (TSS2) of the second musical segment and the end time (TEE1) of the first musical segment coinciding with the end time (TSE2) of the start transition (TS2) of the second musical segment. 11. A method of linking according to one of claims 8 to 10, comprising a step of generating a music recording file (AFE) in which the records of several musical segments are concatenated, with overlapping periods (TSS2 -TEE1), each overlap period extending between the start time (TSS2) of a next musical segment and the end time (TEE1) of a preceding musical segment, and comprising audio data of the music segment previous merged with audio data of the next musical segment, the end time of the preceding musical segment coinciding with the end time (TSE2) of the start transition (TS2) of the next musical segment. 12. A method of linking according to one of claims 8 to 11, comprising steps of: gradually varying the beat period at the end of the first musical segment (SG1) being read, to achieve a modified value approximating the beat period at the start time (TSS) of the second musical segment (SG2), and varying the beat period at the beginning of the second segment, starting from the modified value of the beat period at the end of the first segment being played, so as to reach an initial value of the beat period at the beginning of the second segment. 13. A method of linking according to one of claims 8 to 12, comprising a step of merging information stored in a synchronization file (SF1) associated with the first musical segment (SG1) with information stored in a synchronization file (SF2) associated with the second musical segment (SG2). 14. A method of linking according to one of claims 8 to 13, wherein the second musical segment (SG2) is selected according to at least one of the following criteria: a ratio of average numbers of beats per unit of time in the first and second musical segments, between 0.9xN and 1.1xN, or between 0.9 / N and 1.1 / N, where N is set at 1.2 or 4, a start transition time of the second segment close to an end transition duration of the first segment, the second musical segment belongs to a musical genre of the first musical segment, and the second segment appears at the beginning of a musical recording. 15. System configured to implement the method according to one of claims 1 to 14. Terminal configured to implement the linking method according to one of claims 8 to 14.