FR2958068A1 - Method for synthesizing audio signal from note signal representative of sequence of note levels on cord of e.g. bass guitar, involves producing synthesized audio signal by mixing harmonized tonal signal and contact audio signal - Google Patents

Abstract

The method involves actuating variation of note signal toward a new note level (135) for a note changing and exciting contact. A canalized component of a harmonized tonal signal is modulated at harmonic frequency i.e. synthesis frequency, from the new note corresponding to a harmonic order different from another harmonic order of preceding synthesis frequency (132) of the canalized component. A synthesized audio signal is produced by mixing the harmonized tonal signal and a contact audio signal of the note changing and exciting contact. Independent claims are also included for the following: (1) a device for synthesizing an audio signal (2) a recording medium for recording a computer program comprising a set of instructions for implementing a method for synthesizing an audio signal (3) a computer program comprising a set of instructions for implementing a method for synthesizing an audio signal.

Description

The invention relates to a method for synthesizing an audio signal, called a synthesized audio signal, from a signal, called a signal of a note, representative of a pitch sequence according to which the vibrating member is vibrated, and of a sequencing signal representative of a sequence of contacts imparted on a vibrating member, said contact sequence comprising excitation contacts capable of vibrating the vibrating member, wherein at least one audio contact signal is produced for each of said excitation contacts, and a plurality of frequency components, called modulated components, of at least one audio contact signal, said tonal contact signal, are each modulated successively around frequencies, said synthesis frequencies, each own a harmonic rank of a note pitch of the sequence. The invention extends to a synthesis device. The digital musical instrument interface protocol, called the MIDI protocol, makes it possible to represent the game of an instrumentalist in the form of a sequence of events in time, each described according to a predetermined number of descriptors (instant of triggering of a sound, pitch, intensity ...). This protocol is particularly adapted to describe the play of a keyboardist and has been widely adopted as such. There are known devices, known as traditional MIDI converters, adapted to implement an audio conversion to MIDI, in order to be able to control a compatible MIDI synthesizer, from musical instruments with vibrating organs (eg guitar strings ) entering free vibration following the realization of excitation contacts imparted on the vibrating member. These devices comprise connecting means with at least one microphone mounted on the musical instrument, the microphone providing for each vibrating member, a signal, said microphone signal, representative of the vibration of the vibrating member.

Traditional MIDI converters produce MIDI messages that control a MIDI compatible synthesizer from the microphone signal. To do this, the traditional MIDI converters are adapted to detect at any time whether the microphone signal meets predefined criteria for detecting a given excitation contact on the vibrating member, thereby triggering a synthesized sound. In addition, traditional MIDI converters are adapted to detect at any time during the free vibration of the vibrating member, if the microphone signal meets predefined criteria for detecting a total choking contact on the vibrating member. by the instrumentalist, triggering as such an interruption of a synthesized sound in progress. If necessary, an interruption of the synthesized sound in progress is also controlled following a new detected excitation contact succeeding directly to a previous detected excitation contact, without prior detection of a total choking contact. For example, there are known traditional MIDI converters employing an excitation contact detection, called an attack detection, in which a detected value of overall variation of the vibration intensity of the vibrating member is compared to a predetermined threshold of amplification of the vibratory intensity, said threshold of detection of attack, beyond which a new synthesized sound is produced. There is a need to optimize the sensitivity of the detection of the excitation contacts, without this compromising the fidelity of the representation of the playing of the instrumentalist formed by the synthesized sounds. A faithful translation of the instrumentalist's performance at the output of the MIDI-compatible synthesizer is obtained only when the instrumentalist performs excitation contacts according to a note-attacking technique of the adapted musical instrument taking into account the criteria of attack detection of the traditional MIDI converter. The successive execution of pinches of a guitar string makes it possible, for example, to produce using a traditional MIDI converter for a guitar and a MIDI compatible synthesizer, a sequence of synthesized sounds representative of the sequence of pinches performed. However, each nip must be executed neatly and distinctly, otherwise there will be some nibbling with no audible consequences at the output of the MIDI compatible synthesizer.

An attack detection threshold setting must be made to improve the sensitivity of the traditional MIDI converter's pick-up detection to account for excitation contacts on the vibrating member, resulting in an overall amplification of the drive. relatively low vibratory intensity. Such an adjustment, however, increases the risk of inadvertent detection of an excitation contact while the instrumentalist has not touched the vibrating member. Indeed, an overshoot of the attack detection threshold may then occur under the effect of non-stationary vibratory components in free vibration (out of contact with the vibrating member): phenomena of vibration of the string vibration, slight excitation of the vibrating organ by sympathetic vibration, slight excitation of the vibrating organ by absorption of a sound wave propagating in the air or transmitted to the vibrating organ at the attachment points of the vibrating member on the musical instrument ... The corresponding untimely synthesized sound is all the more unfortunate as it unexpectedly interrupts a synthesized sound that may be present during production. In addition, the effective detection of excitation contacts performed according to other attack techniques that by pinching the rope, in particular ornamentation techniques (rhythmic ornamentations, friction imparted on the rope ...), is uncertain . In any case, the specifications of the MIDI protocol do not provide for translating the game nuances corresponding to the aforementioned techniques of ornamentation, the nuances corresponding to the various techniques of attack of the string (striking, pinching in stop said apoyando, pinching in pulling said tirando, pinching by means of a pick ...), the shades corresponding to the different places on the rope where are performed said techniques of ornamentation and attack ... In particular, the execution of techniques of d ornamentation is likely to lead to the production of synthesized sounds that are often untimely from the instrumentalist's point of view, in any case very unrealistic, that is to say, not at all representative of the sound expected by the instrumentalist in view of the manipulation performed on the rope, so quite inappropriate.

In particular, there are known traditional MIDI converters using a detection of the vibration height of the vibrating member from the microphone signal to produce a sequence of synthesized sounds translating the melodic phrasing imparted on the vibrating member by the instrumentalist. Such detection is advantageously implemented for instruments whose pitch is determined by direct manipulation of the vibrating organ (a guitar string for example), without the aid of a mechanism interposed between the instrumentalist and the vibrating organ. Indeed, a representative synthesis of note pitch changes and modulations of each note pitch performed by the instrumentalist is then implemented with a minimal bulk of the musical instrument. By way of example, in the case of the guitar, said changes in note pitch can be made by left-handed playing techniques and said modulations of the pitch of notes can be made by lateral traction of the string in the case of a guitar. However, any contact on the vibrating member introducing inharmonic vibratory components and / or modifying the relative amplitudes of harmonic vibratory components of the vibrating member, is likely to cause an inadvertent change in the note height detected while the The instrumentalist did not make a voluntary gesture for such a modification. By way of example, contact of the vibrating member in free vibration at a vibration node of a low frequency vibratory mode, for example by impact of a finger at this point, is likely to cause a inadvertent change in detected note pitch for the rest of the vibration duration of the additional detected non-contact cord. In addition, the transient vibration components introduced by any short or prolonged contact on the string may cause temporary untimely changes in the detected note pitch. An inadvertent change in the pitch of the detected note is all the more unfortunate that occurs during the production of a synthesized sound previously triggered while the latter is not interrupted by the contact that caused the inadvertent change Indeed , the inadvertent change then causes an effect of inadvertent change of the pitch of said synthesized sound, or a sound effect, said untimely break of resonance, transient but sudden and unsightly variation of the timbre of said synthesized sound. As a result, in one case as in the other, a diminution of the fidelity of the representation of the play of the instrumentalist formed by the synthesized sounds. In view of the foregoing, the instrumentalist is forced to strip his game of any excitation contact that is not adapted to the predefined attack detection criteria, in particular contacts likely to cause an inadvertent change in the detected note pitch. , in order to be able to control completely the synthesis. This results in a decrease in the expressive potential of the musical instrument. To take the measure of the degree of alteration of the musical discourse, it suffices for example to imagine what would become any piece of guitar after it was stripped of the aforementioned effects: a succession of sounds triggered intensities and variable durations, but all of which have the same timbre. Synthesizers controlled by traditional MIDI converters inherit the very mechanical and disembodied character of cheap electronic keyboards. There is therefore a need to produce for any excitation contact on the vibrating organ, synthesized sounds faithfully and realistically representing said excitation contact, unlimited in the game technique employed. In addition to the guitar given by way of example, this problem is felt more generally for musical instruments without mechanisms interposed between the vibrating organ and the instrumentalist so that a great deal of freedom of execution is granted to the composer. instrumentalist to excite the vibrating organ. The possibilities of play are then no longer countable, the player can freely choose the manipulations to be performed, the objects with which to perform these manipulations if necessary .... The invention aims to overcome these disadvantages.

In particular, the invention aims to enable the performance of any traditional instrumental technique of the musical instrument without the result of effects of inadvertent resonance failure or inadvertent change effect of the height of notes synthesized sounds. In particular, the invention aims to avoid any extension of a detection time of the current pitch. In addition, the invention aims to allow to take into account a greater number of excitation contacts imparted on the vibrating member without compromising the fidelity of the representation of the game of the instrumentalist formed by the synthesized sounds. In addition, the aim of the invention is to carry out this translation in a more representative manner, for each of said excitation contacts imparted taking into account the nuances with which etching techniques and ornamentation techniques can be performed on the vibrating member. .

The invention also aims to provide such a solution that can be implemented at an economical cost price, in particular that it can be implemented on an electronic and / or computer device made up of inexpensive components, in particular computer components and / or generic electronic commerce.

The invention also aims to propose a device and a method for synthesizing an audio signal from a sequence of excitation contacts imparted on a vibrating member and capable of vibrating the vibrating member, which are compatible with a real-time implementation, without unacceptable delay between a contact on the vibrating body and its effect in the audio synthesis signal. The invention also aims to achieve these goals in a technically simple way to develop, especially for several categories of musical instruments without significant additional costs for its adaptation between different categories of instruments, for example the guitar, the bass guitar, the guitar. banjo, mandolin, violin, cello, double bass, viola etc.

To do this, the invention relates to a method of synthesizing an audio signal, said synthesized audio signal, from a signal, said note signal, representative of a pitch sequence of notes according to which the vibrating member is vibrated, and a sequencing signal representative of a sequence of contacts imparted on a vibrating member, said contact sequence comprising excitation contacts able to vibrate the vibrating member, in which process: at least one audio contact signal is produced for each of said excitation contacts, - several frequency components, called modulated components, of at least one contact audio signal, called tonal contact signal, are each modulated successively around frequencies, said synthesis frequencies, each of which is at a harmonic rank of a note pitch of the sequence, characterized in that: - for at least one excitation contact, said excitation contact and of change of note, resulting in a variation of the note signal to a new note pitch in the sequence, at least one modulated component, called channelized component, of a tonal contact signal, called harmonized tonal signal, resulting from a contact of excitation prior to said excitation contact and change of note, is modulated around a harmonic frequency, called new frequency of synthesis, of the new pitch of note, corresponding to a harmonic rank different from the harmonic rank of a previous frequency of synthesis of the channelized component, the synthesized audio signal after said excitation contact and note change is produced by mixing each harmonized tonal signal and the audio contact signal of said excitation contact and change of note. The invention therefore makes it possible to produce, in response to said excitation and change of note contact, a new contact tone signal supplying the synthesized audio signal with frequency contributions distributed according to the harmonic frequencies of the new note height detected, without requiring to interrupt any tonal contact signal resulting from an excitation contact prior to the excitation contact and change of note. Indeed, the modulated components of a harmonized tonal signal according to the invention are, following the change of note pitch, frequency-divided according to the new note pitch so that they harmoniously combine with any tonal signal resulting from the touch contact excitement. This distribution can be provided by means of the invention moderately modulating each channelized component following said change in pitch. Therefore, the invention makes it possible to prevent any unwanted resonance or inadvertent change in the pitch of the harmonized tonal signal resulting from an inadvertent change in pitch by the effect of the excitation contact and the change of note on the note. the vibrating organ. In view of the foregoing, the invention makes it possible to take account of a greater number of excitation contacts imparted on the vibrating member, without compromising the fidelity of the representation of the playing of the instrumentalist formed by the synthesized sounds. . Throughout the text, the term "harmonic frequency" is, unless otherwise stated, used in its broadest sense, referring to frequencies that are not necessarily exact integer multiples of a corresponding fundamental frequency, taking into account possible phenomena of inharmonicity. Moreover, the fundamental frequency forms according to this terminology, a harmonic frequency of rank 1. Advantageously and according to the invention, the note signal is produced from a signal, said signal source of detection of note pitch, representative of the vibration of the vibrating organ. More particularly, advantageously and according to the invention, the note signal is produced by note height detection as the reception of the pitch detection source signal is received, in particular by iterative execution of a step of height detection of the pitch detection source signal.

Throughout the text, the term "signal" is used according to its functional acceptance without limit as to the mode of representation of the signal employed. In particular, one (or more) analog channel (s) and / or digital transmission channel (s) may be used for each signal according to the invention. For example, an audio frequency signal can be transmitted through several channels of analog transmissions each dedicated to the transmission of a sub-band audiofrequency signal. Moreover, each signal can come from one or more sources. For example, excitation detection source signal may be derived from several microphonic circuits sensing the vibration of the vibrating member. Advantageously and according to the invention, the new synthesis frequency is chosen so that it is, among the harmonic frequencies of the new pitch, directly adjacent to said previous synthesis frequency. In particular, advantageously and according to the invention, the new synthesis frequency is chosen so as to be, among the harmonic frequencies of the new pitch, the closest harmonic frequency of the new pitch relative to said previous frequency synthesis, preferably on a logarithmic scale of the frequency. Advantageously and according to the invention, several modulated components of the harmonized tonal signal, called low-rank components, whose synthesis frequencies, called preceding synthesis frequencies, in force for a note pitch preceding the new note pitch in the sequence, are less than the synthesis frequency, in force for said previous note pitch, of another modulated component of the harmonized tonal signal, called the high-rank component, are each modulated towards a harmonic frequency of said new note pitch, selected from so that one of the harmonic frequencies of the new pitch is directly adjacent to the previous synthesis frequency of the low rank component. Advantageously and according to the invention, at least one high-rank component is modulated towards a harmonic frequency of the new note pitch chosen so as to maintain constant the difference between the harmonic ranks of the synthesis frequencies of the high component. rank and an associated low rank component.

Advantageously and according to the invention, a signal, said signal of variation of note, representative of a rate of variation of note pitch is produced from the note signal, and the new synthesis frequency is chosen from the harmonic frequencies of the new note pitch according to the note variation signal. Advantageously and according to the invention, several modulated components of the harmonized tonal signal are modulated each: according to data, called synthetic rank data, representative of a previous harmonic rank, called synthesis rank, to which corresponds a previous frequency method of synthesizing the modulated component, and according to prerecorded data comprising, for a plurality of predefined intervals of note pitch rate, data representative of synthetic rank update values. Advantageously and according to the invention, the sequencing signal is produced from a signal, said excitation detection source signal, representative of the vibration of the vibrating member. Advantageously and according to the invention, the sequencing signal is also representative of an excitation intensity of each excitation contact. More particularly, advantageously and according to the invention, the sequencing signal is produced from a signal, said excitation detection source signal, representative of the vibration of the vibrating member, so that the sequencing signal is representative, for different frequency bands, said excitation bands, of the excitation detection source signal, of at least one excitation intensity value of each excitation contact in each of said excitation bands. More particularly, advantageously and according to the invention, the sequencing signal is furthermore representative of at least one excitation phase value of each excitation contact in each excitation band. More particularly, advantageously and according to the invention, the sequencing signal is produced by excitation detection as the excitation detection source signal is received, in particular by iteratively executing a detection step of the excitation signal. excitation of the excitation detection source signal. In particular, the excitation detection source signal may be formed of the pitch detection source signal. Advantageously and according to the invention, the note signal is produced from a signal representative of the vibration of the vibrating member according to an octave of note pitch detection. Advantageously and according to the invention, the pitch detection source signal is representative of the vibration of a vibrating member devoid of damping member reported. Advantageously and according to the invention, the note height detection source signal is representative of a string mounted under tension. The invention extends to a device for implementing a method according to the invention. The invention therefore relates to a synthesis device comprising at least one processing unit adapted to synthesize an audio signal, called a synthesized audio signal, from a signal, called a note signal, representative of a sequence of pitch levels. wherein the vibrating member is vibrated, and a sequencing signal representative of a sequence of contacts imparted on a vibrating member, said contact sequence comprising excitation contacts adapted to vibrate the vibrating member , Characterized in that the processing unit is adapted to: - produce at least one audio contact signal, for each of said excitation contacts, - modulate several frequency components, called modulated components, of at least one audio signal contact, said tonal signal contact 25, each successively around frequencies, said frequencies of synthesis, each own a harmonic rank of a note's height of the s quence, characterized in that: - for at least one excitation contact, said contact 30 excitation and change of note, resulting in a variation of the note signal to a new pitch of note in the sequence, the unit of processing is adapted to be able to modulate at least one modulated component, called channelized component, of a tonal contact signal, called harmonized tonal signal, resulting from an excitation contact prior to said excitation contact and change of note, around a harmonic frequency, called the new synthesis frequency, of the new pitch of note, corresponding to a harmonic rank different from the harmonic rank of a previous frequency of synthesis of the channelized component, - the processing unit is adapted to produce, after any excitation and note-switching contact, the synthesized audio signal by mixing each corresponding harmonized tonal signal and the audio contact signal of said contact of e xcitation and change of note. Advantageously and according to the invention, the processing unit is able to implement a synthesis method according to the invention. The invention extends to a recording medium - in particular of the removable type (CD-ROM, DVD, USB stick, external electronic hard disk) - adapted to be read in a reader of a computing device, and on which is recorded a computer program adapted to be loaded into the RAM of the computing device when the recording medium is loaded in said reader, characterized in that the computer program comprises portions of program code for execution steps of a method of synthesizing an audio signal according to the invention when the computer program is loaded into the RAM of the computing device. The invention extends to a computer program comprising portions of program code for performing the steps of a method of synthesizing an audio signal according to the invention when said program is executed on a computing device. The invention also relates to a method, a device, a recording medium and a computer program characterized in combination by all or some of the features mentioned above or below. Other characteristic purposes and advantages of the invention appear on reading the following description which refers to the appended figures representing embodiments given by way of non-limiting example, and in which: FIG. 1 is a schematic representation of an exemplary device according to a preferred embodiment of the invention, arranged in operating configuration with an audio amplifier and a guitar having a hexaphonic microphone; FIG. detailed schematic representation of embedded electronic and computer components equipping the device of FIG. 1, - FIG. 3 is a schematic functional diagram of a detection circuit fitted to a synthesizer device of the device of FIG. 1, - FIG. 4 is a diagram. schematic functional diagram of a synthesis circuit equipping said synthesizer device, - Figure 5 is a representation schematic representation in the complex plane of values of frequency sample samples and corresponding predicted frequency sample values, illustrating an attenuation detection principle implemented by the detection circuit of FIG. 3, said values being given in FIG. As an illustrative and nonlimiting example, FIG. 6 is a diagrammatic representation of digital frequency sample values of attenuation and amplitude values of digital excitation frequency samples, produced for different frequency bands. a frequency representation, said values being given by way of illustrative and nonlimiting example; FIG. 7 is a diagram representing the waveform of a digital measurement signal, and the waveform of a signal corresponding prediction produced by the detection circuit of FIG. 3 from the digital measurement signal, said waveforms being As an illustrative and non-limiting example, FIG. 8 shows the waveforms of a first portion and the beginning of a second portion of a filtered digital signal from which recorded tone data is produced. in a memory of the synthesis circuit of FIG. 4, said waveforms being given by way of illustrative and nonlimiting example, FIG. 9 represents an algorithmic diagram according to which generator modules of the synthesis circuit of FIG. 10 represents an algorithmic diagram of a frequency modification step performed iteratively by each generator module; FIG. 11 represents previous values of frequencies of frequency contribution transposition of a modified sound according to FIG. of a current rate of variation of a fundamental value detected, - Figure 12 is a graph representing the trajectories in frequency e of several low-frequency frequency contributions of a modified sound produced by a generator module of the synthesis circuit as successive iterations are made, said trajectories being given for a first illustrative and nonlimiting example, FIG. a graph representing the frequency trajectories of the low frequency frequency contributions of a modified sound, given for a second illustrative and nonlimiting example; FIG. 14 is a diagram showing frequency changes of the filtered digital signal of FIG. 8, generated by a generator module, said modifications being given by way of illustrative and nonlimiting example; FIG. 15 is a diagram representing the waveform of the digital measurement signal of FIG. 7, and the waveform of a resulting tonal synthesis signal produced by the synthesis circuit of FIG. 4, said waveforms and As an illustrative and nonlimiting example, FIG. 16 is a diagram showing the waveform of a disturbance signal produced from the digital measurement signal and the prediction signal of FIG. 7. as well as the waveform of an inharmonic synthesis signal produced from the perturbation signal and the first part of the filtered signal of FIG. 8, said waveforms being given by way of illustrative and nonlimiting example . FIG. 1 represents a device 3 according to a preferred embodiment of the invention connected to a hexaphonic microphone equipping a guitar 1 by means of a suitable transmission cable 4, said micro-hexaphonic being adapted to transmit a microphone signal for each string 2 of the guitar 1 via the transmission cable 4, said microphone signal being representative of the vibration of the string. In particular, the hexaphonic microphone may be of the type comprising a piezoelectric, electromagnetic or optical sensor for each string, for example a hexaphonic microphone marketed by RMC Pickup Co. (USA), Graph Tech Guitar Labs (Canada), Roland corp. (Japan) ... The device 3 of Figure 1 comprises a rigid housing 5 in which is provided a socket 6 equipped with a circuit 200 for receiving the analog measurement signals. The housing 5 encloses a computer central unit 148, equipped with at least one processor 150 running according to an on-board computer program loaded into a live memory 151 associated with the processor 150. In particular, the on-board computer program can be stored in a memory 152 mass and loaded into RAM after powering the device 3.

The housing 5 also contains, for each chord 2 of the guitar, a synthesizer device 136 equipped with a detection circuit 137. Each synthesizer device 136 of the example is furthermore equipped with an analog / digital converter circuit 138 electrically connected to the detection circuit 137 and to an output port of the reception circuit 200 delivering the corresponding microphone signal. Conventionally, the analog / digital converter circuit 138 is adapted to receive the microphone signal of the chord 2, to sample and quantize said microphone signal, and to transmit a resulting digital measurement signal to the detection circuit 137. Note that only three synthesizer devices are shown in Figure 2 for the sake of clarity.

The device 3 of the example further comprises a plug 63 for connecting a support 64 of removable memory, formed in the rigid box 5, and means 65 for gripping-pedal (s), knob (s), button ( s), screen (s) said plug and said input means being connected to the central unit 148 by means of data bus 153 and corresponding input / output interfaces 220. The central unit 148 is adapted to execute in a prerecorded sound loading mode, in which a pre-recorded digital sound file can be selected, via the input means 65, for each chord 2 of the guitar among several digital files previously recorded by any means in the removable memory support 64. The computer unit 148 reads the digital data from the file into the removable memory support 64 and transmits samples, called sound source samples, representative of the waveform of the corresponding sound, said selected sound, to the corresponding synthesizer device 136. via a data bus 149 connecting the computer central unit 148 and said synthesizer device 136. Each synthesizer device 136 of the example further comprises a pretreatment circuit 141 and a synthesis circuit 68, and is adapted to execute, upon receipt of said sound source samples, in a selected sound loading mode. In this mode, said preprocessing circuit 141 produces timbre data from the sound source samples and transmits said timbre data to a synthesis circuit 68 which stores them in a memory 139 on said synthesis circuit 68. In the example, each synthesizer device 136 is further adapted to function, following a corresponding command inputted by the input means 65, and transmitted by the central computer unit 148, in an interactive mode of sound synthesis in wherein the synthesizer device 136 synthesizes in real time a sound sequence signal (not shown) from the digital measurement signal and timbre data loaded into the memory 139 of the synthesis circuit 68. The detection circuit 137 of the example comprises a frequency transforming module 8 implementing a sliding window digital fourier transform method, the module 8 producing, for each of several successive time intervals of the digital measurement signal 7 , digital frequency measurement samples describing a frequency representation of the digital measurement signal 7 in the time interval, said observation window. Preferably, a fast fourier transform is implemented iteratively as and when the measurement digital signal 7 is acquired, according to windows 10, 11, 13, of successive observations overlapping staggered, said no offset, corresponding to a predetermined number of samples of the digital measurement signal 7. Preferably, the fourier transform is implemented according to observation windows 10, 11, 13 of length several times greater than a nominal empty nominal period of the chord 2 corresponding to the synthesizer device 136, so as to allow to use a windowing function improving the efficiency of the separation of the harmonic frequency components of the string obtained in the frequency transform. As an indication, the use of a Hann or Hamming window assumes an observation window length at least twice the fundamental period of the vacuum rope, the use of a Blackman-Harris window. at four terms assumes a length at least four times greater ... Null values can be added following the sequence of samples, according to a traditional technique, called concatenation method with zeros, interpolation of the frequency representation . This method advantageously reduces the frequency sampling step in the Shannon-Nyquist band. In view of the foregoing, each digital frequency measurement sample is representative of a frequency component of the digital measurement signal housed in a frequency band whose width corresponds to the main lobe of the windowing function. The processing circuit 137 further comprises a note pitch detection module 9 adapted to produce a digital signal, called a fundamental signal, from the digital measurement signal, so that said fundamental signal is representative of a value. detected current fundamental for each observation window of the frequency transform.

In the example, the note height detection module 9 comprises a note height detection sub-module (not shown) adapted to provide at each iteration a preliminary fundamental value detected for this iteration. In practice, a traditional note pitch detection technique can be implemented from the samples of the digital measurement signal 7 of the current observation window. In particular, a time-detection technique may be used, for example, a technique for auto-correlation detection of the digital measurement signal (see McLeod and Wyvill, A Smarter Way to Find Pitch, Department of Computer Science, University of Toronto). Otago - New Zealand). Alternatively, a frequency detection technique may be used, for example a detection technique from a frequency representation of the digital logarithmic frequency measurement signal, as taught in Puckette's publications. Brown: An efficient algorithm for the computation of a constant Q transform (Journal of the Acoustical Society of America, Vol 92, No. 5, November 1992) and A high resolution fundamental frequency determination based on phase changes of the Fourier transform (Journal of the Acoustical Society of America, 94, No. 2, point 1, August 1993). Nothing prevents the use of other types of techniques. The note pitch detection module 9 of the example further comprises a sub-module (not shown) for phase vocoder refinement of the detected fundamental preliminary value. In practice, a refined frequency furax value of the frequency can be determined for a digital frequency measurement sample, said sample of maximum, of maximum amplitude value for the current observation window according to the variation of the phase value of said sample of maximum relative to the digital frequency sample corresponding index measurement of the previous iteration. The following formula illustrates this principle: 9 (r, wmax) - 0 (r-1, wmax) + 2 r cst fmax (t) _ fs 2 r 0 In this formula: - A represents the duration of the shift step in number of samples, - r represents the index of the current iteration, - wmax represents the maximum sample index in the frequency representation, - fs represents the sampling frequency in Hertz of the measurement signal, - e represents the phase in the frequency representation, - cste is an integer to be determined so that the refined measurement value f max is closest to a center frequency of the frequency band of the fourier transform to which corresponds the sample index of maximum. The note pitch detection module 9 of the example is adapted to produce at each iteration a digital sample of the fundamental signal according to the ffax value of refined measurement of a value of rank of harmonic 15 to which the sample corresponds. of maximum, considering the preliminary value of fundamental detected. The following formula illustrates this principle: fm (t) f (r) r (2) max max In this formula: - fm (r) represents the current detected fundamental value for the iteration r, - max (t) represents the closest harmonic rank in frequency of the refined measurement value fur considering the preliminary value of fundamental detected. In the example, the note height detection sub module and the phase vocoder refining sub module are adapted to produce a fundamental signal whose values are within an octave detection range extending to treble from a nominal fundamental frequency of the vibration of the rope to empty. The detection circuit 137 of the example further comprises a module, said short-term prediction module 12, for predicting the phase values of the digital frequency measurement samples of the next observation window 13. In practice, the short-term prediction module 12 of the example is adapted to produce at each iteration a digital frequency sample, referred to as the preliminary sample pa (r + 1, w), from each digital frequency measurement sample, said sample m (r, w) current measurement, from the current observation window 10 and the digital frequency measurement sample, said sample m (r -1, w) of previous measurement, corresponding index from the window 11 of previous observation. In practice, the sample pa (r + 1, w) preliminary can be produced according to the principle of the following formula:

in (r, w) 2 I m (r -1, w) I pa (r + 1, w) = m (r -1, w) m (r, w) Given the above, the sample preliminary pa (r + 1, w) is representative of an amplitude value corresponding to that of the current measurement sample m (r, w) and of a phase corresponding to the sum of the phase value of the sample of m (r, w) of current measurement and a phase shift value between the current measurement sample m (r, co) and the previous measurement sample m (r -1, w).

According to a principle of phase coherence, each sample

frequency hosted in the main lobe of the quadrier transform of a sinusoid has a specific phase shift relative to each neighboring frequency sample in the lobe. The following formula defines a theoretical value of the phase shift of a frequency sample of the positive frequency domain of the Shannon-Nyquist band, included in a sinusoid lobe, with respect to a left-handed frequency sample in the lobe: 80 = W-1. NOT

In this formula, W represents the length of the observation windows in number of samples, and N represents the number of frequency samples of the corresponding frequency representations (N> W in case of an implementation of the method of concatenation with zeros). In view of the foregoing, the preliminary values of samples pa (r + 1, w) are out of phase with the values of current measurement samples m (r, w) with different values of a sample pa (r, w). + 1, w) adjoining the other, which is likely to break a phase coherence possibly existing between the current sample values iz (r, w). To overcome this problem, the module 12 short-term prediction of the example treats the preliminary samples p (t + 1, w) according to the principle illustrated in the following formula: e-ibSpa (i + 1, w-1) + Pa (r + 1, w) + eiePa (r + 1, w + 1) wz1 e-ibepa (i + 1, w-1) + pa (r + 1, w) + ei'5BPa (r + 1 , w + 1) In this formula, pa (r + 1, w) designates the value of the predicted sample produced at the output of the short-term prediction module 12. Preferably, the complex factors e 18 and e can be predetermined and pre-recorded in a memory of the short-term prediction module 12 in order to simplify the implementation of the short-term prediction module 12. Furthermore, in the absence of an implementation of the concatenation method with zeros, the imaginary part of the aforementioned complex factors can be neglected. The detection circuit 137 furthermore comprises an attenuation detection module 14 adapted to produce at each iteration a digital frequency sample, referred to as an attenuation sample, representative of an amplitude attenuation factor of the digital measurement signal. from each current measurement sample m (r, w) and from the predicted pa (r, w) sample of corresponding index produced during the previous iteration.

FIG. 5 represents, in the complex plane, three examples of current measurement sample values m (r, w) and three examples of corresponding predicted pa (i, w) sample values. These examples illustrate an attenuation detection principle implemented by the attenuation detection module 14 of the example. According to this principle, the current measurement sample m (t, w) is, if appropriate, representative of the superposition of a component 20, 21, of the digital measurement signal 7, maintained with zero or partial attenuation in the frequency bands of the sample between the previous iteration and the current iteration, with a component 22, 23, of the digital measurement signal 7 supplied between these iterations in the frequency band. In view of the foregoing, a digital frequency sample representative of the attenuation of the held component can be produced by orthogonal projection in the complex plane the value of the current measurement sample m (r, w) on the straight line 25. the corresponding predicted pa (r, w) sample. In the first example of FIG. 5, the corresponding projection component (not shown) is in phase opposition with the predicted pa (r, w) sample. This example therefore corresponds to a total attenuation, between the previous iteration and the current iteration, of the frequency component, called preexisting component, of the sample m (r -1, w) of previous measurement, the sample m (t, w) of current measurement being only representative of a contributed component. In the second example of FIG. 5, the projection component (not shown) is not in phase opposition with the predicted pa (r, w) sample and has a higher amplitude. This example therefore corresponds to a zero detected attenuation of the preexisting component 20 between the previous iteration and the current iteration, the predicted pa (r, w) sample then being representative of the corresponding maintained component. In the third example, the projection component is not in phase opposition with the predicted pa (r, w) sample and has a lower modulus. This example therefore corresponds to a partial attenuation of the pre-existing component between the previous iteration and the current iteration. Therefore, the previously described detection principle makes it possible to detect a partial or total attenuation of the pre-existing component while the intensity is generally amplified in the frequency band between the previous iteration and the current iteration.

The attenuation detection module 14 of the example is adapted to produce the attenuation samples further from data, called damping data, prerecorded in a memory 210 of the detection circuit 137, coupled to the module 14 of FIG. attenuation detection. The damping data of the example are representative of predetermined damping factors of the cord in free vibration from one observation window to the other. Therefore, the attenuation detection module 14 of the example is adapted to produce attenuation samples particularly representative of the attenuation effect resulting from contacts imparted on the string by the instrumentalist, without suffering from the natural damping effect of the chord, due to friction in the air, to the acoustic resistivity of the instrument ... Preferably, the damping data are representative of damping factors specific to different frequency bands for example a damping value / 3 (w) specific to each index w of the frequency representation.

In practice, the attenuation detection module 14 of the example is adapted to produce a digital frequency sample G (r, w), said amplitude gain sample, representative of an amplitude gain value of the preexisting component between the previous iteration and the current iteration according to the principle of the following formula:

(t, w) m (t, w) G (x, w) = a / 3 (w) Pa (t, w) Pa (t,)) In this formula, / 3 (w) denotes the factor of free vibration damping corresponding to the index w of the frequency representation, the symbol • denotes the scalar product operator and the symbol * denotes the conjugate complex. Moreover, the attenuation detection module 14 of the example is adapted to produce said attenuation samples from the amplitude gain samples by thresholding, according to the principle of the following formula:

0 G (r, w) s 0 a (t, w) = 1 G (r, w) z 1 G (r, w) 0 <G (t, w) <1 In this formula, a (r, w) denotes the value of the attenuation sample detected for the frequency band corresponding to the index w of the frequency representation. In view of the foregoing, the predicted pa (r, w) samples and the damping data of the example are representative of a predicted phase value and a predicted amplitude value for each sample m (r , w) of current measurement. Said phase value and said amplitude value defining a predicted waveform of the preexisting component for the current observation window, determined from the waveform of the preexisting component during the previous observation window. The current measurement in (r, w) sample is representative of a real waveform extracted in the corresponding frequency band. In particular, said predicted amplitude value corresponds, in the example, to the amplitude value of the previous measured m (r-1, w) sample attenuated according to the damping factor f3 (w) of the corresponding frequency band. In addition, said predicted phase value is determined in the example, in particular according to a detected value of variation of the phase in the frequency band between a previous iteration, preceding said previous iteration, and the previous iteration, and according to the value of the sample in (r - 1, w) preceding frequency. In the example, the attenuation samples corresponding to a partial attenuation are each representative of the ratio between a maintained amplitude value detected according to said predicted phase value, on said predicted amplitude value. The production of attenuation samples according to the example can be the subject of many alternative embodiments. In particular, an orthogonal projection of the digital frequency sample values 22, 28, 29 of prediction error between the previous iteration and the current iteration can be performed. In addition, the predicted pa (r, w) samples may be generated from the damping data so that each predicted pa (r, w) sample is both representative of said predicted phase value and said d-value. predicted amplitude. Moreover, the principle of projection attenuation detection can be implemented by projection in the time domain, as described below. Moreover, nothing prevents the implementation of the invention according to simplified models of phase prediction and / or intensity. In particular, the natural damping of the chord may be neglected, amplitude gain samples representative of the ratio of the amplitude values of the current measurement samples (t, w) to corresponding predicted amplitude values may be In addition, nothing prevents the implementation of a more sophisticated attenuation detection, for example according to a note height detected in progress and from damping data specific to different guitar scores, representative for each note, depreciation values determined for different ranks of harmonics of the note. In the example, the detection circuit 137 further comprises a calibration module 140, and the synthesizer peripheral 136 is adapted to function, following a corresponding command transmitted by the central computer unit 148, in a calibration mode. guitar. In this calibration mode, the attenuation detection module 14 is executed by means of reset damping data of zero value. Said calibration module 140 produces the damping data from the resulting attenuation samples, so that said damping data thus produced is representative, for each index of the frequency representation, of an average value of attenuation of the rope in free vibration in the corresponding frequency band.

In particular, the calibration module 140 can implement a method for detecting the stationarity of the digital measurement signal between two successive observation windows, in order to take into account only the attenuation samples produced while the 7 digital signal measurement meets minimum stationarity criteria. In addition, the data specific to each index of the frequency representation can be produced according to a minimum intensity threshold of the corresponding frequency component, below which the corresponding attenuation frequency samples are ignored. Preferably, the calibration module 140 may further execute from the fundamental signal so that the damping data produced is representative of harmonic frequency component damping factors of the digital measurement signal. Such a calibration makes it possible to determine precise damping data following the execution of different notes along the neck of the guitar, allowing the cord to resonate freely in prolonged vibration between each note.

In combination or alternatively, all or part of the damping data can be entered via the means 65 for gripping the device. Furthermore, nothing prevents the use of predetermined damping data by any means independently of the instrument actually connected to the device, in particular predetermined data for a predefined instrument model or a predefined range of instrument models whose the acoustic resistivity varies little or not significantly depending on the conditions of use (temperature, humidity ...) and one copy the other. The detection circuit 137 of the example further comprises an excitation detection module 30 adapted to produce, at each iteration for each current measurement sample m (r, w), a frequency sample, called excitation sample, representative of the component 19, 22, 23, brought between the current iteration and the previous iteration. In particular, each excitation sample can be produced from the current measurement sample m (r, w), the corresponding predicted pa (r, w) sample, the damping data, and the sample. corresponding attenuation of the current iteration. In practice, the

excitation detection module 30 may be implemented according to the principle of the following formula: (r, w) = m (r, w) - (')) pa (i, w) w In this formula, (r, w) represents the value of the excitation sample. Preferably, account may be taken of the effect of

damping in free vibration. To do this, the excitation detection module 30 according to the principle of the following formula: e (t, w) = mn (r, w) - a (t, w) Pa (r, w) Nothing prevents producing the excitation samples é (r, w) according to a more approximate method of detecting the contributed component, for example so that each sample is representative of a perturbation component 22, 28, 29 corresponding to the superposition component 19, 22, 23, contributed and a component 26, 27, removed. The detection circuit 137 further comprises a module, called the medium-term prediction module 47, for predicting the digital frequency measurement sample values corresponding to a window 48 for observation subsequent to the current observation window. The mean prediction module 47 can be made identically to the short-term detection module 12 in order to produce a sample b (r + 0, co) predicted for each measurement frequency sample of said subsequent observation window 48. except that the corresponding preliminary h (-t +, co) samples are produced according to the following formula: WW m (t, w) m (r -1, w) IA Pb (r + 0 'w) The detection circuit 137 further comprises a module, referred to as an inverse transform module 263, adapted to produce representative digital samples at each iteration. a predicted waveform 50 for the subsequent observation window 48 from the corresponding pb (z + 0 'co) predicted samples. In practice, a fast method of inverse fourier transform can be implemented, with truncation of the last samples -in the case of an implementation of the traditional method of concatenation with zeros-, then weighting of the values of the resulting samples according to the windowing function and according to an amplitude normalization factor determined according to the following formula: d = W w (n) 2 n

In this formula, w represents represents the windowing function and d represents said normalization factor. The module detection circuit 137 further comprises a module 264 for constructing a digital signal, said prediction signal 51, adapted to produce, at each iteration, a sequence of digital samples, called the predicted sequence 52, of the signal 51 of prediction, corresponding to the interval of the shift step to the next observation window 13, said interval 49 following shift. Sample values of sample sequences 53, 54, 55

Numerical waveforms 50, 56, 57, predictors of the current iteration and previous iterations, corresponding to said interval 49 following shift are summed two by two for this purpose. The digital samples of the waveform 50 predicted from the current iteration are stored in a memory (not shown) of the building module 264 for subsequent iterations.

The detection circuit 137 of the example also comprises a short-time disturbance detection module 58 adapted to produce, at each iteration, a sequence of digital samples, called a disturbance sequence 59, of a signal S n) disturbance detected. The detection module of the example is adapted to produce said perturbation sequence 59 from the

a digital sample sequence 60 of the digital measurement signal 7 extending in the interval, called the offset current interval 61, of the shift step between the previous observation window 11 and the current observation window 10, and from the predicted sequence 62 determined during the previous iteration, so that said perturbation sequence 59 is representative of the difference between the waveform of the digital measurement signal 7 and the waveform of the prediction signal 51 during the interval 61 offset current. The preprocessing circuit 141 of the example is adapted to implement a digital filter so as to produce a digital signal filtered from the sound source samples, the filtered digital signal being representative of the sound waveform after filtering the sound source. its pre-recorded in the A digital file selected. The filter of the circuit 141 of the example is designed to

at least partially compensate, during the operation of the synthesizer device 136 in interactive mode of sound synthesis, a natural imbalance of the relative intensities of the frequency components of the string in favor of the low frequencies compared to the acute frequencies. As an indication, the following formula defines the relative theoretical intensity in decibels for different harmonics of a free-resonance chord, following an ideal point pinch conferring a triangular profile to the chord before the vibration is triggered: sin (ro ' 7r) l PdB (r, G) = 201og10 2 r In this formula, a represents the ratio of the distance between the vertex of the triangular profile of the string and the bridge, along the length of the string.

In view of the above, the filter of the pretreatment circuit 141 may be designed to provide an equalization gain of the harmonic frequency components, called prerecorded harmonics, of the prerecorded sound corresponding to the square of the rank of each prerecorded harmonic. Preferably, the equalization gain values can be capped to a predefined maximum gain threshold value to prevent the emergence beyond a tolerable auditory threshold of harmful sound elements caused by a microphonic background noise. high frequencies and / or high frequency background noise from the selected sound recording. In practice, this threshold can be adjusted via means 65 for inputting the device 3. Alternatively, this threshold can be determined according to measurement values of the signal-to-noise ratio of the micro-hexaphonic and / or measurement values. the signal-to-noise ratio of the selected sound.

In practice, the filter of the pretreatment circuit 141 can be adapted to determine the coefficients of a finite impulse response digital filter, called FIR filter, and of linear phase, according to a traditional iterative method of digital filter design starting from a filter mask defined according to the values of equalization gains and data, called original note height data, representative of the fundamental fv of the selected sound. In practice, the original note pitch data may be produced and prerecorded by any means in the form of meta data in the selected pre-recorded digital sound file, and transmitted to the synthesizer device 136 along with the sound source samples. Alternatively, the original note pitch data may be inputted via the input means 65. Alternatively or in combination, the pretreatment circuit 141 may be adapted to implement a pitch height detection method to produce said original note pitch data from the sound source samples. In addition, the preprocessing circuit 141 of the example is adapted to implement a sliding window Fourier Fourier transform method according to the frequency transforming module 8 of the detection circuit 137 so as to produce digital frequency samples. said pre-recorded samples (represented by the symbol V in FIG. 4), according to the same length of observation window, the same windowing function, and the same shift step, from the filtered digital signal whose values of digital samples corresponding to a first part 66 are reduced beforehand to a zero value. Therefore, the prerecorded samples are descriptive of frequency representations of successive segments, said prerecorded segments 69, 70, of a second portion 67 of the filtered digital signal, overlapping each offset by the shift. The original note pitch data, the pre-recorded samples and a so-called initial sequence of samples of the filtered digital signal corresponding to said first part 66 form the timbre data transmitted to the synthesis circuit 68 and recorded, such as previously described in the memory 139 of the synthesis circuit 68. The detection circuit 137 of the example is electrically connected to the synthesis circuit 68 so as to transmit on the fly, the fundamental signal, the disturbance signal samples, the attenuation samples and the excitation samples produced at each iteration.

The synthesis circuit 68 of the example comprises several generator modules 142, each adapted to produce, at each iteration, modified digital frequency samples, called modified samples, from the pre-recorded samples corresponding to a prerecorded segment 69, 70. In particular, each module 142 generator is adapted to be able to produce, during successive iterations, modified samples from segments 69, 70, prerecorded successive. To do this, the generator sub-module 142 may be adapted to increment at each iteration, an index stored in a memory space of the memory 139, specific to said generator module 142, said index being representative of the current prerecorded segment for the generator module 142 . Each generator module 142 of the example implements, at each iteration, a step 158 of frequency changes, in particular with a view to frequency transposing said prerecorded harmonics according to the fundamental signal, and in particular to weight the amplitude of said prerecorded harmonics according to attenuation samples produced for the current iteration. In view of the foregoing, each generator module 142 produces for several successive iterations, modified samples corresponding to successive prerecorded segments 69, 70, said modified samples being descriptive of frequency representations of segments, so-called modified, successive segments of a modified sound from the selected sound. Said step 158 of frequency changes comprises in particular an interpolation sub-step 143 and a translation sub-step 146 executed for each prerecorded harmonic of the selected sound. These sub-steps implement a frequency transposition of the prerecorded harmonic to a harmonic frequency, called transposition frequency 39, 40, 41, in accordance with the current value of the fundamental signal. Therefore, the modified sound consists of several components, called frequency contributions, each resulting from a prerecorded harmonic, and each modulated in frequency as successive iterations, according to corresponding transposition frequency values successively in force. during said successive iterations. The following formula expresses the shift 71, 72, 73, 32, 74 in frequency of the transposition to be implemented at each iteration and for each prerecorded harmonic according to a transposition rank in force for this prerecorded harmonic: bco (r , r) = (p ~ (c, r) • fm (r) -1 - fv) N 1 fs In this formula: - prl (r, r) denotes the rank of transposition of the prerecorded harmonic of rank r , in force during the iteration i and for the generator module - bco77 (x, r) designates said frequency shift of the transposition to be implemented, in terms of frequency samples. Each generator module 142 is adapted to implement a step 154 of initialization in the memory 139 of transposition rank data specific to the generator module 142, and representative of a value of rank of transposition in force for each prerecorded harmonic. This step is executed during each iteration corresponding, for this module, to a modification of a first prerecorded segment 69. Preferably, the transposition rank data can be initialized so as to transpose, in the step 158 of frequency changes executed for the first prerecorded segment 69, each prerecorded harmonic to a harmonic frequency of corresponding rank. The synthesis circuit 68 of the example further comprises a selection module 75 adapted to extract from the memory 139, during each iteration, a set of update data representative of a new value of rank of transposition for each value previously in effect during the previous iteration. The selection module 75 is adapted to select said update data set from among several prerecorded update data sets in the memory 139 of the synthesis circuit 68. The graph in Figure 11 illustrates a principle that produces each update data set. This graph presents oblique lines corresponding each to a value of rank of transposition in force during the previous iteration, said line representing the corresponding frequency of transposition, said preceding frequency, as a function of the rate (in abscissa) of variation of the fundamental detected current relative to the previous iteration.

Said preceding frequencies 76, 77, 78, 79 are distributed in frequency proximity intervals 80, 81, 82, each extending around a frequency 39, 40, 84, which is harmonic with respect to the current value of the signal. fundamental. This distribution is constant within intervals, referred to as subdomains 87, 88, 89, of a domain 90 for detecting the rate of change.

In the example, an update data set is predetermined for each of said subdomains 87, 88, 89, so as to define, for several rank values of low frequency transpositions (four in the example of FIG. 11), a new value of effective translation rank which corresponds to the frequency 39, 40, 84, harmonic of the proximity interval accommodating the corresponding frequency 76, 77, 78, 79, preceding. Therefore, each corresponding low frequency frequency contribution is modulated, during the current iteration, to the nearest harmonic frequency in force according to the logarithmic scale. In addition, each set of update data is predetermined in the example so as to define for each value of higher transposition rank at low frequency transposition ranks, a new value of transposition rank in force according to the principle of the formula. Next: Pn (r, r) = PTI (r, Rp) + r - Rp, r> RP In this formula Rp denotes a predefined number of low frequency transposition ranks according to which to determine the update sets.

In the example, the selection module 75 is adapted to produce at each iteration from a current sample and from a preceding sample of the fundamental signal, a current sample of a signal representative of the variation rate of the signal of the signal. fundamental. In addition, the selection module 75 is adapted to execute from predetermined threshold data by any means and prerecorded in the memory 139 of the synthesis circuit 68 with the update data sets. The threshold data of the example are representative of values 91, 92, 93, thresholds of the rate of variation beyond which the frequency 76, 77, 78, 79, preceding of at least one frequency contribution low frequency flips one subdomain 87, 88, 89, to another. In particular, the threshold data and the update data sets can be predetermined by any means as a function of the note height detection range of the note height detection module 9 and the number of low frequency transposition ranks. desired. The update data sets extracted at each iteration are transmitted to each generator module 142 of the synthesis circuit 68.

Each generator module 142 of the example is further adapted to implement a step 155 of updating the data of transposition ranks. This step is executed from the update data set received during each iteration, called iteration of subsequent modifications, corresponding for this module, to a modification of a prerecorded segment 70 subsequent to the first prerecorded segment 69. In this step, the data of transposition ranks are recorded, once updated, in the memory 139 for the next iteration. The following formula illustrates the execution of the steps of initialization and update of the data of ranks of implementation as implemented in the example:

In this formula: r, p (r -1, r) denotes a set value of transposition rank day defined by the update data set in effect for the current iteration, 25 - p (c -1, r) denotes the rank of transposition of the prerecorded harmonic of rank r, in effect when of a previous iteration r-1 for the generator module 'q, 20' r0 (n) denotes the most recent iteration, called triggering iteration, of changes in frequency of the first prerecorded segment 69 for the generator module T I. FIGS. 12 and 13 show the frequency paths of low frequency frequency contributions of an example of the modified sound produced from an iteration 94 for triggering the modified sound and as the following iterations are measured. In these figures, the harmonic frequencies in effect during each iteration are represented by circles spaced along the axis 95 of the frequency (in Hertz) as a function of the value 134, 135, current of the fundamental signal. The frequency trajectory of each frequency contribution is represented by a sequence of arrows, each extending from a harmonic frequency value corresponding to the transposition frequency in effect during the previous iteration, to a harmonic frequency value corresponding to the transposition frequency in effect during a current iteration. FIG. 12 represents an example in which the detected fundamental progressively varies so that the rate of variation of the detected fundamental is accommodated at each iteration in the subdomain, said subdomain 88 of progressive variation, of variation corresponding to the lowest values of the domain of variation in absolute value. Therefore, each frequency contribution is modulated according to the frequency of the same rank of harmonic as and iteration. The device of the example is therefore adapted to be able to produce a sound sequence signal reproducing the tone of the selected sound in a particularly faithful manner and quite representative of modulation of the pitch of the string by gradually changing its voltage. It should be noted that the graph of Figure 11 and Figures 12 and 13 relate to different examples. Indeed, Figure 11 is an example given for four rows of low frequency harmonics, while Figures 12 and 13 relate to examples given for nine rows of low frequency harmonics.

The subdomain 88 of progressive variation which applies for the examples of FIGS. 12 and 13 is therefore in fact narrower than that illustrated in FIG. 11, all things being equal. FIG. 13 represents an example in which the detected fundamental varies abruptly so that the rate of variation of the fundamental is housed outside said subdomain 88 of progressive variation during an iteration 131. Such a variation occurs in particular when a note pitch change performed by the instrumentalist. In this case, the values of the transposition ranks of the frequency contributions are adjusted during this iteration 131 with respect to the previous iteration 130. As illustrated in this figure the frequency modulation of the frequency contributions which results is relatively progressive, despite the abrupt variation of the detected fundamental. The principle illustrated in FIG. 11 can be the subject of numerous variants of implementation. In particular, a minimization of the frequency modulation can be carried out according to other frequency scales, in particular a scale more accurate than the logarithmic scale, of the sensitivity of the human ear to the frequency differences. In the interpolation sub-step 143, a beam of digital frequency samples, called the harmonic beam, is produced from pre-recorded samples, called original samples, located in and around the band 96, 97, 98 , 99, of the main lobe frequencies of the windowing function centered around the frequency 100, 101, 102, 103, of the pre-recorded harmonic, so that the samples of said harmonic beam are representative of interpolated values of samples frequency according to indices of the frequency representation shifted according to a rounding remainder value of the offset value 71, 72, 73, 74, of frequency transposition. Examples of values (in solid lines) of amplitudes of original samples and values (in dashed lines) of corresponding interpolated amplitudes are shown in FIG. 14. In particular, the interpolation sub-step can be implemented in FIG. works according to a linear interpolation method. This interpolation step makes it possible to compensate in part for the approximation error occurring due to an integer shift shift of frequency samples, as implemented 37 during the translation sub-step 146 described herein. -after. The following formula illustrates the principle of interpolation sub-step 143: ## EQU1 ## In this formula : - wl (r) denotes the index in the frequency representation corresponding to the sample of index 1 of the harmonic beam, - L denotes a number of digital samples of which each bundle is constituted, this value depending on the width of the main lobe of the windowing function, - & o (r, r) denotes the rounded value of the frequency shift shift in number of frequency samples. The frequency modification step 158 further comprises a sub-step 144 of weighting performed for each harmonic beam produced following the interpolation sub-step 143, in which a beam of digital frequency samples, called the contribution beam, is produced from the harmonic beam, phase adjustment data and a beam, called modulation beam, of complex value digital samples. During each iteration of modifications of the first prerecorded segment 69, the generator module 142 is adapted to implement, following step 154 of initialization of the data of transposition ranks and prior to the step 158 of changes in frequency, a step 156 of initialization of the modulation beams. In this step, the samples of the modulation beams corresponding to the various prerecorded harmonics are initialized and recorded in the memory 139. In the example, each modulation beam is initialized from a beam, called the excitation beam. current excitation samples housed in the band 35, 36, 37 of frequencies of the main lobe of the windowing function centered around the frequency 39, 40, 41, of transposition in force for the prerecorded harmonic taking into account the data of transposition ranks and the current value of the detected fundamental signal. Preferably, the amplitude values 104, 105, 106 of the samples of each excitation beam can be equalized, during the step 156 of initialization of the modulation beams, so that amplitude values of the beam of resulting modulation has a flat frequency profile. To do this, each sample value of the excitation sample beam can be weighted according to the value of the ratio of the value 104, 106, of the maximum amplitude of the excitation sample beam to the amplitude value. of the sample. During each iteration of modifications of a prerecorded segment 70, the generator module 142 is adapted to execute, following step 155 of updating the data of transposition ranks and prior to step 158 of frequency changes. , a step 157 of attenuation. In this step, the modulation beams stored in memory during the previous iteration are updated according to attenuation samples produced during the current iteration. In this regard, the synthesizer circuit 68 further comprises an attenuation control module 159 adapted to produce, at each iteration, data, called harmonic attenuation data, from the fundamental signal and the attenuation samples. current and provide said harmonic attenuation data to generator modules 142. In particular, the harmonic attenuation data of the example comprise representative low-frequency attenuation data, for each rank of a predefined number of low frequency transposition ranks (this number possibly being different from that predefined for the first time). generation of update datasets), the attenuation sample value 32, 33, 34 corresponding to the lowest attenuation in the main lobe frequency band 35, 36, 37 of the windowing function centered around the corresponding transposition frequency 39, 40, 41 (taking into account the current detected fundamental value).

In addition, the harmonic attenuation data of the example includes high frequency attenuation data, produced from the low frequency attenuation data. In particular, the low frequency attenuation data of the example are representative of attenuation factors of the high frequency ranks advantageously correspond to the average of the attenuation factors of the low frequency ranks: Ry Y (r, P) = Ry Y (r, X) (p> Ry) x = 1 In this formula y (r, p) represents the attenuation factor applying for prerecorded harmonics corresponding to the transposition rank p, and Ry designates a predefined number of ranks low frequency transposition (Ry can be different from Rp). The samples of the modulation beams updated during the attenuation step 157 are stored in the memory 139 for the next iteration. The following formula illustrates the principle of attenuation of the amplitude values of the modulation beams as implemented in the example: T1 (r, r, l) = T1 (r -1, r, l). y (r, p (r, r)), (z> r0 (n)) In this formula, zTl (r, r, l) denotes the value of the numerical sample of index 1 of the modulation beam corresponding to the prerecorded harmonic of rank r for the generator module Ti. Furthermore, during each iteration of modifications of the first prerecorded segment 69, the generator module 142 is adapted to execute, following the step 154 of initialization of the transposition rank data and before step 158 of modifications to frequency, a step 160 of initializing the phase adjustment data. Furthermore, during each iteration of modifications of a pre-recorded segment 70, the module 142 generator 25 is adapted to perform, following step 155 of updating the transposition rank data and prior to step 158 frequency changes, a step 161 of updating the phase adjustment data. The adjustment data of the phase of the example is representative of a complex coefficient for ensuring phase continuity between the modified samples produced during the current iteration and those produced during the previous iteration. This data is recorded in the memory 139 during each execution of the initialization step 160 or the updating step 161, for the next iteration. The following formula illustrates the principles of initialization and updating of the complex phase adjustment coefficients implemented in the corresponding steps: i2ar Scot (r, r) • AN r = r0 (T1) i2ar • Scon ( r, r) • A 71 (r -1, r) e n r> r0 (ri) In this formula l? (r, r) denotes the coefficient 77

adjustment of the phase. In the sub-step 144 of weighting, the digital sample values of the harmonic beam are multiplied two by two with the sample values of the modulation beam in force for this iteration so as to modulate the values of amplitude and of phase of the digital samples of said harmonic beam. In addition, the digital sample values of the harmonic beam are multiplied according to the complex adjustment coefficient of the corresponding phase. The following formula illustrates the principle of weighting sub-step 144: ## EQU1 ## In this formula , éTl (r, r, l) denotes the sample of index 1 of the contribution beam and i (r, r, l) denotes the harmonic beam produced following the interpolation sub-step 143. Figure 14 shows the amplitudes 107 of the sample samples of contribution beams corresponding to the first ranks of prerecorded harmonics. ~ Tl (r, r) = e In the translation sub-step 146, the modified samples, corresponding, in the frequency representation, to the frequency indices shifted by a value 71, 72, 73, 74, rounded by the transposition shift by the indices of the original samples corresponding to the prerecorded harmonic are produced from the samples of the corresponding contribution beam. In particular, from one execution of the translation sub-step to the other, the sample values of two contribution beams corresponding to the same indices of the frequency transform are added in the example. On the other hand, sample values of contribution beams exceeding the Schannon-Nyquist frequency are ignored in the example. In addition, the contribution beam sample values overflowing to the negative frequencies of the frequency representation are folded back to the positive frequencies in the example. Preferably, the modified samples corresponding to the negative frequencies of the frequency representation are produced from the modified samples of the positive frequencies, by symmetry with respect to the origin. Fig. 14 shows examples of modified sample amplitude values 108 corresponding to a low frequency portion of the frequency representation. In addition, during the attenuation step 157 and during the step 156 of initialization of the modulation beams, data, said data of remaining energy, representative of a value of global energy of remaining amplitude for the samples of the modulation beams following their initialization or attenuation, are produced from the corresponding samples, and transmitted to a module, said trigger control module 147, of the synthesis circuit 68. Said trigger control module 147 is adapted to transmit at each iteration from the corresponding remaining energy data transmitted by the generator modules 142, a trigger control signal to the generator module 142 corresponding to the lowest overall energy value. of remaining amplitude. Each generator module 142 of the example also executes, during each iteration before any other step, a step 145 of updating the index of the current pre-recorded segment. In this step, the index is reset so that it designates the first prerecorded segment 69 in the case where the generator module 142 has received the trigger control signal during the previous iteration, the index being incremented in the opposite case. Therefore, the modified sounds likely to have the highest loudness are maintained from one iteration to the other while a modified sound likely to have the lowest loudness is interrupted at each iteration to release a generator module for the production of a new modified sound according to the excitation samples of the iteration.

The synthesizer module 68 further comprises a superposition module 162 adapted to mix the modified frequency samples produced by the frequency modification modules 142 during each iteration, so as to produce descriptive samples of a current frequency representation of a digital signal, said tonal synthesis signal 110, in which the frequency contributions of the current modified sounds are superimposed. The synthesis circuit 68 further comprises an inverse transform module 163 adapted to produce, at each iteration, digital samples representative of a waveform, called the tonal waveform 109, from said mixed samples. In practice, a fast method of inverse digital quadrant transformation can be used, with truncation of the last samples -in the case of an implementation of the concatenation method with zeros-, then weighting with the windowing function and according to said amplitude normalization factor. The synthesis circuit 68 further comprises a module 164 for constructing the tonal synthesis signal 110 adapted to produce at each iteration a sequence of digital samples, called the tonal sequence 116, of the tonal synthesis signal, corresponding to the current interval 61. offset. The sample values of the digital sample sequences 111, 112, 113, waveforms 109, 114, 115, current and past tonal corresponding to this interval 61 are superimposed for this purpose. The building module 164 is further adapted to record the tonal waveform 109 of the current iteration in a memory (not shown) of the building module 164 for use in subsequent iterations. The synthesis circuit 68 of the example further comprises a convolution module 117 adapted to produce a sequence of digital samples, called convolution sequence 118, of samples by convolution from the iteration disruption sequence 59 current and the initial sequence of the stamp data. In particular, the convolution can be implemented according to a traditional technique of rapid convolution (see The digital signal processing handbook, Vijay Madisetti, Douglas Bennett Williams, 1998, CRC Press LLC, pp. 8-1 to 8-4). The selection and tripping module 117 is furthermore adapted to produce, at each iteration, a sequence of digital samples, called the driving sequence 122 corresponding to the offset interval 61, by superposition of the forms of wave convolution sequences 118, 119, 120, 121 from the most recent iterations, shifted with respect to each other according to the shift step according to the principle of linear combination of the convolution product. Said driving sequence 122 is representative of a current portion of a digital signal, called an inharmonic synthesis signal 123. The detection circuit further comprises an output module 124 adapted to produce the sound sequence signal as iterations proceed. In particular, the detection circuit of the example is adapted to mix the sequence 122 of current attack samples with the current tone sequence 116 according to respective factors of superposition gains, so as to produce a corresponding sequence of samples , said final synthesis sequence, of the sound sequence signal. In particular, said superposition gains can be determined by the user from the input means 65. The peripheral synthesis circuits 68 are electrically each connected to a port of an output circuit 170 of the device 3, said output circuit 170 being adapted to produce a digital output signal by mixing the received sound sequence signals, transmitted by said circuits 68 of synthesis. Preferably, said output circuit 170 comprises a module (not shown for digital / analog conversion connected to an analog socket 125 formed on the rigid housing 5 so as to be able to connect the device to an amplifier 126 equipped with an analog plug corresponding input for audible rendering of the digital output signal in real time The output circuit 170 may further comprise an encoder module adapted to translate the digital output signal into any other suitable digital or analog transmission format For practical purposes, the detection circuit 137, the pretreatment circuit 141, and the synthesis circuit 68 of the example may be realized by means of components of FIG. traditional digital electronic devices, such as programmable logic circuits-in particular FPGA-type, dedicated digital components -logical ports, flip-flops, specialized integrated circuits known as ASIC of the memories, flash memory, microcontrollers .... In addition, each circuit can be realized by means of one or more microprocessors, executing according to a loaded program in an associated random access memory, in particular one or more processors specialized in digital signal processing. The architecture of the synthesis circuit 68 may in particular be adapted according to the number of modules 142 generators to implement. In this regard, the applicant has found by experience that the use of a small number of generator module makes it possible to obtain satisfactory sound results in the free vibration regime. In addition, the applicant has been able to ascertain by experience the use of a larger number, but modest enough for real-time implementation, to obtain at any time sound results representative of the game of the game. instrumentalist, including in the course of prolonged forced roping of the rope (continuous friction, introduction of inharmonic forced components, noises, etc.). In view of the above, the excitation samples produced by the detection circuit 137 for one or more observation windows, called excitation contact windows, extending in part at least over a time interval of the signal 7 digital measurement during which an excitation contact is imparted on the string, are representative of values of excitation amplitudes and excitation phases of said excitation contact. In particular, each excitation amplitude value and each excitation phase value is detected in the frequency band of the corresponding digital measurement frequency sample. Where appropriate, the excitation samples produced by the detection circuit 137 are representative of a detected sequence of excitation contacts imparted on the vibrating member, and form as such a sequencing signal according to the invention. In the example, each detected excitation contact causes the production of modified samples representative of one or more modified sounds, called modified contact sounds, each triggered following a corresponding excitation contact window, called the trigger window. . The modified samples of each modified contact sound form an audio contact signal according to the invention, produced in response to the corresponding excitation contact.

In the example of the preferred embodiment of the invention, the modified samples of each modified contact sound are produced from the excitation samples of the corresponding trigger window, and sound source samples from the selected sound. . In particular, each frequency contribution of the modified contact sound is initially produced by transposing the corresponding prerecorded harmonic around an initial transposition frequency and by weighting the amplitude of the corresponding prerecorded harmonic according to an amplitude value of d. excitation, called initial gain value, detected in a frequency band housing the initial transposition frequency. Moreover, each frequency contribution is then produced by transposition of the corresponding prerecorded harmonic as iterations follow, around the transposition frequencies successively in effect for the frequency contribution during these iterations. During said subsequent iterations, the frequency contribution is furthermore produced by weighting the prerecorded harmonic according to said initial gain value attenuated cumulatively according to successive attenuation values, each detected in a frequency band of the digital measurement signal 7 accommodating the transposition frequency in effect at the corresponding iteration. In particular, each modified contact sound is attenuated as appropriate according to harmonic attenuation data representative of a partial attenuation of the modified sound produced for one or more observation windows subsequent to the modified sound trigger window. , each extending at least in part over a time interval of the digital measurement signal during which a detected excitation contact, subsequent to contact with the modified sound, is imparted on the string.

Moreover, in view of the foregoing, the modified samples of the modified contact sounds corresponding to an excitation contact extending over several observation windows -froting the chord for example-combine so that the sequence signal sound is representative of a resulting composite sound representing in a particularly faithful and realistic way the prolonged excitation contact. The modulation of the prerecorded harmonic phases according to excitation phase values detected around the corresponding initial transposition frequencies makes it possible to ensure a phase continuity between each modified sound corresponding to the same excitation contact. In addition, the modified sounds triggered out of contact with the string have an intensity proportional to the non-stationary components which are at the origin, so that their sound impact is negligible for the instrumentalist. Moreover, any partial attenuation effect produced by a partial contact during several observation windows is translated shortly after the iteration corresponding to the first observation window of said contact. Furthermore, in view of the foregoing, the sound sequence signal of the preferred embodiment of the invention exhibits at all times a harmonic content which varies in accordance with the instrumentalist's play without the effects of untimely changes in note pitch or inadvertent resonance failure despite any possible variation of the fundamental signal corresponding to an inadvertent change in note pitch detected.

In the case of an impact of the finger, said harmonic contact, at a vibration node of a vibratory low frequency mode of the string, a variation of the fundamental signal corresponding to an inadvertent change of note pitch risk of occurring. For example, the string of E bass, called El, guitar played empty, after impact of the finger above the seventh fret from the nut of the neck of the guitar, a frequency signature quite similar at so, says B2, of the upper octave. A fundamental value corresponding to the note of si, said B1, of the lower octave is thus possibly detected by the note height detection module 9 of the example. In this case, taking into account the adjustment of the resulting ranks of transposition (see FIG. 13), each low frequency frequency contribution whose transposition rank is three before the harmonic contact is at least attenuated and modulated by negligibly in spite of such an inadvertent change of pitch. Consequently, said low-frequency frequency contributions are attenuated in a similar way in the presence or absence of an inadvertent change in the note pitch detected following the making of a harmonic contact in the preferred mode of the invention. The applicant has been able to ascertain by experience that the sound sequence signal of the preferred embodiment of the invention represents, in a particularly faithful manner, the melodic phrasing imparted on the string, including in the presence of prolonged contacts causing a strongly noisy vibration. In addition, the sound sequence signal of the device 3 of the example does not present any sound dissonances, other than those specific to the prerecorded sound if appropriate, during the free vibration of the string. Furthermore, the variation of the amplitude envelope of each prerecorded harmonic is neither accelerated nor slowed by the effect of the transposition as implemented in the example. Therefore, the device of the example makes it possible to restore the tone of the selected sound in a particularly faithful manner. The inharmonic synthesis signal 123 of the example is produced by convolving the disturbance signal with the first part of the filtered digital signal. Said first part comprising the transient attack components of the selected sound, the inharmonic synthesis signal 123 makes it possible to translate with sharpness the percussive effect of each contact produced on the string. In the example, this inharmonic synthesis signal 123 compensates for a phenomenon of blunting of the attacks of the modified sounds, resulting from the modifications in frequency carried out by the generators modules 142, the blunting being all the more marked as the observation window is long. In the example of the preferred embodiment of the invention, the disturbance signal sE (n) of the example can be produced at short notice from the digital measurement signal, so that each contact is followed by 'an audible consequence without perceptible delay for the instrumentalist. Indeed, the pitch of shift can be defined for any string of the guitar so as to correspond to a negligible delay, including for the bass chords requiring a long observation window to provide a desired attenuation detection accuracy. The pretreatment circuit 141 could be improved to extract more precisely the transient components of the selected sound. Any suitable traditional method could be implemented in this respect. The applicant has been able to ascertain from experience that good results can be obtained for a modest number of low frequency transposition ranks and a modest number of generator modules 142, a real-time implementation being therefore possible without delay. crippling. The device of the preferred embodiment of the invention may be subject to various variants. In particular, each generator module can be adapted to perform a selection step, among several sets of stamp data from different digital prerecorded sound files, a set of stamp data from which to produce the modified sound. The selection can be performed according to predefined criteria, for example according to a scale of excitation intensity so that a more faithful reproduction of the dynamic variation of the timbre of a musical instrument whose sound is to be reproduced can be obtained , for example an acoustic piano.

According to another variant of the preferred embodiment, the pretreatment circuit may be adapted to produce patch data further comprising original frequency data for each prerecorded harmonic representative of a refined central frequency value of the harmonic prerecorded in frequency. Therefore, the original samples from which to produce each frequency contribution can be selected precisely, even for selected sounds with marked inharmonicity. In addition, the device of the example may be adapted to take into account phenomena of vibration inharmonicity of the rope, such as for example when the rope is very strongly pinched, by the effect of the use of ropes diameters In practice, the phase vocoder refining sub module of the note pitch detection module 9 can be adapted to determine, for each harmonic rank, a refined measurement value of the frequency from a digital frequency measurement sample corresponding to a local maximum amplitude located near a preliminary frequency value of the harmonic rank defined by the detected fundamental preliminary value. The following formula represents the frequency shift of the transposition to be implemented in this variant of the preferred embodiment of the invention: i fm (i, p ~ (i, r)) -1 r 'fv (1) In this formula: fm (c, p (-r, r)) denotes the refined measurement value of the transposition frequency of the prerecorded harmonic of rank r, fv (r) denotes the refined central frequency value of the prerecorded harmonic of rank r. In addition, other synthesis techniques can be implemented alternatively or in combination with the generator modules described for the example of the preferred embodiment of the invention, such as, for example, additive synthesis techniques, subtractive synthesis, synthesis techniques by physical modeling ... & off (r, r) = N fv (r) fs In a variant, the detection circuit 137, the preprocessing circuit 141 and the synthesis circuit 68 can be realized at the same time. means of a single circuit equipped with a processor running according to a program adapted to implement features of the modules of said circuits, loaded into a random access memory associated with the processor. In particular, the program may be prerecorded in the mass memory 152 and transmitted by the central computer unit to the synthesizer device 136 via the corresponding data bus 149, following the powering up of the device 3. In practice, said program may be prerecorded in the mass memory 152 at the factory, or by the user by means of a removable memory medium, by downloading by means of a communication device of the device 3 (not shown) ... Nothing does not prevent to implement a synthesis device according to the invention comprising a single processing unit adapted to treat, according to the method of the invention, one or more vibration signals specific to different vibrating organs of a musical instrument . In particular, the invention can be implemented by means of a generic microcomputer equipped with a microprocessor and an associated random access memory, running according to an operating software loaded in RAM after powering up. of the microcomputer, and according to a program loaded in RAM from a mass memory of the microcomputer, said program being adapted so that the microcomputer implements a method of synthesizing an audio signal according to the invention following the loading of said program, so that at least one vibration signal according to the invention, the acquisition of which is performed by a sound card equipping the microcomputer, is processed by the central unit to produce a synthesized audio signal according to the invention as output of said sound card. Nothing prevents to implement all or part of the method of the invention by analog processing. In the example, the microphone signal is transmitted to the input of the device 3 by analog means. Nothing prevents, to implement a device according to the invention treating one or more signals representative of the vibration of a vibrating member received digitally. In particular, a digitization of each signal can be implemented by means of a digitization circuit embedded on the musical instrument. In addition, nothing prevents coupling the digitizing circuit of the musical instrument with an on-board frequency conversion circuit so that a signal representative of the vibration of the vibrating member is provided at the input of a device. according to the invention, according to a frequency representation of said signal. In addition, nothing prevents the invention from being implemented in a simplified embodiment from a sequencing signal conforming to the MIDI standard. Moreover, nothing prevents the implementation of a device according to the invention adapted to produce a synthesized audio signal according to the invention from a sequencing signal according to the invention and a note signal according to the invention. the invention produces from a detection circuit interposed between the instrument and the device according to the invention. Nothing prevents the implementation of a simplified design device without short-time disturbance detection module 58 and convolution module 117.

Claims (1)

CLAIMS1 / - A method of synthesizing an audio signal, said synthesized audio signal, from a signal, said note signal, representative of a sequence of notes heights according to which the vibrating member is set in vibration, and a sequencing signal representative of a sequence of contacts imparted on a vibrating member (2), said sequence of contacts comprising excitation contacts able to vibrate the member (2), in which process: at least one audio contact signal is produced for each of said excitation contacts, - several frequency components, called modulated components, of at least one contact audio signal, called tonal contact signal, are each modulated successively around frequencies, said frequencies (132, 133) of synthesis, each of which is at a harmonic rank of a pitch (134; 135) of the sequence, characterized in that: - for at least one excitation contact, said contact ex citation and change of note, resulting in a variation of the note signal to a new pitch (135) in the sequence, at least one modulated component, called channelized component, of a tonal contact signal, called harmonized tonal signal, resulting from an excitation contact prior to said excitation contact and change of note, is modulated around a harmonic frequency, said new frequency (133) of synthesis, the new pitch of note (135), corresponding to a harmonic rank different from the harmonic rank of a preceding frequency (132) of synthesis of the channelized component, the audio signal synthesized after said excitation contact and note change is produced by mixing each tonal signal harmonized and contact audio signal of said excitation contact and note change. 2 / - Method according to claim 1, characterized in that the new frequency (133) of synthesis is chosen so that it is, among the harmonic frequencies of the new pitch (135), directly adjacent to said previous frequency (132) of synthesis.3 / - Method according to one of claims 1 or 2, characterized in that several modulated components of the harmonized tonal signal, called low-rank components, whose synthesis frequencies, said frequencies (132) preceding syntheses, effective for a note pitch (134) preceding the new note pitch (135) in the sequence, are less than the synthesis frequency, effective for said previous note pitch (134), of a another modulated component of the harmonized tonal signal, said high-rank component, are each modulated towards a harmonic frequency (133) of said new note pitch (135), selected so as to be among the harmonic frequencies of the new note pitch (135), directly adjacent to the previous synthesis frequency (132) of the low rank component. 4 / - Method according to claim 3, characterized in that at least one high-ranking component is modulated to a harmonic frequency of the new note pitch (135) chosen so as to maintain constant the difference between the ranks of harmonics of the synthesis frequencies of the high rank component and an associated low rank component. 5 / - Method according to one of claims 1 to 4, characterized in that a signal, said note variation signal, representative of a rate of change of note pitch is produced from the note signal, and the new synthesis frequency (133) is chosen from the harmonic frequencies of the new note pitch (135) according to the note variation signal. 6 / - Method according to one of claims 1 to 5, characterized in that a plurality of modulated components of the harmonized tonal signal are each modulated: - according to data, said data of synthetic rank, representative of a previous rank of harmonic , said synthesis rank, which corresponds to a previous frequency (132) of synthesis of the modulated component, and according to prerecorded data comprising, for a plurality of predefined intervals (87, 88, 89) of variation rate of height of Note, data representative of update values of synthetic ranks.7 / - Method according to one of claims 1 to 6, characterized in that the sequencing signal is produced from a signal, said signal ( 7) excitation detection source, representative of the vibration of the vibrating member (2), so that the sequencing signal is representative, for different frequency bands, called excitation bands, of the ignal (7) excitation detection source, at least one excitation intensity value (104, 105, 106) of each excitation contact in each of said excitation bands. 8 / - Method according to one of claims 1 to 7, characterized in that the note signal is produced from a signal (7) representative of the vibration of the vibrating member according to an octave of detection of height of note. 9 / - Synthesis device comprising at least one processing unit (136) adapted to synthesize an audio signal, said synthesized audio signal, from a signal, said note signal, representative of a sequence of pitch levels according to which vibrating member is vibrated, and a sequencing signal representative of a sequence of contacts imparted on a vibrating member (2), said contact sequence comprising excitation contacts adapted to put the member ( 2) vibrating vibration, characterized in that the processing unit (136) is adapted to: - produce at least one audio contact signal, for each of said excitation contacts, - modulate several frequency components, called modulated components, at least one audio contact signal, said tonal contact signal, each successively around frequencies, said frequencies (132, 133) of synthesis, each one of a harmonic rank of a note pitch (134; ) of the sequence, characterized in that: - for at least one excitation contact, said excitation contact and note change, causing a variation of the note signal to a new pitch (135) in the sequence , the processing unit is adapted to be able to modulate at least one modulated component, called channelized component, of a tonal contact signal, called a harmonized tonal signal, resulting from an excitation contact prior to said excitation and change contact. of note, around a harmonic frequency, called the new synthesis frequency (133), of the new pitch (135), corresponding to a harmonic rank different from the harmonic rank of a previous frequency (132) synthesis of the channelized component, the processing unit is adapted to produce, after any excitation contact and note change, the audio signal synthesized by mixing each corresponding harmonized tonal signal and the audiometric signal. o contact said excitation contact and note change. 10 / - Recording medium - in particular removable type (CD-ROM, DVD, USB stick, external electronic hard disk) - adapted to be read in a reader of a computer device, and on which is recorded a program of computer adapted to be loaded into the RAM of the computing device when the recording medium is loaded into said reader, characterized in that the computer program comprises portions of program code for performing the steps of a method of synthesizing an audio signal according to one of claims 1 to 8 when the computer program is loaded into the RAM of the computing device. 11 / - Computer program comprising portions of program code for performing the steps of a method of synthesizing an audio signal according to one of claims 1 to 8 when said program is executed on a computer device.