FR3142639A1 - METHOD FOR CHARACTERIZING A FILTER FOR PROCESSING THE VOICE OF AN INDIVIDUAL, COMMUNICATION DEVICE - Google Patents

Abstract

METHOD FOR CHARACTERIZING A FILTER FOR PROCESSING A VOICE OF AN INDIVIDUAL, COMMUNICATION DEVICE Method for characterizing a filter (F1) for processing a voice of a given subject, said voice being acquired by a microphone arranged in a mouthpiece, said method comprising: Acquisition of a first signal corresponding to a distorted subject's voice; Acquisition of a second signal corresponding to an undistorted subject's voice; Sampling of the first signal; Estimation of average values of the amplitude of the first signal; Estimation of signal speed values using a finite-time homogeneous differentiator method; Calculation of a first set of coefficients; Calculation of an error between a set of samples of the second signal and the samples obtained at the output of the first model, Iteration(s) of the calculation of the coefficients of the first model. Figure for abstract: Fig. 1

Description

METHOD FOR CHARACTERIZING A FILTER FOR PROCESSING THE VOICE OF AN INDIVIDUAL, COMMUNICATION DEVICE Field of the invention

The invention relates to a method for processing and correcting an altered audio stream coming from an individual having, for example, a mouthpiece in their mouth allowing them to breathe underwater and comprising a microphone. The field of the invention relates to methods implemented by a computer embedded in electronic equipment to be executed in real time, without an external connection and configured to make an audio stream distorted by the presence of an element held in place more intelligible. mouth.

State of the art

We know the document US2012/0213034 describing a regulator mouthpiece to be placed in the mouth and comprising a microphone as well as a bone conduction earpiece vibrating an acoustic conduction plate to pass the signal through the user's teeth.

A fault is that the signal captured is hardly intelligible due to the labial part of the mouthpiece preventing the user from moving their lips sufficiently. There is a need to reconstruct intelligible sound using a voice processing process.

A first problem is that the audio flow depends a lot on the individual who pronounces the sentences which are distorted by the presence of the mouthpiece. A second problem is that certain algorithms require significant calculation times which do not allow real-time processing of an individual's voice, on an embedded system, thus making communication between two individuals by means of this difficult. type of equipment. At most it is possible to recover the voice of a swimmer and process it offline.

The invention proposes a method implemented by computer, in particular an embedded computer, such as an electronic control unit, called MCU, a microprocessor, a microcontroller or an FPGA. The method of the invention allows the reconstruction and/or correction of certain syllables by real-time processing of the audio stream acquired by a microphone located in the vicinity of a diver's mouth. The method of the invention makes it possible in particular to transmit or process the audio stream in reception so that another diver can listen intelligibly to the audio stream emitted by a first diver.

• Acquisition d’au moins un premier signal correspondant à la prononciation d’un premier ensemble de phonèmes par ledit sujet donné ; ledit premier signal correspondant à une voix d’un sujet acquise par le micro du dispositif d’aide à la respiration sous-marine lorsque ce dernier est porté par ledit sujet ;
• Acquisition d’au moins un second signal correspondant à la prononciation du premier ensemble de phonèmes par ledit sujet ; ledit second signal correspondant à une voix d’un sujet acquise par un micro sans que le sujet porte un dispositif d’aide à la respiration sous-marine ;
• Pour le premier signal, exécution des étapes suivantes :
  • Echantillonnage du premier signal ;
  • Estimation d’un premier ensemble de valeurs moyennes de l’amplitude du premier signal sur une pluralité de premières fenêtres prédéfinies, chaque première fenêtre comportant une pluralité d’échantillons du premier signal ;
  • Estimation d’un second ensemble de valeurs de vitesse du signal à partir d’une méthode du différenciateur homogène en temps-fini sur la pluralité de premières fenêtres prédéfinies ;
  • Calcul d’un premier ensemble de coefficients d’un système d’équations différentielles représentant un premier modèle dont les solutions sont le premier et le second ensemble de valeurs ;
• Calcul d’une erreur entre au moins un ensemble d’échantillons du second signal et les échantillons obtenus en sortie du premier modèle exécuté avec des valeurs du premier signal et les coefficients estimés,
• Itération(s) du calcul des coefficients du premier modèle en minimisant à chaque itération ladite erreur calculée, le nombre d’itérations étant configuré pour minimiser l’erreur en dessous d’un seuil prédéfini de manière à définir des coefficients d’exploitation, lesdits coefficients d’exploitation définissant les coefficients du filtre.

According to a first aspect, the invention relates to a method for characterizing a filter for processing a voice of a given subject, said voice being acquired by a microphone arranged in an underwater breathing aid device intended to be worn close to the mouth of a subject, said method comprising:

• Acquisition of at least a first signal corresponding to the pronunciation of a first set of phonemes by said given subject; said first signal corresponding to a voice of a subject acquired by the microphone of the underwater breathing aid device when the latter is worn by said subject;
• Acquisition of at least one second signal corresponding to the pronunciation of the first set of phonemes by said subject; said second signal corresponding to a subject's voice acquired by a microphone without the subject wearing an underwater breathing aid device;
• For the first signal, perform the following steps:
  • Sampling of the first signal;
  • Estimation of a first set of average values of the amplitude of the first signal over a plurality of first predefined windows, each first window comprising a plurality of samples of the first signal;
  • Estimation of a second set of signal speed values from a finite-time homogeneous differentiator method over the plurality of first predefined windows;
  • Calculation of a first set of coefficients of a system of differential equations representing a first model whose solutions are the first and the second set of values;
• Calculation of an error between at least one set of samples of the second signal and the samples obtained at the output of the first model executed with values of the first signal and the estimated coefficients,
• Iteration(s) of calculating the coefficients of the first model by minimizing at each iteration said calculated error, the number of iterations being configured to minimize the error below a predefined threshold so as to define operating coefficients, said operating coefficients defining the filter coefficients.

One advantage is to allow characterization of an audio filter in order to then reconstruct an intelligible sound when a voice is distorted by a mouthpiece. An advantage of the method of the invention is to propose an algorithm that requires little computing power and memory to be implemented on an on-board computer. Another advantage is to allow encoding in simple lines of code and comprising a minimum of lines of code. Another advantage is to allow an algorithm that is adaptive to the user and their environment.

According to one embodiment, the characterization method comprises a step of filtering the first signal carried out prior to sampling.

An advantage is to allow a characterization of an audio filter that eliminates noise, that is to say a characterization of the induced distortion of the voice due to the equipment present in the mouth.

• Analyse de la densité spectrale du signal afin de déterminer des seuils caractéristiques d’amplitude ou de puissance ;
• Sélection d’un ensemble de fréquences caractéristiques pour lesquelles les seuils sont dépassés ;
• Echantillonnage de segments du premier signal autour de chaque fréquence caractéristique ;
• Génération d’un ensemble d’échantillons pour le traitement du premier signal.

According to one embodiment, sampling the first signal comprises the following steps:

• Analysis of the spectral density of the signal in order to determine characteristic amplitude or power thresholds;
• Selection of a set of characteristic frequencies for which the thresholds are exceeded;
• Sampling of segments of the first signal around each characteristic frequency;
• Generation of a set of samples for processing the first signal.

An advantage is to focus the calculation times on the areas of interest of the voice, that is to say the characteristic frequencies of the voice of an individual, in particular that which provides a maximum of useful signal.

According to one embodiment, sampling produces between 400 and 96,000 samples for 1 second.

An advantage is to allow different operating modes depending on whether we wish to prioritize calculation time or signal quality.

Sampling at 4000 samples per second makes it possible to keep a voice spectrum between 300 Hz and 1300 Hz and reduce computing power.

According to one embodiment, the method is repeated for a plurality of acquisitions of first signals, each first acquired signal corresponding to the pronunciation of a new set of phonemes by the same subject, the calculation of the error and its minimization being carried out for each new first signal acquired.

An advantage is to carry out the method of the invention for a few phonemes, that is to say a few sentences to allow a filter to be characterized quickly. An advantage is to allow rapid training of the algorithm from 3 or 4 sentences spoken by an individual. Thus, the equipment can be quickly ready, configured and trained just before use, for example just before making a dive.

According to one embodiment, the system of differential equations is of the first order.

An advantage is the simplification of calculations and therefore the saving in calculation time.

• Estimation d’un troisième ensemble de valeurs d’accélération du signal à partir de la méthode du différenciateur homogène en temps-fini sur la pluralité de premières fenêtres prédéfinies ;
• Calcul d’un premier ensemble de coefficients d’un système d’équations différentielles du premier ordre représentant un premier modèle dont les solutions sont le premier et le second ensemble de valeurs.

According to one embodiment, the characterization method comprises:

• Estimation of a third set of signal acceleration values using the finite-time homogeneous differentiator method over the plurality of first predefined windows;
• Calculation of a first set of coefficients of a system of first order differential equations representing a first model whose solutions are the first and second set of values.

An advantage is the improvement in the characterization of the filter allowing better reconstruction of the deformed path of an individual.

• Estimation d’un N^ièmeensemble de valeurs d’une dérivée d’ordre N du signal à partir de la méthode du différenciateur homogène en temps-fini sur la pluralité de premières fenêtres prédéfinies ;
• Calcul d’un premier ensemble de coefficients d’un système d’équations différentielles du N^ièmeordre représentant un premier modèle dont les solutions sont le premier et le second ensemble de valeurs.

According to one embodiment, the characterization method comprises:

• Estimation of an ^Nth set of values of an N-order derivative of the signal using the finite-time homogeneous differentiator method over the plurality of first predefined windows;
• Calculation of a first set of coefficients of a system of ^Nth order differential equations representing a first model whose solutions are the first and the second set of values.

An advantage is the improvement in the characterization of the filter allowing better reconstruction of the deformed path of an individual.

• ,

According to one embodiment, the first model is a linear model and is written:

• ,

where G, F and H are matrices whose coefficients are to be determined, the dimension of the matrix being determined by the order of the linear system of differential equations, y(t) is the signal that we seek to identify approaching the second signal, and x(t) is a state variable internal to the first model.

According to one embodiment, the first model is a non-linear model and is written:

• le procédé comportant en outre, un calcul d’un ensemble de coefficients d’un réseau de neurones à partir d’une régression configurée pour minimiser l’erreur.

where G, F and H are matrices whose coefficients are to be determined, the dimension of the matrix being determined by the order of the system of differential equations, NN designates a non-linear filter, y(t) is the signal that 'we seek to identify approaching the second signal, and x(t) is a state variable internal to the first model,

• the method further comprising a calculation of a set of coefficients of a neural network from a regression configured to minimize the error.

According to one embodiment, the calculation of the error minimization step is carried out from the implementation of a neural network whose coefficients are calculated by a regression to minimize the error.

An advantage is the simplicity and speed of learning.

According to one embodiment, at each iteration a first condition is verified, said first condition being the verification that the error is less than a predefined threshold and/or that a second condition is verified, said second condition being the verification that a predefined number of iterations was carried out, the method comprising a step of generating the filter coefficients as soon as one of the two conditions is verified.

According to another aspect, the invention relates to a computer program product comprising a memory and a calculator for executing code instructions implementing the method of the invention.

An advantage is to allow characterization of a filter on any type of equipment, for example a smart phone, called a Smartphone, or any other mobile electronic equipment from which the characterization filter can be developed then transferred to a diving equipment.

• Acquisition d’un signal correspondant à une voix acquise par un micro agencé dans dispositif d’aide à la respiration sous-marine destiné à être porté dans la bouche d’un sujet, la voix acquise étant définie par un signal audio acquis en temps réel correspondant à une voix d’un sujet déformée par la présence de l’embout buccal en bouche ;
• Application d’un premier modèle entrainé selon la méthode de caractérisation du filtre de l’invention, ladite application visant à obtenir un second signal de voix transformée à partir du premier signal de la voix déformée par un dispositif d’aide à la respiration sous-marine.

According to another aspect, the invention relates to a method of processing a voice of a given subject comprising:

• Acquisition of a signal corresponding to a voice acquired by a microphone arranged in an underwater breathing aid device intended to be worn in the mouth of a subject, the acquired voice being defined by an audio signal acquired in real time corresponding to a subject's voice distorted by the presence of the mouthpiece in the mouth;
• Application of a first model trained according to the filter characterization method of the invention, said application aiming to obtain a second voice signal transformed from the first voice signal distorted by a device for assisting underwater breathing. Marine.

An advantage is to equip equipment in advance with a pre-configuration of the filter and to adapt the training with the voice of a given individual.

• un micro pour acquérir un signal acoustique provenant de la voix du sujet ;

According to another aspect, the invention relates to a communication device comprising a mouthpiece intended to be worn in the mouth of a subject comprising:

• a microphone for acquiring an acoustic signal from the subject's voice;

an electronic card comprising an audio filter, an analog-digital converter, a calculator for sampling the digitized signal, a memory for recording the coefficients of a filter to be applied to the sampled signal and a calculator for generating an output signal, said coefficients recorded being calculated from a method of the invention.

Description of the invention

The invention relates to a first computer-implemented method for configuring a filter for processing and correcting an audio stream. This process is called the process of characterizing a filter for processing a voice.

The invention also relates to a second computer-implemented method intended to be embedded and executed in electronic equipment to communicate in particular underwater. This second method implements a filter characterized by the first method.

These two processes can be carried out as an extension of the other successively or spaced out over time. However, it should be noted that the first process can be executed before the second process or an alternative to the second process. However, it achieves the same end goal of processing the audio stream of a diver with a mouthpiece in their mouth as they speak to make it intelligible.

The invention also relates to the communication device forming an element of an underwater Walkie-Talkie system. This is a voice communication device between divers. This device generally includes a tip containing the electronics necessary to record and reproduce sound. The tip includes an electronic interface such as a cable allowing it to be connected to a box. An electronic card arranged in the housing includes components for recording the coefficients of the filter defining its characterization and for carrying out data processing and transmission and reception of the audio signal.

An advantage of the invention is to create a voice communication device for scuba diving with little or no modification to the diver's equipment. This constraint involves using a regulator tip as a means of receiving and emitting the audible sound signal. The voice being modified by the mouthpiece and its intelligibility thus being reduced, the invention makes it possible to recover part of the degraded intelligibility thanks to the filter which was characterized by the first process.

Microphone

According to one embodiment, the microphone can be fixed on a mechanical part conducting the air flow created by the diver's voice, this mechanical part comprises at least one channel or a light extending between the mouthpiece and the regulator. Thus, according to different embodiments, the microphone can be fixed between the mouthpiece and the regulator so as to capture the audio stream spoken by a diver. The invention relates to all of these embodiments.

According to another example, the microphone can be integrated into an elastic covering which can be protected by a protective wall arranged frontally on the microphone sensor. The protective wall advantageously reduces the difficulties of integrating the microphone into the covering.

The mouthpiece 1 preferably comprises a labial portion 42. The labial portion 42 is intended to be covered by the lips of the user during use. Said labial portion 42 comprises a light or lights allowing the passage of air on either side of the labial portion. Advantageously, this light or these lights allow the passage of air from the channel of the regulator towards the mouth of the subject during use. The labial portion 42 makes it possible to create with the user's lips a tight closure between the external environment and the air passage light(s).

The labial portion 42 extends between a breathing zone comprising the acoustic conduction plates 20 and a connection means 43. The connection means 43 advantageously allows the mouthpiece to cooperate with a diving regulator. According to one embodiment, the connection means 43 is of tubular shape whose walls are rigid or flexible.

Characterization process / treatment process

Figures 1 and 2 relate more particularly to the steps of a process for processing an audio stream PROC ₁ implementing a filter characterized by a method for characterizing the filter according to the invention.

The audio stream processed by the PROC ₁ processing method is denoted S _A (t). This audio stream is acquired and filtered using the filter F ₁ characterized by a first method of characterizing the filter CARAC ₁ . The application of the F ₁ filter makes it possible to reconstruct part of the intelligibility lost by the presence of the EM ₁ tip in an individual's mouth.

The invention therefore relates to a process denoted PROC ₁ on the which consists of applying APP ₁ the filter F ₁ to an acquired input ACQ ₀ of an audio stream corresponding to the voice of an individual, the filter F ₁ being characterized by a first process denoted CARAC ₁ which is described below thanks to in Figures 3 and 4.

Figures 3 and 4 therefore relate more particularly to the steps of a CARAC ₁ characterization process of a filter from a few sequences spoken by a user with a mouthpiece in the mouth and the same sequences pronounced without a mouthpiece in the mouth. The filter is then used for the diver's equipment.

The audio streams processed by the CARAC ₁ characterization process of a filter F ₁ are denoted S ₁ (t) and S ₂ (t).

The CARAC ₁ process includes learning a function characterizing the filter in order to be adapted to correct the deformation induced by the presence of a mouthpiece and to take into account a type of voice, its nature, its range, its stamp, etc.

There represents in particular the passage of the audio and sound flow in coding within different components. For this purpose, the represents the mouthpiece EM ₁ , also noted 1 on the , of an underwater communication device. It may include a portion to be held in the mouth, for example formed by two acoustic conduction plates 20. The tip 1 preferably comprises a labial portion 42 and a connection 43 with, for example, a regulator (not shown). Openings 61 and 62 are preferably provided for the passage of air between the mouth and the regulator. A MIC microphone (not shown on the ) makes it possible to acquire the sounds produced by an individual speaking in the mouthpiece 1 before transmitting the signal to an analog digital converter CAN ₁ . The invention relates to any other mechanical architecture of a mouthpiece 1 intended to be connected to a regulator. In particular, the device of the invention can include different variants, in particular on the number of channels envisaged, on the presence or not of acoustic conduction plates, their arrangement and their material.

In the example of the , a vibrating element 3 can be used to transcribe the audio emitted by another individual and thus allow listening by bone conduction. Another system could, however, be used to transcribe the voice and its intelligibility which has been processed by the method of the invention.

Phonemes for learning

There describes two branches of a process. The first branch comprises a step ACQ ₁ of acquiring an incoming audio stream S ₁ (t) comprising a first set of phonemes distorted by the mouthpiece 1. The second branch comprises the acquisition of an incoming audio stream S ₂ (t) comprising the same set of phonemes defining a second audio stream not distorted by the presence of a mouthpiece 1. Preferably, the phonemes are the same and are preferentially acquired by the same user. Indeed, in order to allow learning and definition of the coefficients of the optimal filter, it is necessary to isolate as much as possible the influence of the deformation linked to the presence of the tip 1 in the mouth. Thus, the invention is particularly relevant in the case where the two sequences of phonemes are substantially the same. Learning will also be even more efficient when the same individual pronounces the two audio streams S ₁ (t) and S ₂ (t)

The phonemes can for example correspond to a few sentences to be pronounced with and without a mouthpiece 1. The invention thus proposes an embodiment in which an individual pronounces two same groups of sentences with and without a mouthpiece.

Phonemes can correspond to a grouping of letters, syllables or sentences. The two groups preferably include the same group of phonemes and the same ordering of these phonemes within the sequence pronounced by an individual, with and without a device.

We understand that learning an algorithm based on a few spoken sentences, for example between 1 and 10 sentences, allows rapid learning for the user.

The recording of the audio stream made with the mouthpiece is processed in particular by a differentiator. The recording of the audio stream made without the mouthpiece is used as the desired output when creating the filter.

A differentiator of the invention may comprise a set of components for performing the mathematical operations of differentiation, i.e. providing an output proportional to the derivative of the input with respect to one or more variables.

According to a first variant, the invention makes it possible to consider in the algorithm all of the pronounced sentences comprising a plurality of letters, syllables or sentences. In this case, they are all processed together.

According to a second variant, the invention makes it possible to consider each sentence in the algorithm sequentially and independently of each other. In this case, each spoken sentence includes a plurality of letters, syllables and each sentence is processed independently of the others.

The sentences can be defined in such a way as to obtain a distribution of syllables most representative of the language that can be used in an underwater environment and in a given language.

According to an exemplary embodiment, the sentences selected for learning and characterization of the filter F ₁ comprise a representative sample of the different labials, in particular syllables comprising the following letters: B, P, M, F, V.

According to one embodiment, the sentences can be displayed on a display of the diving equipment so that the latter reads them. According to another example, the sentences are displayed on a display of an electronic terminal such as that of a Smartphone by means of an application allowing a language to be selected. Thus, we understand that the CARAC ₁ characterization method can be carried out according to the embodiments directly on the diving equipment or on a Smartphone. When the filter is thus characterized from a Smartphone, different possibilities can be implemented to transfer the characterized filter to a memory of an electronic card placed on the diving equipment, such as the mouthpiece.

According to one embodiment, the display of sentences includes a cursor making it possible to indicate a reading speed. The display may include graphic indications making it possible to indicate to a user a reading sequence with or without a mouthpiece in order to acquire the two sets of sentences.

F1 filter calculation resources

In order to carry out this preliminary step, according to a first embodiment, a user can use an electronic terminal such as a Smartphone, that is to say a smart phone or any other arrangement comprising a microphone. The two sequences of spoken sentences can be recorded using the microphone of the electronic terminal and transmitted to a remote server to carry out the method of the invention. According to this mode, the process is carried out remotely and can benefit from significant calculation resources.

In particular, according to a preferred mode, the characteristics of the filter can be calculated by a remote server having significant calculation resources without hardware constraints linked to a particular mobile equipment. For this purpose, the voice signal emitted by a user can be acquired by a mobile electronic device, such as a Smartphone, otherwise called a smart phone, and transmitted to a remote server. The filter calculated by the remote server can then be re-transmitted to the Smartphone or on-board electronic diving equipment.

According to a second embodiment, a user can use diving equipment comprising an electronic module comprising a computer and a memory. In this case, learning is carried out using fewer resources than a server. The algorithm is then optimized to be executed in a constrained environment. An advantage is not needing a network connection, this case can be interesting in a sea diving environment for example when you are on a boat far from the coast. Another advantage is to use the microphone of the diving equipment. Thus, learning will be all the more effective as it overcomes the gaps linked to differences in equipment between the equipment used for learning and that used for operating the characterized filter.

According to a third hybrid embodiment, the filter can be calculated on an electronic terminal of the Smartphone type from a downloaded application making it possible to exploit the calculation resources of the electronic terminal. An advantage is to produce the filter with a little more resources than the case of an electronic card embedded in the diving equipment while allowing the operation to be carried out locally, that is to say without implementing a 'a remote server.

When the operation is carried out by an electronic terminal of the Smartphone type, a wireless connection can be provided allowing the loading of the filter F ₁ calculated by a resource other than that of the diving equipment so that it is recorded in a memory or resource of said diving equipment.

Audio signal acquisition mode

The signal from the microphone with the mouthpiece arranged in the mouth of an individual is called the first signal S ₁ , the signal acquired by a microphone without the mouthpiece being arranged in the mouth of an individual is called the second signal S ₂ . These signals S ₁ and S ₂ are acquired with the aim of characterizing an audio filter.

Preferably, the same and single microphone is used to acquire the two signals S ₁ and S ₂ . This embodiment makes it possible to overcome the intrinsic characteristics of the microphone and the influence of the latter on the acquired signal. According to one embodiment, this microphone is the microphone of the diving equipment, for example that arranged in the part of the distal tip located between the part intended to be positioned in the mouth and the regulator. According to another embodiment, the microphone is the microphone of an electronic terminal such as that of a smartphone type mobile telephone. In the latter case, an individual positions the mouthpiece 1 in the mouth and says the few sentences near the microphone of the electronic terminal to train the algorithm executed by the latter. According to one embodiment, a known corrective filter can be used to correct the signal acquired by the Smartphone.

Other embodiments can be implemented, in particular modes involving the use of signals acquired by different microphones.

According to one embodiment, the two signals S ₁ and S ₂ encoded digitally after their conversion in an analog – digital converter, called CAN. The conversion and encoding advantageously include sampling of each signal.

In the same way during the acquisition of the signal in the processing method using a characterized filter, an analog - digital conversion step is implemented and sampling is carried out of the acquired audio signal.

According to one embodiment, the first signal S ₁ corresponding to the sentences pronounced with the mouthpiece in the mouth are sampled according to a first ECH ₁ sampling. The sampling of the signals implemented in the CARAC ₁ characterization process of the F ₁ filter and in the PROC ₁ processing process can be identical, that is to say that the sampling frequency, the measurements carried out on the samples by grouping the latter by considering time windows are configured identically. However, the invention could implement sampling specific to the characterization of the filter F ₁ and sampling specific to the process of processing the audio stream spoken by a diver. Typically, in this embodiment, the sampling frequencies could be different. An advantage is to consume less information during the processing stages during a dive in which we wish to keep the communication link as long as possible while in the learning and characterization phase of the filter F ₁ , sampling finer, that is to say allowing more samples to be obtained, makes it possible to obtain a more efficient F ₁ filter.

According to an exemplary embodiment, sampling leading to the recording of digital values is configured for a high value of 96,000 samples per second. However, a lower number of samples can also allow very good results to be obtained with a value between 4000 and 8000 samples per second.

When acquiring the audio signal, a number of sliding samples is kept in memory during real-time calculation cycles of the average values. A number of samples of between 200 and 400 samples is kept for each real-time calculation cycle.

Additional band-pass or low-pass filter

According to an exemplary embodiment, prior filtering can be carried out before sampling. This last filtering aims to eliminate noise or restrict the signal analysis band.

According to one example, the invention comprises the implementation of a low-pass filter whose cutoff frequency depends on the sampling frequency. According to an exemplary embodiment, a suitable filter makes it possible to filter the effect produced by the bubbles during breathing or when emitting a sound through the mouth of a diver. The specific frequency can be filtered so as to improve intelligibility upstream of the construction of the characteristic filter F ₁ .

Sample selection

According to an exemplary embodiment, in order to reduce the calculation time, a selection of samples is chosen. To this end, the selection of a set of samples amounts to choosing a set of time windows Tf ₁ each comprising a subset of samples from which the average values will be calculated.

In order to select the best samples or the best groups of samples to define an efficient characteristic filter F ₁ , a step of calculating characteristic frequencies of the spectrum of the acquired signal can be carried out. An advantage is to select a few frequencies comprising either the greatest signal intensities or the frequencies having the greatest spectral density, for example by considering the spectral power of the acquired signal. These selected frequencies make it possible to select an interval around the frequency to retain samples characteristic of the audio stream.

This method saves time and calculations and therefore makes it possible to embed the method of the invention in an electronic component on which learning can be carried out to define the characteristic filter F ₁ quickly.

According to an example between 5 and 20 characteristic frequencies can be configured to determine between 5 and 20 intervals around each frequency making it possible to define between 5 and 20 sample time windows.

Speed and acceleration of spoken sounds

There represents an output S ₁ (t) corresponding to the audio stream acquired with a tip and an output S ₂ (t) corresponding to the audio stream acquired without a tip. We recall that the sentences made up of phonemes in each signal S ₁ (t) and S ₂ (t) are identical in this example. We notice in the graph of the that the duration of the audio stream S ₁ (t) is longer than the duration of the audio stream S ₂ (t) due to the fact that the speed of speech is slowed down by the presence of the mouthpiece in the mouth.

The method of the invention makes it possible to take into account in the model used the distortion of the sound induced by the speed of speech and therefore of the sound of the audio stream which is recorded and filtered.

An advantage of the invention is to take into consideration in the characterization of the filter the signal artifacts linked to the deformation of the signal induced by the change of speed and the acceleration of the sound pronounced by a diver.

Estimation of average quantities

The process of the invention is based on the implementation of a method based on the theory of homogeneity by the definition of a homogeneous differentiator. This method makes it possible to carry out a finite-time estimation of noisy signals. An advantage of the method of the invention is that the algorithms implemented are simple, particularly fast and non-asymptotic in nature unlike traditional methods.

This method allows the estimation of the derivatives of the signal according to the chosen order by considering all of the sampled data. In order to smooth the estimates and reduce artifacts linked to acquisition errors or noisy signals, average values are calculated in order to make the method of the invention more robust.

The method makes it possible to estimate EST ₁ a first set of values of data relating to the signal acquired over predefined time windows Tf ₁ , each first window Tf ₁ comprising a given number of samples of the signal.

Thus, according to an example, the physical quantity relating to the acquired signal is the amplitude of the signal. According to one embodiment, the method makes it possible to estimate the average values of the amplitude of the signal taken over a set of samples of a given time window.

Other physical quantities can be used or processed according to the method of the invention, however the embodiment if detailed afterward is relative to other processing of the average values of the amplitudes of the acquired signal.

The method takes into consideration a value linked to the dynamics of the signal: its speed. For this purpose, the method of the invention makes it possible to estimate EST ₂ a second set of physical quantities relating to the dynamics of the signal. According to one example, the speed values of the first signal S ₁ are estimated. Speed refers to how quickly sentences are spoken. In order to calculate the speed of the first signal S ₁ , a method called the finite-time homogeneous differentiator DHT is implemented. The differentiator is applied to a plurality of first predefined time windows each comprising a set of samples.

According to an improved example, the method takes into consideration another value linked to the dynamics of the signal: its acceleration. For this purpose, the method of the invention makes it possible to estimate EST ₃ a third set of physical quantities relating to the dynamics of the signal. According to one example, the acceleration values of the first signal S ₁ are estimated. Acceleration refers to variations in the speed with which sentences are spoken. In order to calculate the acceleration of the first signal S ₁ , an equivalent method called the homogeneous finite time differentiator DHT is also implemented. The differentiator is applied to a plurality of first predefined time windows each comprising a set of samples. The samples are preferably the same as those considered to estimate the speeds.

In the same way, during the PROC ₁ processing method, the values of the derivatives and accelerations of the input signal S _A (t) which comes from a signal corresponding to an audio stream of a diver acquired with a mouthpiece in the mouth can be calculated. An interest is to generate the output stream S _B (t) from the filter F ₁ characterized by the CARAC ₁ characterization method obtained in the learning phase.

Dynamic system modeling

The method of the invention makes it possible to define a model based on the modeling of dynamic systems and to solve the problem posed by obtaining an approximate solution by optimizing the calculations by iteration while minimizing an error. Error can be defined by knowing the system input and system output. A problem solved by the invention is to define a model which converges quickly enough to define an operational and efficient filter F ₁ making it possible to subsequently reconstruct the audio stream distorted by the presence of the tip.

The signal S ₂ (t) designates the normal input voice which has not undergone distortion, and can define the desired output of the model. The model of distortion of the voice due to the presence of the mouthpiece can be represented by this equality .

An advantage of modeling by a dynamic system is to introduce a differentiation operator with respect to the time variable t.- In such modeling, we can write , where p is the differentiation operator with respect to the time variable, p = d/dt.

The signal S ₂ (t) defines the output of the dynamic system in the model learning phase. It is therefore appropriate to define the inverse problem consisting of finding S ₂ (t) from S ₁ (t).

The inverse problem amounts to calculating the inverse function W ^-1 which represents a learning model that we seek to optimize in order to calculate the signal S _B (t) represented on the corresponding to the corrected audio stream when the known signal S ₂ (t) will no longer be available. The first audio sequences S ₂ (t) therefore make it possible to carry out training of the model in order to parameterize a trained audio filter F ₁ , that is to say whose coefficients have been produced by an algorithm comprising a regression operation or d iteration aimed at minimizing an error.

The coefficients thus produced are the operating coefficients making it possible to define an operational F1 filter.

The invention therefore makes it possible to generate a “learned model” to subsequently use a characteristic filter F ₁ within the equipment to correct the audio stream distorted by the presence of the mouthpiece in the mouth.

The opposite problem can be written as: .

Linear model

Which can be written in terms of the theory of dynamic systems as the following equation according to a linear model MOD ₁ where x(t) is a variable internal to the system:

According to one embodiment, in order to solve this system, a first hypothesis is to consider a sufficiently smooth signal S ₂ (t) in order to consider a differentiable function having a set of derivatives constituting different orders of the function S ₂ (t) = y(t) where y(t) is the output of the MOD ₁ or MOD ₂ model. Thus, the set of derivatives is written [y ⁽ⁿ⁾ ], “n” being the order of the derivative of the sampled signal and being greater than or equal to 1.

The desired output can be modeled by a vector z defined by the system of differential equations that we seek to estimate:

z = [y ⁽¹⁾ , y ⁽²⁾ , y ⁽³⁾ , … y ^{(n -1 )} ] ^T and

y ⁽ⁿ⁾ (t) = Theta(t)

This is the set of first, second, third derivatives and so on up to the ^nth derivative. The number of derivatives considered defines the dimension of the system.

By considering an order of derivatives retained equal to 2 to simplify the calculations, it is possible to define a simple MOD ₁ model of the system which is efficient. The following is described for an order n in order to illustrate a more general embodiment.

We can define the following equivalent system:

In which MA is defined by the following matrix:

And “C” is a vector defined by the following vector:

C = [1, 0, 0…, 0]

And is a vector defined by the following vector:

Phi(t) = [0, …, Theta(t)] ^T

We notice , where alpha is strictly greater than 0 and x is a real number.

The finite-time homogeneous differentiator is defined by the following system for i = 2, …, n-1:

Or is the estimate of z

The coefficients k ₁ , …, k _n form a Hurwitz polynomial.

It is possible in the resolution of this system to choose in an appropriate manner to ensure the homogeneity of the system the coefficients (a ₁ , …a _n ) in the space of strictly positive real numbers of dimension n.

According to an example, for a given A in the interval [1 – 1/(n-1), 1], the sequences (r ₁ , …, r _n ) and (a ₁ , …a _n ) can be chosen as follows :

, with 1 ≤ i ≤ n

, with 1 ≤ i ≤ n

In order to ensure convergence in finite time, according to an example, the method can be modeled by taking into account the conditions necessary for the theory of homogeneity. In this hypothesis, it is possible to write that

^{( j -1)} (t) = _j (t), j = 1, …, n after a finite time of the transients.

• (1)

The dynamics of differentiation errors takes the form of the following system:

• (1)

We know, by having chosen the coefficients which can define a Hurwitz polynomial, that there exists A in the interval [1-1/(n-1), 1] sufficiently close to 1 such that equation (1) is generally stable at finite time.

One of the main advantages of stable homogeneous systems in finite time is the speed of their convergence rate and their robustness with respect to different disturbances.

Due to the term Theta(t), it is impossible to get the error to converge to zero without having additional knowledge about this Theta(t) signal. This problem can be overcome by assuming that y(t) is locally polynomial and that over a small time interval, Theta(t) = 0. In this case, it is then possible to recover the derivatives.

According to one embodiment, the method comprising a number of iterations aimed at repeating the coefficient calculation sequence until a calculated error is less than a given threshold.

In order to quantify the performance of the results obtained with the final algorithm, an error is defined based on the standard of the difference of the reconstructed sounds and original sounds.

According to one embodiment, at each iteration a first condition C ₁ is verified. According to an example, the first condition C ₁ corresponds to the verification that the error is less than a predefined threshold. According to one example, a second condition C ₂ is verified. The second condition C ₂ is the verification that a predefined number of iterations has been carried out.

Advantageously, the method includes a step of generating the filter coefficients as soon as one of the two conditions C ₁ or C ₂ is verified.

The succession of iteration steps corresponds to the learning step of the algorithm which results in producing filter coefficients making it possible to reconstruct the signal S _B (t) from an input signal S _A (t ) according to the PROC ₁ treatment method of the invention.

Nonlinear model

According to an alternative embodiment, a nonlinear MOD ₂ model can be used to implement the invention.

Where NN denotes a non-linear filter.

In this implementation the filter F ₁ = NN corresponds to an additional correction which makes it possible to correct the modeling error of the linear model. In this case, the error is minimized by taking into consideration the nonlinear part NN.

In the case of an implementation of a nonlinear model MOD ₂ , the filter F ₁ is a nonlinear filter. According to one example, a method for calculating the coefficients of the nonlinear filter F ₁ can be implemented using a neural network. For this purpose, a forward propagation neural network is produced with the result obtained with the linear filter. According to one example, the invention comprises the implementation of a CNN type network, defining a convolutional neural network. In this example implementation, a conventional feedforward network with a hidden layer that uses sigmoidal activation functions achieves good learning performance.

In the case of a non-linear filter, learning also includes the phase of calculating the network coefficients by a regression carried out at each iteration.

According to other examples, other neural networks can be used.

The method of the invention is based on the implementation of an algorithm allowing the calculation of coefficients integrating iterative phases comprising the calculation of an error. This algorithm is called a “learning algorithm” or “learning function” insofar as it takes into account intermediate steps making it possible to converge towards a result leading to the construction of the characteristic filter F ₁ .

Once the filter F ₁ has been determined, all the parameters of the MOD ₁ model for the linear model and of the MOD ₂ model for the non-linear model are determined. The models can then be applied to predict the sorite S _B (t) on the .

Figures 3 and 4 represent the step of calculating the error er(t) at each iteration, considering S ₂ (t) acquired directly and the output y(t) estimated.

According to one embodiment, an initial filter Fi is pre-trained and recorded in a device memory. Such an initial filter Fi can be trained with different voice profiles such as female voice profiles, male voice profiles, child voice profiles. Certain vocal range or timbre can be used to pre-train an initial Fi filter within the equipment. An interest is to define initial coefficients making it possible to improve the convergence of the error below a given threshold during learning.

In order to define an initial filter Fi, the CARAC ₁ characterization method of the invention can be used with a first configuration. The first configuration can benefit from additional calculation resources, thus the order of the system can be increased, as well as the samples or even a greater number of training sentences can be available. In this configuration, a non-linear MOD ₂ model can be used to define the initial filter Fi and one or the linear model can be used to train the model with the real voice of the diver.

According to another embodiment, the PROC1 processing method of the invention can be implemented to correct an error also during the diving phase, for example with supervised or unsupervised training.

For example, a repeated sentence or a signal indicating poor understanding can be used to label model outputs in order to improve the construction of a filter F2 during the PROC ₁ processing process.

According to an exemplary embodiment, the method of the invention can comprise three successive trainings of the model or models which can be carried out at times spaced apart from each other. A first training of the model makes it possible to define an initial filter Fi from a first configuration of the CARAC ₁ characterization process, for example with a pre-selected voice corresponding to a first signal S2(t).

A second training of the model makes it possible to define a characteristic filter F ₁ from a second configuration of the CARAC ₁ characterization process with the diver's voice corresponding to a second signal S2(t). This second training can be carried out with initial model coefficients which were calculated from the first training.

• Les répétitions de phrases prononcées à des mêmes vitesses ou à des vitesses différentes ;
• L’identification de phrase prédéfinie telle que « pas compris », « je ne comprends pas », « peux-tu répéter », etc ;
• Un changement de respiration ou une détection d’un taux de bulles ;
• Etc.

Finally, a third training can be carried out during communications using the PROC1 processing method and automatic labeling of the outputs which can be deduced from the analysis of the audio stream of one or both divers, such as for example:

• Repetitions of sentences spoken at the same or different speeds;
• Predefined phrase identification such as “don’t understand”, “I don’t understand”, “can you repeat”, etc.;
• A change in breathing or detection of a level of bubbles;
• Etc.

Conversion of the filtered and transmitted signal for bone conduction

According to one embodiment, the output filtered by the filter F ₁ characterized by the CARAC ₁ characterization method of the filter is transmitted via a wireless communication interface to another communication device of another swimmer. In one example, the filtered audio output is modulated to a carrier frequency. The invention is compatible with any form of wireless transmission by electromagnetic means such as a radio link or by emission of sound or ultrasonic waves. Different modulations can be used, whether frequency or amplitude modulation. According to one embodiment, the filtered audio signal is transmitted by being modulated by a frequency escape technique.

The mouthpiece 1 may comprise signal transmission means connected to the microphone 5 and/or the vibrating element 3. The transmission means may comprise electronic cables and/or a conduction layer. These transmission means can be at least partially integrated into an elastic covering.

When the signal transmitted from one piece of equipment to another is demodulated on reception, the audio signal can be converted to vibrate a bone conduction plate. There represents an example of an acoustic conduction plate 20 arranged so as to be bitten by the user wearing the mouthpiece 1. According to this example, the mouthpiece 1 further comprises a vibrating element of the transducer type. The vibrating element is connected to the conduction plate so as to transmit vibrations from the vibrating element to the acoustic conduction plate.

The vibrating element 3 is configured to convert audio input to acoustic output and acoustically couple to the upper and/or lower teeth of the diver to conduct the acoustic output from the upper teeth of the diver through the skull to propagate the acoustic signal up to the diver's inner ear via the skull and jaw bones when the diver wears the tip in the mouth.

The acoustic conduction plate 20 is designed and arranged to acoustically couple to the upper teeth of the user to conduct the acoustic output of the upper teeth of the diver through the skull to the cochlea to generate audible sound in at least one of the diver's inner ears when the diver wears the mouthpiece 1.

The acoustic conduction plate 20 is configured to acoustically engage and couple with the surface of the plunger teeth and is configured to conduct the vibrations of the vibrating element 3 in response to an electrical signal. The vibration of the vibrating element 3 produces an acoustic output signal which is acoustically conducted to the diver's teeth, via the acoustic conduction plate, then through his jaw bones and skull to the inner ear , including the cochlea, where it is perceived as a sound.

Claims (15)

Characterization process (CARAC₁) of a filter (F₁) for processing (PROC₁) of a voice of a given subject, said voice being acquired by a microphone arranged in an underwater breathing aid device intended to be worn near the mouth of a subject, said method comprising:

• Acquisition (ACQ1) of at least a first signal (S ₁ ) corresponding to the pronunciation of a first set of phonemes by said given subject; said first signal (S ₁ ) corresponding to a voice of a subject acquired by the microphone of the underwater breathing aid device when the latter is worn by said subject;
• Acquisition (ACQ2) of at least one second signal (S ₂ ) corresponding to the pronunciation of the first set of phonemes (PHR ₁ ) by said subject; said second signal (S ₂ ) corresponding to a subject's voice acquired by a microphone without the subject wearing an underwater breathing aid;
• For the first signal (S ₁ ), execution of the following steps:
  • Sampling (ECH ₁ ) of the first signal (S ₁ ) to generate a plurality of samples from the first signal (S ₁ );
  • Estimation (EST ₁ ) of a first set (ENS ₁ ) of average values of the amplitude of the first signal (S ₁ ) over a plurality of first predefined windows, each first window comprising a plurality of samples of the first signal (S ₁ );
  • Estimation (EST ₂ ) of a second set (ENS ₂ ) of speed values of the signal (S ₁ ) from a finite-time homogeneous differentiator method over the plurality of first predefined windows;
  • Calculation (EST _F ) of a first set of coefficients of a system of differential equations representing a first model (MOD _1, MOD ₂ ) whose solutions are the first and the second set of values (ENS ₁ , ENS ₂ ) ;
• Calculation of an error (Er) between at least one set of samples of the second signal (S₂) and the samples obtained at the output of the first model (MOD₁, MOD₂) executed with values of the first signal (S₁) and the estimated coefficients (G, F, H),
• Iteration(s) of calculating the coefficients of the first model (MOD₁, MOD₂) by minimizing at each iteration said calculated error (Er), the number of iterations being configured to minimize the error below a predefined threshold so as to define operating coefficients, said operating coefficients defining the coefficients of the filter (F₁).

Characterization method according to claim 1, characterized in that it comprises a step of filtering the first signal (S ₁ ) carried out prior to sampling. Characterization method according to any one of claims 1 to 2, characterized in that the sampling of the first signal (S₁) includes the following steps:

• Analysis of the spectral density of the signal (S ₁ ) in order to determine characteristic amplitude or power thresholds;
• Selection of a set of characteristic frequencies for which said determined thresholds are exceeded;
• Sampling of segments of the first signal (S ₁ ) around each selected characteristic frequency;
• Generation of a set of samples (ECH ₁ ) for processing the first signal (S ₁ ) from each sampled segment.

Characterization method according to any one of claims 1 to 3, characterized in that the sampling (ECH ₁ ) produces between 400 and 96,000 samples for 1 second. Characterization method according to any one of claims 1 to 4 characterized in that the method is repeated for a plurality of acquisitions of first signals (S ₁ ), each first signal (S ₁ ) acquired corresponding to the pronunciation of a new set of phonemes by the same subject, the calculation of the error (Er) and its minimization being carried out for each new first signal (S ₁ ) acquired. Characterization method according to any one of claims 1 to 5 characterized in that the system of differential equations is of the first order. Characterization method according to any one of claims 1 to 6 characterized in that it comprises:

• Estimation (EST ₃ ) of a third set (ENS ₃ ) of acceleration values of the signal (S ₁ ) using the finite-time homogeneous differentiator method over the plurality of first predefined windows;
• Calculation (EST_F) of a first set of coefficients of a system of first order differential equations representing a first model (MOD₁, MOD₂) whose solutions are the first and second set of values (ENS₁, ENS_2,ENS₃).

Characterization method according to any one of claims 1 to 7 characterized in that it comprises:

• Estimation (EST _N ) of an ^Nth set (ENS _N ) of values of a derivative of order N of the signal (S ₁ ) using the finite-time homogeneous differentiator method over the plurality of first predefined windows ;
• Calculation (EST_F) of a first set of coefficients of a system of differential equations of N^thorder representing a first model (MOD₁, MOD₂) whose solutions are the first and second set of values (ENS₁, ENS_{2, …,}ENS_NOT).

Characterization method according to any one of claims 1 to 8 characterized in that the first model is a linear model (MOD1) and is written:

• ,

where G, F and H are matrices whose coefficients are to be determined, the dimension of the matrix being determined by the order of the linear system of differential equations, y(t) is the signal that we seek to identify approaching the second signal (S₂), and x(t) is a state variable internal to the first model (MOD₁). Characterization method according to any one of claims 1 to 8 characterized in that the first model is a non-linear model (MOD₂) and is written:

• ,

where G, F and H are matrices whose coefficients are to be determined, the dimension of the matrix being determined by the order of the system of differential equations, NN designates a non-linear filter, y(t) is the signal that 'we seek to identify approaching the second signal (S₂), and x(t) is a state variable internal to the first model (MOD₂),

• the method further comprising a calculation (EST _NN ) of a set of coefficients of a neural network from a regression configured to minimize the error (Er).

Characterization method according to any one of claims 1 to 10 characterized in that the calculation of the minimization step (EST _F , EST _NN ) of the error (Er) is carried out from the implementation of a neural network whose coefficients are calculated by regression to minimize the error (Er). Characterization method according to any one of claims 1 to 11 characterized in that at each iteration a first condition is verified, said first condition being the verification that the error is less than a predefined threshold and/or that a second condition is verified, said second condition being the verification that a predefined number of iterations has been carried out, the method comprising a step of generating the filter coefficients as soon as one of the two conditions is verified. Treatment process (PROC₁) in a voice of a given subject including:

• Acquisition of a signal (S _A (t)) corresponding to a voice acquired by a microphone arranged in an underwater breathing aid device intended to be worn in the mouth of a subject, the acquired voice being defined by an audio signal acquired in real time ( _SA ) corresponding to a subject's voice distorted by the presence of the mouthpiece in the mouth;
• Application (APP ₁ ) of a first model (MOD ₁ ) trained according to the method of characterizing the filter (F ₁ ) of any one of claims 1 to 12, said application (APP ₁ ) aiming to obtain a second signal ( S _B ) of voice transformed from the first signal (S _A ) of the voice distorted by an underwater breathing aid device (EB ₁ ).

Communication device comprising a mouthpiece (1) intended to be worn in the mouth of a subject comprising:

• a microphone for acquiring an acoustic signal from the subject's voice;

an electronic card comprising an audio filter, an analog-digital converter, a calculator for sampling the digitized signal, a memory for recording the coefficients of a filter (F₁) to be applied to the sampled signal (S_HAS(t)) and a calculator making it possible to generate an output signal (SB(t)), said recorded coefficients being calculated from a method according to any one of claims 1 to 12. Computer program product comprising instructions which, when the program is executed by the communication device, cause said communication device to implement the steps of the method of any one of claims 1 to 12.