FR2566658A1 - Multichannel auditory prosthesis. - Google Patents

Abstract

The invention relates to a multichannel auditory prosthesis of the type comprising a microphone 1 and an amplifier 2, followed by at least two processing channels 4, 5 each provided with a separating filter 9, 10, at least one dynamics compressor 11, 12 and a compensating filter 13, 14, the said channels being reunited at an adder 6 followed by a power amplifier and an output transducer, in particular an earpiece 8; according to the invention, the prosthesis comprises a prior compression stage 3 mounted downstream of the preamplifier 2 and upstream of the separating filters 9, 10, with a view to providing, before the differential processing, a prior compression of the signal over the full frequency range; the filter/separator 9, 10 of each channel is furthermore adapted in order to have a steep slope at least equal to 20 dB/octave and preferably between 40 and 52 dB/octave.

Description

MULTI-CHANNEL HEARING ROUTHESIS
The invention relates to a multi-channel hearing aid comprising a microphone, a preamplifier, at least two differentiated processing channels combined in a summing device, a power amplifier and an output transducer, in particular a headphone.

 Almost all of the existing hearing aids are of the single channel type, comprising a single processing chair which performs the amplification, filtering and possibly dynamic compression functions for the entire input signal from the microphone. Dynamic compression prostheses have appeared more recently and are making significant progress compared to previous prostheses because dynamic compression makes it possible to adapt the dynamics of natural sounds to the acoustic dynamics of the hearing impaired (most hearing impairments resulting in effect a change, depending on the frequency, the perception and pain thresholds: "recruitment effect").

 However in a single-channel prosthesis with dynamic compression, the dynamic adaptation is only carried out over a limited frequency range. To overcome this insufficiency, so-called multi-path prostheses have been designed, comprising at least two treatment chains in parallel, carrying out treatments differentiated according to the frequencies. The diagram of these prostheses comprises atasi, at the output of a preamplifier directly in series with a microphone, a division of the signal through filters and several differentiated processing channels succeeding these filters, each channel being provided with a compressor. dynamic in the most recent versions (DE patent 2641675 Bosch; US patent 4025723 Blackledge).

It is thus hoped to improve the performance of the prostheses at the cost of greater complexity of the circuits and therefore of increased cost.

 However, such a multi-channel design raises difficulties in terms of the very principle of the functioning of the prosthesis.

 Indeed, a first idea would consist in carrying out, at the entry of each channel, a filtering of the radi cal type (steep filter) which clearly separates the frequency band into distinct ranges. However, in this design, poor performance is achieved, in particular in the most frequent case where the significant signal to be heard is temporally offset from the background noise: in this case the compression carried out on the significant signal channel has a preset rate , insufficient in noisy atmosphere, which makes lose most of its interest in dynamic compression.

 Another idea consists in carrying out, at the entry of each channel, a filtering with low slope, the signals of different frequencies then crossing at least partly each channel; a multi-channel prosthesis of this type is described in the above-mentioned patent DE 2641675. However, in this case, we lose in part, in terms of compression, the advantage of the multi-channel design since the treatments operated according to the frequencies are no longer as clearly differentiated.

In fact, such prostheses are practically equivalent as regards compression, to single-track prostheses (with however an additional advantage, consisting in allowing greater flexibility in the choice of the response curve); the slight improvement in performance combined with the surcharge caused by the multi-channel structure explains why this type of prosthesis has not been the subject of dissemination on the market at present.

 The present invention proposes to provide a new multi-channel prosthesis, making it possible to take full advantage of the multi-channel design without having the aforementioned defects.

 An objective of the invention is in particular to obtain a relevant compression of the dynamics of the signals in particular in noisy surroundings, while carrying out on each of the channels a truly differentiated processing.

 Another objective of the invention is to provide a universal multichannel prosthesis, which can be adjusted beforehand by the practitioner according to the specific impairments of the hearing impaired.

To this end, the hearing aid targeted by the invention comprises a microphone and a preamplifier, followed by at least two differentiated processing channels, each provided with a separating filter capable of separating a predetermined frequency band, from at least one compressor. dynamics and a compensating filter, said channels being combined in an adder followed by a power amplifier and an output transducer; according to the present invention
on the one hand, the prosthesis comprises a stage of prior compression, located downstream of the preamplifier and upstream of the separating filters, in order to ensure before the differentiated treatment a prior compression of the
the signal on all / frequency range,
on the other hand, the filter-separator of each channel is adapted to have a slope at least equal to 20 dB / octave, preferably between 40 and 52 dB / octave.

Thus in such a prosthesis, is combined with the compression carried out on each channel, a prior compression carried out before separation on the entire signal from the preamplifier. This prior compression, associated with sufficiently severe filtering and selective compressions on each channel, makes it possible, as will be understood better below, to carry out a global processing which, at the same time, takes account of the signal ratio
noise and has a clearly differentiated character as a function of the frequency ranges so as to best adapt the output dynamics to the dynamics of the hearing impaired, a function of frequency. It is essential to note that the differentiated compression is carried out on each channel on a signal already compressed during a prior phase so that the significant signals are in each channel at levels sufficient to trigger the compressors of said channel with a relevant compression ratio, even if the background noise is in a frequency range different from that of the channel considered.

 The preliminary compression stage is advantageously constituted by a compressor (or by several compressors in series), arranged in series with the preamplifier upstream of the channel separation node.

 According to a preferred embodiment, the prosthesis comprises two channels, the filter-separator of one channel being a low-pass filter having a cut-off frequency between 1,800 and 2,500 Hz, the filter-separator of the other channel being a high pass filter having substantially the same cutoff frequency. Experiments have shown that such a prosthesis generally allows good speech intelligibility (the frequency of around 2000 Hz being a pivotal frequency for the vast majority of impairments).

 In addition, according to another characteristic of the invention, the prior compression stage is adapted to have a compression ratio of between 1.5 and 2. This compression ratio represents in practice an optimum rate for improving the signal ratio by noisy atmosphere, while avoiding a significant narrowing of the dynamic range of the significant signal at the output of the prosthesis.

 Furthermore, each channel preferably comprises an attenuator1 located on said channel downstream of the compressor or compressors thereof and providing variable attenuation, adjustable between approximately 0 and 60 dB. It is thus possible to adjust the transfer function of the prosthesis according to the pathology considered.

Other characteristics, objects and advantages of the invention will emerge from the description which follows with reference to the appended drawings which present an embodiment of prosthesis and diagrams illustrating its operation; on these drawings
FIG. 1 is a block diagram of a multi-channel prosthesis according to the invention,
FIG. 2 is a diagram illustrating the compression curves of the three compression sets of the prosthesis,
- Figures 3a, 3b, 3c, 3d, 3e and 3f are diagrams illustrating the operation in a particular case.

 The multi-channel prosthesis illustrated in FIG. 1 comprises a miniaturized microphone 1, in particular of the electret type, a preamplifier 2 adjusting the range of the amplitudes of the electrical signals coming from the microphone to the operating range of the following means, a prior compression stage 3, followed by two processing channels, bass channel 4 and acute channel 5, a summator 6 in which the two channels 4 and 5 meet, a power amplifier 7 with adjustable gain and a headphone 8.

 Each processing channel 4 or 5 comprises a separating filter 9 or 10, with a steep slope between 40 and 52 dB / octave; the filter of the low channel 4 is in the example a low-pass filter with a slope equal to 48 dB / octave as long as a cut-off frequency between 800 Hz and 2,500 Hz, in particular equal to 2,000 Hz; the filter of the treble channel 5 is in the example a high-pass filter with a slope equal to 48 dB / octave having a cut-off frequency substantially equal to that of the previous filter.

Each channel 4 or 5 comprises, downstream of the filters 9 or 10, a dynamic compressor 11 or 12 having a variable compression ratio between 1.5 and 3 for the bass channel, and between 1.5 and 5 for the acute channel. Curves B and
C of FIG. 2, they illustrate the compression rates of the compressors 11 and 12 for predetermined settings. (The curve
D illustrates for comparison a linear amplification).

The compressors 11 and 12 are followed by compensating filters 13 and 14 of the same type and substantially the same cut-off frequency as the separating filters 9 and 10. The slope of each of these compensating filters is chosen to be of the order of P / t where p is the slope of the filter sé
t parator of the channel considered and t a value chosen from the range of compression rates of the compressor of said channel.

The value of p being 48 dB / octave and the value preferentially chosen for t being 2, the slope of the compensating filters may advantageously be chosen equal to 24 dB / octave. This slope gives in practice an optimal compensation for the loss of selectivity on each of the two channels (loss due to compression).

Following these compensating filters, each channel comprises a variable attenuator 15 or 16, adapted to achieve an adjustable attenuation between approximately
O and 60 dB.

 The prior compression stage 3 is formed by one or more compressors in series, located upstream from the channel separation node. The compression ratio of this or these compressors is adjusted between 1.5 and 2, in particular to a value substantially equal to 1.7. Curve A in Figure 2 illustrates this compression ratio.

I1 it should be noted that the curves A, B,
These are compressor curves of a type known per se. One can also use compressors of the same rate, but having improved performance, as described in the patent application filed simultaneously in the name of the applicant.

 The combination of prior compression, high slope filtering and differentiated compression on each channel makes it possible to obtain dynamics of adjustable small width, compatible with those of the vast majority of hearing impaired, as illustrated by the numerical examples. below.

Example 1: Limit example of a serious signal in an acute parasitic noise
This example is based on the realistic hypothesis of superimposed signals for the ear (ie separated 2 by very weak times less than / 10/10 second) but in fact distinct from a few hundredths of a second.

 Pre-compression ratio: 1.7.

 Compression ratio of the low track adjusted to 1.5.

 Compression ratio of the acute channel adjusted to 5.

 The low signal is assumed at a sound level of 50 dB at the entrance to the prosthesis (whispered male voice).

 The high signal is assumed at a sound level of 90 dB at the entrance of the prosthesis (sawmill noise, sheet metal, factory, considered to be very loud).

 The difference of 40 dB between the signals at the input of the pre-compressor 3 is reduced to 23 dB at the input of the two channels, the signals having electrical levels corresponding to 90 and 67 dB respectively.

 Differentiated compression (clearly due to the slopes of filters 4 and 5) reduces this difference to 4 dB at the end of the two channels, since the signals have sufficient levels in each of the two channels to trigger compression.

On output (after summation), the signals have levels equivalent to
- 130 dB for treble,
- and 126 dB for bass.

We thus obtain a difference between the low and high signals, compatible with the very narrow dynamics of the hearing impaired.
For comparison, in the absence of prior compression and assuming all the other parameters adjusted to obtain the same type of dynamic, we would obtain an output
- 130 dB for treble,
- and 113 dB for bass
It can be seen that the low signals are 13 dB lower than those of the prosthesis according to the invention; these signals are barely perceptible, whereas those obtained in the target prosthesis are perfectly heard
Example 2: Acute signal in a serious parasitic noise.

This example is similar to the previous one, with the following settings
Compression ratio of the low track, adjusted to 1.5.

 Compression ratio of the algae channel, adjusted to 3.

 Severe input signal: 90 dB.

 Acute input signal: 50 dB.

(This corresponds to a high voice whispered in an ambient background noise like street, cocktail ...)
The difference of 40 dB between the signals at the pre-compressor input 3 is reduced as before to 23 dB.

 The differentiated compression reduces this difference to 3 d / B at the end of the two channels.

The output signals have levels equivalent to
- 130 dB for bass,
- 127 dB for treble.

In the absence of prior compression, we would obtain an output
- 130 dB for bass,
- and 122 dB for treble (audible limit for an acute signal in a dynamic range of 8 dB)
Example 3: Strictly simultaneous low and high signals.

 This case illustrated in FIGS. 3a to 3f is not a current image of the sound reality but makes it possible to illustrate more clearly the need for high slope separator filters as provided in the invention (signals in solid lines in the figures ) in comparison with a known multi-channel prosthesis having filters of low slope (signals in broken lines).

In FIG. 3a, the low signal at a level of 90 dB and the high signal at a level equal to 60 dB at the input of the two processing channels are illustrated. (In order to show the influence of the slope of the separating filters, it is assumed that these low and high signals are the same at the input, of the two channels of a conventional prosthesis with low separation slope: in particular prosthesis described in the patent
DE Bosch No. 2641675 already mentioned).

 FIGS. 3b and 3c give the appearance of the signals, at the output of the separating filters respectively for the bass channel and for the acute channel. The dotted lines illustrate the slopes of the separating filters: 48 octave in the case of the invention, 12 dB / octave in the known case.

 On the low track (FIG. 3b), the signals are little different in the case of the invention (solid lines) and in the known case (broken lines).

 On the other hand on the acute channel (figure 3c), the known prosthesis lets pass a big part of the serious signal (which, as we will see, will be the source of the bad differentiation of the known prosthesis).

Figures 3d and 3e illustrate the signals respectively at the compressor outlet of the track
this example equal to 5. / grave and at the output of the compressor of the acute channel whose rates are in /
On the acute channel (Figure 3e), compression is triggered in both cases - invention and known case - by the low signal; but this serious signal has a very different level in the two cases so that the amplification gain obtained is much greater for the prosthesis of the invention than that obtained for the known prosthesis.

Thus the high signal is amplified to a level equivalent to approximately 126 dB in the case of the invention, and only to a level equivalent to 105 dB in the known case.

 Figure 3f shows the signals after summation and illustrates the dynamic band of the hearing impaired (D).

In the invention, the low and high signals belong to this dynamic, whereas, in the known case, the high signal is clearly outside of it.

Claims (9)

 1 / - Multichannel hearing aid, comprising a microphone (1) and a preamplifier (2), followed by at least two differentiated processing channels (4,5) each provided with a separating filter (9,10) able to separate a predetermined frequency band, of at least one dynamic compressor (11,12) and a compensating filter (13,14), said channels being combined in a summator (6) followed by a power amplifier (7 ) and an output transducer (8), said hearing aid being characterized in that it comprises a preliminary compression stage (3) located downstream of the preamplifier (2) and upstream of the separating filters (9,10) , in order to ensure before the differentiated processing a prior compression of the signal over the whole frequency range, and in that the filter-separator (9,10) of each channel is adapted to present a slope at least equal to 20 dB /octave.  2 / - Prosthesis according to claim 1, characterized in that the filter-separator (9,10) of each channel is adapted to have a slope of 40 to 52 dB / octave.  3 / - Prosthesis according to one of claims 1 or 2, comprising two channels (4,5), the filter-separator of a channel (4) being a low-pass filter (9) having a cutoff frequency between 1,800 and 2,500 Hz, the separator filter of the other channel (5) being a high-pass filter (10) having substantially the same cutoff frequency.  4 / - Prosthesis according to one of claims 1, 2 or 3, characterized in that the prior compression stage comprises at least one compressor (3) located in series with the preamplifier (2) upstream of the separation node of tracks (4,5).  5 / - Prosthesis according to one of claims 1, 2, 3, or 4, characterized in that the prior compression stage (3) is adapted to have a compression ratio between 1.5 and 2.  6 / - Prosthesis according to one of claims 1, 2, 3, 4 or 5, characterized in that the compensating filter (13,14) of each channel is a filter of the same nature and substantially of the same cutoff frequency as the separator filter (9, -10) of the channel considered, said compensator filter having a slope of the order of P / t where p is the slope of the filter separator of the channel considered and t a value chosen from the rate range of compression of the compressor (s) (11, 12) preceding said filter on said channel.  7 / - Prosthesis according to one of claims 1, 2, 3, 4, 5 or 6, in which each channel (4,5) comprises an attenuator (15,16), located on said channel downstream of the compressor (s) (11,12) and adapted to achieve variable attenuation, adjustable between approximately 0 and 60 dB.  8 / - Prosthesis according to claim 3, characterized in that the compressor (s) (12) of the channel (5) comprising the high-pass filter (10) has a variable compression ratio between 1.5 and 5, and the or the compressors (11) of the other channel (4) has a variable compression ratio between 1.5 and 3.  9 / - Prosthesis according to claims 2, 3, 4, 5, 6 taken together, characterized in that:  the compressor (s) (3) located upstream from the channel separation node (4,5) have a compression ratio of the order of 1.7,  the filter-separator (9,10) for each channel has a slope of the order of 48 dB / octave and a twist cut-off frequency of 2,000 Hz,  the compensating filter (13,14) of each channel has a slope of the order of 24 dB / octave and a cut-off frequency of the order of 2,000 Hz.