WO2017117984A1

WO2017117984A1 - Signal processing method and system for enhancing temporal presentation in cochlear implant

Info

Publication number: WO2017117984A1
Application number: PCT/CN2016/092067
Authority: WO
Inventors: 孟庆林; 郑能恒; 李霞
Original assignee: 深圳大学
Priority date: 2016-01-07
Filing date: 2016-07-28
Publication date: 2017-07-13
Also published as: CN105662706A; CN105662706B

Abstract

A signal processing method for enhancing temporal presentation in a cochlear implant. The method comprises: a filtering step, in which a band-pass filter bank filters an acquired audio signal; a step of transforming temporal presentation of a signal, in which, if a frequency range of a current electrode channel is higher than the upper limit f1 of temporal pitch of the electrode channel, and a bandwidth is less than or equal to a difference between the upper limit f1 of temporal pitch and an upper limit f2 of a frequency range of temporal periodicity of an envelope, then the frequency of the band-pass filtered signal passing through the current electrode channel is shifted downward to between the upper limit f2 of the frequency range of temporal periodicity of the envelope and the upper limit f1 of temporal pitch of the electrode channel; and a band-pass filtered signal transferring step, in which the band-pass filtered signal having undergone the transformation of temporal presentation is further processed, and then outputted as an electrical current waveform. The method introduces temporal fine structure information without interfering with an envelope, thus providing cochlear implants with improved auditory perception performance.

Description

Method and system for processing cochlear signal with enhanced time domain expression

Technical field

The invention belongs to the field of medical instruments, and in particular relates to a method and system for processing a cochlear signal which enhances time domain expression.

Background technique

Cochlear Implant (CI) It is an electronic device that can help people with severe hearing loss recover part of their hearing. Most CI The implanter can already perform good one-to-one verbal communication (without visual information) in a quiet environment, or even make a phone call. But this is based on the speaker's articulation, standard, and moderate speed. Fast speech, continuous pronunciation, and occasional voicing, these conditions may be no problem for normal hearing, but will bring great difficulties to the implanter. In addition, The anti-noise, music perception, and pitch resolution of CI implanters are still poor.

Current CI Most of the signal processing strategies are based on the time domain envelope design, in which only a small number of limited time (12-24) channels of time domain envelope information are retained, and the time domain fine structure is discarded (temporal Fine Structure, TFS ) Information. The lack of TFS has led to the CI shortcomings mentioned above. There have been a lot of studies trying to add TFS to CI signal processing strategies, but in practice In the CI experiment, none of these strategies have demonstrated the ability to significantly improve the actual CI effect. Limitations include electro-acoustic time-domain pitch limits, mutual interference between envelopes and TFS, envelopes and TFS Decompose the morbidity of the problem.

In order to break through the above three limitations, the technology proposes a new signal processing strategy, named as time domain limited coding (temporal) Limits encoder, TLE) strategy. The core innovation of the TLE strategy is to shift the bandpass signal down to the upper limit of the electro-audio time domain pitch, which overcomes the electro-acoustic time-domain pitch limit, while this frequency shifting operation avoids explicit envelopes and The TFS decomposition operation avoids the nuisance caused by it.

However, one problem left out by TLE is the mutual interference between the envelope and the TFS. Since TFS was moved to 300Hz Below, this range overlaps with the frequency range (50 Hz - 300 Hz) in which the time domain periodicity in the envelope lies. Envelope and TFS at 300Hz simultaneously The following expressions in the time domain form inevitably cause mutual interference.

technical problem

The technical problem to be solved by the present invention is to provide a method and system for processing a cochlear signal with enhanced time domain expression, aiming at introducing time domain fine structure information while avoiding interference to an envelope, thereby improving the sound perception effect of the cochlear implant.

Technical solution

The invention provides a method for processing a cochlear signal with enhanced time domain expression, comprising the following steps:

Filtering step: filtering the collected sound signal by using a preset band pass filter bank to obtain a band pass signal;

Time domain signal expression conversion step: if the current electrode channel frequency range is greater than the electrode channel time domain pitch upper limit f1 and the bandwidth is less than or equal to the electrode channel time domain pitch upper limit f1 and the envelope time domain periodicity upper limit of the frequency range When the difference of f2 is, the band pass signal passing through the current electrode channel is frequency-shifted down to an upper limit f2 of the frequency range in which the time domain of the envelope is located and an upper limit f1 of the time domain pitch of the electrode channel;

The band pass signal transmission step: further processing the band pass signal after the time domain signal expression conversion processing and expressing it on the current waveform and transmitting it.

Further, the upper limit of the time domain pitch f1 of the electrode channel is 450 Hz. ~850Hz, the upper limit of the frequency range in which the time domain of the envelope is located is in the range of 250 Hz to 450 Hz, and f1 - f2 ≥ 100 Hz.

Further, the filtering step further includes a pre-processing step, specifically performing pre-emphasis and noise reduction processing on the collected sound signal.

Further, the method for further processing the band pass signal in the band pass signal transmission step is: performing nonlinear compression on the band pass signal.

Further, the current waveform in the band pass signal transmission step is a single pulse interval sampling waveform or an analog current waveform;

Specifically, if a single pulse interval sampling waveform is used, the band pass signal is sampled by a single pulse, and then the stimulation is generated at intervals between different channels;

If an analog current waveform is used, the bandpass signal is expressed as an analog current.

The invention also provides a cochlear signal processing system for enhancing time domain expression, comprising:

a filtering module, configured to filter the collected sound signal by using a preset band pass filter bank;

a time domain signal expression conversion module, configured to perform a time domain signal expression conversion process on the band pass signal output by each band pass filter; specifically, if the current electrode channel has a frequency range greater than an upper limit of the time domain pitch of the electrode channel, f1 When the bandwidth is less than or equal to the difference between the upper limit f1 of the time domain pitch of the electrode channel and the upper limit f2 of the frequency range in which the time domain of the envelope is located, the band pass signal passing through the current electrode channel is down-shifted to the time domain of the envelope. The upper limit of the frequency range where the periodicity is f2 and the upper limit of the time domain pitch of the electrode channel f1;

The band pass signal transmission module is configured to further process the band pass signal after the time domain signal expression conversion processing and express the signal to the current waveform for transmission.

Further, a pre-processing module is further included for performing pre-emphasis and noise reduction processing on the collected sound signal.

Further, the band pass signal transmission module is specifically configured to perform non-linear compression on the band pass signal subjected to the time domain signal expression conversion processing and express the signal to the current waveform.

Further, the current waveform in the band pass signal transmission module is a single pulse interval sampling waveform or an analog current waveform;

Beneficial effect

Compared with the prior art, the present invention has the beneficial effects that the present invention provides a method and system for processing a cochlear signal that enhances time domain expression, and shifts the band pass signal to a time domain periodicity slightly higher than the envelope. The upper limit of the frequency range, which avoids the ill-posed problem of envelope and TFS (time domain fine structure) decomposition, and transmits TFS at a moderately usable speed, which can avoid excessive interference of TFS on the envelope; Single pulse interval sampling or analog current stimulation provides higher auditory sensitivity than traditional multiphase pulse interval sampling stimulation, and minimizes residual charge that may damage the auditory nerve.

DRAWINGS

1 is a flowchart of a method for processing a cochlear signal with enhanced time domain expression according to an embodiment of the present invention;

2 is a schematic diagram of a frequency shifting operation used in time domain signal expression conversion according to an embodiment of the present invention;

3 is a schematic diagram of a cochlear signal processing system for enhancing time domain expression according to an embodiment of the present invention;

FIG. 4 is a schematic diagram showing comparison of CIS, TLE1, and TLE2 in time domain signal expression by using a composite sound composed of a double pure tone as an example according to an embodiment of the present invention; FIG.

Figure 5 is a schematic view showing the macro comparison of the electrodes of CIS, TLE1, and TLE2 of Figure 4;

Fig. 6 is a partially enlarged view showing the electrode patterns of CIS, TLE1, and TLE2 in Fig. 5.

Embodiments of the invention

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

A method for processing a cochlear signal with enhanced time domain expression, specifically comprising: a pre-processing step of pre-emphasizing and noise-reducing a sound signal collected by a microphone; and a filtering step: using a pre-designed band-pass filter bank pair The obtained sound signal is filtered; the time domain signal expression conversion step: performing time domain signal expression conversion processing on the band pass signal output by each band pass filter, so that the time domain of the envelope is preserved while introducing the time domain fine structure information Periodic information; bandpass signal transmission step: nonlinearly compressing the converted bandpass signal in each channel and presenting it on the current waveform in a certain waveform form. The overall flow chart is shown in Figure 1.

The method for transforming the time domain signal expression in the time domain signal expression conversion step includes at least one of the following three methods:

1. If the frequency range of the current electrode channel is smaller than the upper limit f1 of the time domain pitch of the electrode channel, the band pass signal is directly processed without any processing.

2. If the frequency range of the current electrode channel is greater than the upper limit f1 of the time domain pitch of the electrode channel and the bandwidth is less than or equal to f1-f2, the bandpass signal that passes through is frequency-shifted down to the frequency range in which the time domain of the envelope is periodic. The upper limit f2 is between the upper limit of the time domain pitch of the electrode channel f1;

The time-domain pitch upper limit f1 of the electrode channel ranges from 450 Hz ~850Hz; the upper limit of the frequency range in which the time domain of the envelope is located is in the range of 250 Hz to 450 Hz; and f1 - f2 ≥ 100 Hz.

3. If the frequency range of the current electrode channel is greater than the upper limit of the time domain pitch of the electrode channel f1 and the bandwidth is greater than or equal to f1- For f2, one of the following two methods is selected to process the passed bandpass signal:

In the first method, the band pass signal passing through the current electrode channel is frequency-shifted to a range above the upper limit f2 of the frequency range in which the time domain of the envelope is located, and the band-pass signal after the frequency shift is extracted.

In the second method, the envelope signal extracted through the current electrode channel is directly extracted.

The frequency shifting operation used in the time domain signal expression conversion is as shown in FIG. 2.

In the above-described band pass signal transmission step, the band pass signal is subjected to nonlinear compression processing; specifically, the output result of the time domain expression conversion is acoustically mapped. The traditional compression function is a power function and a logarithmic function; the present invention is different from the principle of a conventional envelope (short time energy) based strategy, which uses a smaller compression factor to better express TFS information.

In the band pass signal transmission step, If the time domain signal expression conversion process adopts the first or second method, the current waveform in the band pass signal transmission step corresponds to a single pulse interval sampling waveform or an analog current waveform; if the time domain signal expression conversion process adopts the third method, Then, the current waveform in the band pass signal transmission step corresponds to the multiphase pulse interval stimulation waveform.

Specifically, if a single pulse interval is used to sample the waveform, the band pass signal is sampled with a single pulse (only one positive rectangular pulse or one negative rectangular pulse), and then the stimulus is generated at intervals between different channels.

If an analog current waveform is used, the bandpass signal is directly expressed as an analog current.

If the traditional rectification plus multi-phase pulse interval stimulation waveform is used, a higher stimulation rate should be used when using pulse expression (recommended above 2000 pps); here pps is pulses per The abbreviation of second, translated as the number of pulses per second. In terms of current quantization accuracy, higher (>8 bits) quantization accuracy is recommended to better express TFS information.

The so-called conventional multiphase (most commonly used biphasic) pulse interval stimulation waveform has multiple phases in one pulse; in the case of a biphasic pulse, it is a positive and negative wave, that is, a positive rectangular pulse and a negative rectangle. Pulses occur sequentially (can also be negative before and after). The absolute charge of two different phase pulses is equal. Because they have opposite signs, they can cancel each other out and avoid residual charge. (Note: residual charge may damage the auditory nerve) .

The invention also provides a cochlear signal processing system for enhancing time domain expression, as shown in FIG. 3, comprising:

Filter module 1: for filtering the collected sound signal by using a preset band pass filter bank;

Time domain signal expression conversion module 2: for performing time domain signal expression conversion processing on the band pass signal output by each band pass filter;

The band pass signal transmission module 3 is configured to further process the band pass signal subjected to the time domain signal expression conversion processing and express the signal to the current waveform for transmission.

The time domain signal expression conversion module 2 includes at least one of the following three submodules:

The time domain signal expression module 21 directly outputs a band pass signal if the frequency range of the current electrode channel is smaller than the time domain pitch upper limit f1 of the electrode channel;

The frequency shifting module 22, if the frequency range of the current electrode channel is greater than the upper limit of the time domain pitch of the electrode channel f1 and the bandwidth is less than or equal to f1- At f2, the frequency shifting module 22 shifts the passed band pass signal down to an upper limit f2 of the frequency range in which the time domain periodicity of the envelope is located and an upper limit f1 of the time domain pitch of the electrode channel;

The envelope extraction module 23, if the frequency range of the current electrode channel is greater than the upper limit f1 of the time domain pitch of the electrode channel and the bandwidth is greater than or equal to f1- For f2, one of the following two sub-modules is selected to process the passed bandpass signal:

The first envelope extraction module is configured to frequency-shift the band-pass signal passing through the current electrode channel to a range above the upper limit of the frequency range f2 in which the time domain of the envelope is located, and perform the band-pass signal after the frequency shift Envelope extraction

The second envelope extraction module is configured to perform envelope extraction directly on the band pass signal passing through the current electrode channel.

The following is a comparison of the time domain signal representations of CIS, TLE1, and TLE2 in a specific embodiment:

At present, most of the commonly used strategies in the clinic only retain the time domain envelope information and discard the time domain fine structure (TFS) information, the most typical of which is continuous interval sampling (continuous Interleaved sampling, CIS) strategy. Another comparison method is the time domain limited encoder proposed by the prior art. The encoder) strategy is represented here by TLE1. An important difference between the present invention and TLE1 is that the selected frequency shift range is different, and TLE1 shifts the band-limited signal to below 300 Hz. The present invention shifts the band-limited signal to more than 300 Hz. Here, the method proposed by the present invention is represented by TLE2. .

Figure 4 compares the time domain signal representations of CIS, TLE1, and TLE2 with a composite sound composed of double pure tones. The left column is the time domain signal representation, the gray solid line represents the original or frequency shifted band limit signal, and the black dotted line represents the corresponding Hilbert envelope. The time domain expression signal of each channel of TLE1 is the left middle diagram solid line, the time domain expression signal in each channel of TLE2 is the left lower diagram solid line, and the time domain expression signal in each channel of CIS is the upper left diagram dashed line ( That is, the Hilbert envelope). The column on the right is the amplitude spectrum corresponding to the left solid line.

It can be seen that the amplitude spectra of the TLE1 and TLE2 signals in Fig. 4 all retain the shape of the original amplitude spectrum, with the only difference being the frequency range. The Hilbert envelope of the time domain signals extracted by TLE1 and TLE2 retains the envelope information of the original band-limited signal. The time domain signal extracted by the TLE method is slower than the original band limit signal, and the time domain signal extracted by TLE2 is slightly faster than the time domain signal extracted by TLE1. Since the TFS is moved below 300 Hz in TLE1, this range overlaps with the range to which the envelope belongs, and mutual interference is easily generated. When TLE2 is moved above 300Hz, this interference can be avoided to some extent. At the same time, according to the existing research, 300Hz is not the absolute upper limit of the sound level of the electro-acoustic time domain. Higher time domain rate of change, such as 300 to 1000 Hz, is also perceived as possible. This ensures that the TLE2 method retains the advantages of the TLE1 method while avoiding envelope and TFS interference.

Figure 5 is a macroscopic comparison of the electrode patterns of CIS, TLE1, and TLE2. The original voice is a mandarin saying "how can his phone not work?"; Figure 6 shows the signals in the three small long boxes in Figure 5. A map obtained by partial enlargement. It can be seen from Fig. 5 that the three methods do not have significant differences in the overall energy distribution. However, as can be seen from Fig. 6, there is a significant difference in local features; TLE2 retains the envelope information in CIS better than TLE1, while TLE2 increases the available TFS information than CIS.

The present invention shifts the bandpass signal to a range slightly above the upper limit of the electrical auditory pitch (about 300 Hz), and can pass the TFS at a moderately usable speed while avoiding the ill-conditioned problems of envelope and TFS decomposition. To avoid excessive interference of TFS on the envelope. In addition, in terms of the form of current stimulation, the present invention proposes to use a single pulse interval sampling or analog current form because the time domain signal extracted by TLE2 does not have a direct current component and does not generate harmful charge accumulation. These advantages make the present invention (TLE2) enhance the time domain expression of the cochlear implant, which in turn may improve the acoustic perception of the cochlear implant.

The invention is applicable to a cochlear implant, and may also be applied to other artificial hearing devices, including auditory brainstem implantation, auditory midbrain implantation, deep brain implantation and the like.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

A method for processing a cochlear signal that enhances time domain expression, comprising the steps of:

Filtering step: filtering the collected sound signal by using a preset band pass filter bank to obtain a band pass signal;

Time domain signal expression conversion step: if the current electrode channel frequency range is greater than the electrode channel time domain pitch upper limit f1 and the bandwidth is less than or equal to the electrode channel time domain pitch upper limit f1 and the envelope time domain periodicity upper limit of the frequency range When the difference of f2 is, the band pass signal passing through the current electrode channel is frequency-shifted down to an upper limit f2 of the frequency range in which the time domain of the envelope is located and an upper limit f1 of the time domain pitch of the electrode channel;

The band pass signal transmission step: further processing the band pass signal after the time domain signal expression conversion processing and expressing it on the current waveform and transmitting it.
The cochlear signal processing method according to claim 1, wherein the upper limit of the time domain pitch f1 of the electrode channel is 450 Hz. ~850Hz, the upper limit of the frequency range in which the time domain of the envelope is located is in the range of 250 Hz to 450 Hz, and f1 - f2 ≥ 100 Hz.
The method for processing a cochlear signal according to claim 1, wherein the filtering step further comprises a pre-processing step, in particular, pre-emphasizing and noise-reducing the collected sound signal.
The cochlear signal processing method according to claim 1, wherein the band pass signal is further processed in the band pass signal transmitting step by nonlinearly compressing the band pass signal.
The cochlear signal processing method according to claim 1, wherein the current waveform in the band pass signal transmitting step is a single pulse interval sampling waveform or an analog current waveform;

Specifically, if a single pulse interval sampling waveform is used, the band pass signal is sampled by a single pulse, and then the stimulation is generated at intervals between different channels;

If an analog current waveform is used, the bandpass signal is expressed as an analog current.
A cochlear signal processing system for enhancing time domain expression, comprising:

a filtering module, configured to filter the collected sound signal by using a preset band pass filter bank;

a time domain signal expression conversion module, configured to perform a time domain signal expression conversion process on the band pass signal output by each band pass filter; specifically, if the current electrode channel has a frequency range greater than an upper limit of the time domain pitch of the electrode channel, f1 When the bandwidth is less than or equal to the difference between the upper limit f1 of the time domain pitch of the electrode channel and the upper limit f2 of the frequency range in which the time domain of the envelope is located, the band pass signal passing through the current electrode channel is down-shifted to the time domain of the envelope. The upper limit of the frequency range where the periodicity is f2 and the upper limit of the time domain pitch of the electrode channel f1;

The band pass signal transmission module is configured to further process the band pass signal after the time domain signal expression conversion processing and express the signal to the current waveform for transmission.
The cochlear signal processing system according to claim 6, wherein the upper limit of the time domain pitch f1 of the electrode channel is 450 Hz ~850Hz, the upper limit of the frequency range in which the time domain of the envelope is located is in the range of 250 Hz to 450 Hz, and f1 - f2 ≥ 100 Hz.
The cochlear signal processing system of claim 6 further comprising a pre-processing module for pre-emphasizing and noise-reducing the collected sound signals.
The cochlear signal processing system according to claim 6, wherein the band pass signal transmission module is specifically configured to nonlinearly compress the band pass signal after the time domain signal expression conversion processing and express the current waveform on the current waveform. Pass it out.
The cochlear signal processing system according to claim 6, wherein the current waveform in the band pass signal transmission module is a single pulse interval sampling waveform or an analog current waveform;

Specifically, if a single pulse interval sampling waveform is used, the band pass signal is sampled by a single pulse, and then the stimulation is generated at intervals between different channels;

If an analog current waveform is used, the bandpass signal is expressed as an analog current.