CN103393484A - Voice processing method used for electrical cochlea - Google Patents
Voice processing method used for electrical cochlea Download PDFInfo
- Publication number
- CN103393484A CN103393484A CN2013103288270A CN201310328827A CN103393484A CN 103393484 A CN103393484 A CN 103393484A CN 2013103288270 A CN2013103288270 A CN 2013103288270A CN 201310328827 A CN201310328827 A CN 201310328827A CN 103393484 A CN103393484 A CN 103393484A
- Authority
- CN
- China
- Prior art keywords
- signal
- hilbert
- passage
- huang transform
- frequency
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Prostheses (AREA)
Abstract
The invention discloses a voice processing method used for an electrical cochlea. The voice processing method used for the electrical cochlea comprises the steps of pre-processing, namely carrying out pre-processing on a voice signal input through a microphone; Hilbert-Huang transformation, namely carrying out Hilbert-Huang transformation on the pre-processed voice signal to obtain Hilbert-Huang transformation signals of multiple channels; modulation, namely processing the Hilbert-Huang transformation signal of each channel to obtain an amplitude signal and a frequency signal which correspond to the Hilbert-Huang transformation signal of the channel, modulating a biphase pulse by means of the frequency signal corresponding to the Hilbert-Huang transformation signal of each channel to obtain a biphase pulse modulation frequency signal corresponding to the Hilbert-Huang transformation signal of the channel, and modulating the biphase pulse modulation frequency signal corresponding to the Hilbert-Huang transformation signal of each channel by means of the amplitude signal corresponding to the Hilbert-Huang transformation signal of the channel to obtain a channel modulation signal used for driving an electrode corresponding to the channel.
Description
Technical field
The present invention relates to a kind of cochlear implant, relate in particular to the method for speech processing for cochlear implant.
Background technology
Cochlear implant is a kind of medical apparatus that utilizes functional electric stimulation to recover severe or total deafness patient part audition, and it is unique effective way for the treatment of at present phonosensitive nerve deafness.Up to now, the whole world has at least had 100000 people to carry out the implant surgery of cochlear implant.The progress of development, the especially microelectronic of science and technology, make cochlear implant experience from one pole to multipole, from analog to digital, and the development from simple feature extraction to complicated time domain treatment of details.Along with constantly bringing forth new ideas of signal processing technology and reaching its maturity of voice coding strategy study, the cochlear implant system not only can provide the very phonetic recognization rate of high level, and can make implantation person quietly carry out communication on telephone under environment.Yet, the problems such as identification for the speech recognition under implantation person's noise circumstance of present existence, intonation speech recognition, " cocktail party effect " (cocktail partyeffect) and music but never are well solved, its main cause is that existing speech processes scheme is not still very perfect, can not well simulate the function of audition peripheral-system.
The voice coding strategy is the core technology of cochlear implant, and according to the different information and the hearing mechanism that stress exciting electrode, people have developed a lot of voice coding strategies, and studying carefully that its characteristics can be divided three classes is modeling scheme, amplitude modulation schemes and amplitude-frequency modulation scheme.
To be mainly the voice signal that will gather directly be sent to the electrode stimulating acoustic nerve to modeling scheme after simple bandpass filtering and suitable dynamic range compression, and this scheme mainly comprises CA(Compressed Analog).Amplitude modulation(PAM) is similar with the principle of amplitude-frequency modulation scheme, is all voice signal to be passed through a plurality of band filters after pretreatment, carries out a series of processing, finally the output waveform of each band filter is sent to the electrode stimulating acoustic nerve.What difference was that the AM scheme mainly transmits is the amplitude information of raw tone, utilizes the amplitude information that obtains to modulate the diphasic pulse of fixed frequency, and then is sent to acoustic nerve.This strategy mainly comprises: simulation simultaneously stimulates SAS(Simultaneous Analog Stimulation), maximum chromatograph process device SMSP(Spectral Maxima Sound Processor), spectrum peak method SPEAK(Spectral peak), consecutive intervals sampling CIS(Continuous Interleaved Sampling) etc.The amplitude-frequency coding strategy has also extracted its frequency (phase place or fine structure) information when extracting the primary speech signal amplitude information, utilize amplitude and frequency (phase place or the fine structure) combined coding of raw tone, reach the purpose of improving speech recognition, these class methods mainly comprise amplitude-frequency combined coding (Frequency Amplitude Modulation Encoding, FAME), the small echo zero passage stimulates (Wavelet Transform Zero-crossing Stimulation, WTZS) etc.
Research shows, the identification of voice under noise circumstance, and the frequency, phase information and the time domain fine structure that comprise in the identification of competition voice and music and voice signal are closely related.And the existing voice coding strategy some can only transmit the amplitude information (SPEAK, CIS etc.) of voice; Although some coding strategies is coded amplitude and frequency information but information after processing is too coarse makes acoustic nerve can't optionally obtain useful information (as CA) to cause the speech recognition poor effect simultaneously; The coding strategy such as FAME, WTZS no doubt can be sent to acoustic nerve with the amplitude of raw tone and frequency simultaneously, and through experimental verification, really can improve speech recognition capabilities, but its principle design defect and adaptivity fail to be well solved.
Summary of the invention
Problem for the existing voice coding strategy exists, the present invention proposes a kind of method of speech processing for cochlear implant, possesses good adaptivity.
Method of speech processing for cochlear implant of the present invention, it comprises the following steps: pre-treatment step, the voice signal through the mike input is carried out pretreatment; The Hilbert-Huang transform step, carry out Hilbert-Huang transform to voice signal after pretreatment, obtains the Hilbert-Huang transform signal of a plurality of passages; Modulation step,, for each passage, process the Hilbert-Huang transform signal of this passage, obtains the range signal corresponding with the Hilbert-Huang transform signal of this passage and frequency signal; With the frequency signal that the Hilbert-Huang transform signal with this passage is corresponding, diphasic pulse is modulated and obtained the diphasic pulse modulation frequency signal corresponding with the Hilbert-Huang transform signal of this passage; Modulate with the diphasic pulse modulation frequency signal that range signal corresponding to the Hilbert-Huang transform signal with this passage pair and the Hilbert-Huang transform signal of this passage are corresponding, obtain for the channel modulation signal that drives the electrode corresponding with this passage.
Preferably, described pre-treatment step comprises preemphasis and minute frame processing.
Preferably, described pre-treatment step comprises preemphasis, end-point detection and minute frame processing.
Preferably, described diphasic pulse is the diphasic pulse of this channel central frequency.
Preferably, the described diphasic pulse pulse corresponding with the mistake zero position of intrinsic mode function corresponding to this passage Hilbert-Huang transform signal that be pulse position.
Preferably, in modulation step, the Hilbert-Huang transform signal of each passage is divided into first via signal and the second road signal; Obtain the range signal corresponding with the Hilbert-Huang transform signal of this passage by first via signal; Obtain the frequency signal corresponding with the Hilbert-Huang transform signal of this passage by the second road signal.
Preferably, according to the instantaneous amplitude formula, first via signal is carried out range signal and extract, the range signal that then will extract carries out low-pass filtering, thereby obtains the range signal corresponding with the Hilbert-Huang transform signal of this passage.
Preferably, according to the instantaneous frequency formula, the second road signal is carried out frequency signal and extract, thereby obtain the frequency signal corresponding with the Hilbert-Huang transform signal of this passage.
Preferably, before with the diphasic pulse pair frequency signal corresponding with the Hilbert-Huang transform signal of this passage, modulating, according to people's ear, the scope of frequency perception is carried out limit bandwidth to the frequency signal that extracts and process.
Preferably, after with the diphasic pulse pair frequency signal corresponding with the Hilbert-Huang transform signal of this passage, modulating, further carry out low-pass filtering, obtain the diphasic pulse modulation frequency signal corresponding with the Hilbert-Huang transform signal of this passage.
method of speech processing for cochlear implant of the present invention, compare and do not need to arrange band-pass filter group or choose wavelet basis function with conditional electronic cochlea voice coding strategy, it is according to the own characteristic of voice signal, obtain the local narrow band signal IMF of different scale feature by screening layer by layer, use on this basis HT, accurately extracting instantaneous amplitude and instantaneous frequency parameter encodes, whole process is not used any priori, broken away from the restriction of conventional filter group and wavelet basis window function and uncertainty principle, possesses good adaptivity, the fundamental characteristics that has kept primary speech signal, its Performance Ratio wants superior many based on the voice coding strategy of bandpass filtering and wavelet analysis.
Description of drawings
Fig. 1 is the schematic diagram of the method for speech processing for cochlear implant of the present invention;
Fig. 2 is an embodiment of the method for speech processing for cochlear implant of the present invention;
Fig. 3 is the simulating, verifying conceptual scheme of the method for speech processing that is used for cochlear implant of Fig. 2;
Fig. 4 a is the primary speech signal figure of the method for speech processing that is used for cochlear implant of Fig. 2;
Fig. 4 b is eight passage IMF of the method for speech processing that is used for cochlear implant of Fig. 2;
Fig. 4 c is the instantaneous amplitude figure corresponding to eight passage IMF of the method for speech processing that is used for cochlear implant of Fig. 2;
Fig. 4 d is eight instantaneous frequency figure corresponding to passage IMF of the method for speech processing that is used for cochlear implant of Fig. 2;
Fig. 4 e is primary speech signal and the synthetic speech signal comparison diagram of the method for speech processing that is used for cochlear implant of Fig. 2.
The specific embodiment
Below, the present invention is described in detail by reference to the accompanying drawings.
A Hilbert transform (Hilbert Transform, HT)
For real signal x (t)=A (t) cos φ t arbitrarily, its analytical form s (t) can be expressed as:
Wherein
Form Hilbert transform pairs with x (t), s (t) can be expressed as again so:
s(t)＝A(t)e
jφt （2）
Wherein
Like this, the Hilbert conversion provides the definition instantaneous amplitude of a uniqueness and the function of phase place, and the instantaneous frequency f (t) of primary signal x (t) is:
For non-linear, non-stationary signal, although HT be convenient, effectively, instrument efficiently, it can not be applied separately.For the output that makes HT has accurate physical significance, the input that requires HT must be narrow band signal, A (t) like this, and φ (t) and f (t) could represent the time-varying characteristics of primary signal x (t) from different perspectives accurately.That is to say, before multicomponent signal HT to, must process by narrow band filter.
B Hilbert-Huang transform (Hilbert-Huang Transform, HHT)
HHT is Huang has proposed a kind of new non-linear, non-stationary signal in 1998 Time-Frequency Analysis Method.Need carry out two basic steps while applying this method: at first, with empirical mode decomposition (Empirical Mode Decomposition, EMD) method becomes a series of to signal decomposition and satisfies condition (a) and intrinsic mode function (b) (Intrinsic Mode Function, IMF).Then, the IMF component that decomposition is obtained carries out the Hilbert conversion, thereby draws the Energy distribution on time-frequency plane.
(a) in whole data sequence, the quantity of extreme point must equate with the quantity of zero crossing, or differ at most can not be more than one.
(b) on a time point in office.The envelope meansigma methods of the local maximum of signal and local minimum definition is zero.
First qualifications is very obvious; It is similar to the definition of traditional stationary Gaussian process about arrowband.Second condition is that traditional overall situation is limited and becomes local the restriction.This restriction is necessary, and it can remove the fluctuation of the instantaneous frequency that causes because waveform is asymmetric.
EMD is commonly called Empirical mode decomposition, is the Algorithm of Signal Decomposition that Chinese American NE.Huang proposed in 1996, and this is mainly the process of IMF of isolating in sophisticated signal, also referred to as screening process (The Sifting Process).Its catabolic process is as follows:
(1) at first find out all maximum points of x (t) and it is become the coenvelope line e of former data sequence with Cubic Spline Functions Fitting
max(t): and all minimum points and it is become the lower envelope line e of former data sequence with Cubic Spline Functions Fitting
min(t);
(2) calculate the average m of up and down envelope
11(t)=(e
max(t)+e
min(t))/2, deduct this average with primary signal x (t) and can obtain a new data sequence h who removes low frequency
11(t)=x (t)-m
11(t);
If h
11(t) also do not meet two primary conditions of IMF, it is repeated screening process step (1) and (2) k time as echo signal, until meet end condition 0.2≤SD≤0.3,
At this moment, h
1k(t)=h
1 (k-1)(t)-m
1k(t), m wherein
1k(t) be the envelope average of k iteration, h
1 (k-1)(t) be signal and (k-1) difference of inferior envelope average.
(3) definition c
1(t)=h
1k(t) be first IMF, it separated r from primary signal
1(t)=x (t)-c
1(t);
(4) with r
1(t) carry out above screening process as new initialize signal and extract the IMFs of x (t);
(5) work as r
nWhile (t) meeting end condition, prompting can not be isolated IMF again from primary signal x (t), and final x (t) can be expressed as the combining form of IMFs and residual error:
According to above analysis as can be known, these IMF meet the requirement of narrow band signal in theory, can be directly used in instantaneous amplitude and instantaneous frequency that HT obtains primary signal, and these instantaneous parameterses has clear and definite physical significance, simultaneously the time-varying characteristics of energy accurate expression primary signal.The EMD decomposition is the adaptive decomposition that the information of basis signal itself is carried out, and namely its catabolic process depends on code book and economizes the change information that comprises, the variation of the responsive reflected signal of energy.Its screening thought has embodied the filtering of multiresolution analysis simultaneously, and each IMF component has certain physical significance usually, and comprises the characteristic dimension of certain limit, therefore can utilize this feature to carry out filtering to echo signal.
C voice coding strategy of the present invention
Theoretically, cochlea can be considered to consist of the band filter of one group of spatial distribution, and the quality factor q of the wave filter ratio of bandwidth (mid frequency with) is approximately constant.According to this characteristic, band-pass filter group and wavelet function are widely used in existing cochlear implant voice coding policy-simulative people's peripheral auditory system.From the angle of signal processing, no matter be bandpass filtering or wavelet transformation, its essence is all Fourier transformation, inevitably is subjected to the restriction of window function and uncertainty principle, thus the time-frequency characteristic of expression signal accurately.In addition, band filter and wavelet basis parameter are once determining just no longer change when processing signals, and adaptivity is very poor.The adaptive multiresolution analysis characteristic of HHT mentioned above is very similar to the cataloged procedure of voice to cochlea, and therefore, it is feasible utilizing the multiresolution analysis function of this simulated behavior cochlea of HHT to encode to voice.The present invention proposes a kind of new cochlear implant voice coding strategy (Fig. 1) according to above analysis, to improve the identification of speech recognition, intonation speech recognition, " cocktail party effect " (cocktail party effect) and music under noise circumstance.
Fig. 1 is the schematic diagram of the method for speech processing for cochlear implant of the present invention.Fig. 2 is an embodiment of the method for speech processing for cochlear implant of the present invention.Voice signal carries out pretreatment after Mike's elegance enters system, pretreatment comprises preemphasis, end-point detection and minute frame etc.In actual process, preemphasis is mainly in order to promote the HFS of voice, makes the frequency spectrum of signal become smooth, so that carry out spectrum analysis or channel parameters analysis (by the preemphasis digital filter, realizing).Carry out end-point detection for intonation voice especially Chinese etc. and can realize the efficient processing of voice, non-stationary and time variation due to voice signal, usually its segmentation or minute frame are realized approximate local or calm disposing in short-term, the frame number of general per second is about 33-100, is determined on a case-by-case basis.After pretreatment, voice signal decomposes and obtains the intrinsic mode function IMF that frequency is arranged in order from low to high through EMD
n, IMF
n-1IMF
2, IMF
1, thereafter these intrinsic mode functions are carried out Hilbert transform.After conversion, signal parallel is divided into two-way, and the instantaneous amplitude formula of leading up to accurately extracts amplitude corresponding to each passage IMF and carries out low-pass filtering treatment and obtains amplitude modulation information; Frequency corresponding to each passage IMF accurately extracted by the instantaneous frequency formula in another road, according to people's ear, the scope of frequency perception being carried out limit bandwidth to each passage instantaneous frequency processes, after processing, signal is used for frequency modulation(PFM) (not changing the amplitude of bidirectional pulse) is carried out in the diphasic pulse of respective channel mid frequency, through low-pass filtering treatment, obtains frequency or phase-modulated information.With the instantaneous amplitude of each passage, corresponding frequency or phase information are carried out amplitude modulation(PAM) and then are sent to the electrode E that implants cochlea stimulating acoustic nerve finally.
Fig. 3 is the simulating, verifying conceptual scheme of the method for speech processing that is used for cochlear implant of Fig. 2, be with Fig. 2 difference, instantaneous frequency is carried out frequency modulation(PFM) to the sinusoidal signal of mid frequency, and finally the warbled information of each channel amplitude adds and rear synthetic speech signal output.
Fig. 4 a-4e is based on an instantiation of the cochlear implant voice coding strategy voice simulation of Hilbert-Huang transform.Primary speech signal Fig. 4 a obtains intrinsic mode function Fig. 4 b of eight passages after EMD, carry out after Hilbert transform and corresponding signal processing obtaining instantaneous amplitude hum pattern 4c and instantaneous frequency or phase information Fig. 4 d of respective channel, through amplitude modulation(PAM) and finally obtain synthetic speech signal output map 4e.
For the nonlinear and nonstationary voice signal, its signal frequency is time dependent, will analyze exactly the frequency change rule of this type of signal, single yardstick single resolution analysis method in other words is no longer applicable, and it is indispensable to possess the signal processing technology of ability of multiresolution analysis.the new cochlear implant voice coding strategy based on HHT that the present invention proposes, compare and do not need to arrange band-pass filter group or choose wavelet basis function with conditional electronic cochlea voice coding strategy, it is according to the own characteristic of voice signal, obtain the local narrow band signal IMF of different scale feature by screening layer by layer, use on this basis HT, accurately extracting instantaneous amplitude and instantaneous frequency parameter encodes, whole process is not used any priori, broken away from the restriction of conventional filter group and wavelet basis window function and uncertainty principle, possesses good adaptivity, the fundamental characteristics that has kept primary speech signal, its Performance Ratio wants superior many based on the voice coding strategy of bandpass filtering and wavelet analysis in theory.
Claims (10)
1. method of speech processing that is used for cochlear implant, it comprises the following steps:
Pre-treatment step, carry out pretreatment to the voice signal through the mike input;
The Hilbert-Huang transform step, carry out Hilbert-Huang transform to voice signal after pretreatment, obtains the Hilbert-Huang transform signal of a plurality of passages;
Modulation step,, for each passage, process the Hilbert-Huang transform signal of this passage, obtains the range signal corresponding with the Hilbert-Huang transform signal of this passage and frequency signal; With the frequency signal that the Hilbert-Huang transform signal with this passage is corresponding, diphasic pulse is modulated and obtained the diphasic pulse modulation frequency signal corresponding with the Hilbert-Huang transform signal of this passage; Modulate with the diphasic pulse modulation frequency signal that range signal corresponding to the Hilbert-Huang transform signal with this passage pair and the Hilbert-Huang transform signal of this passage are corresponding, obtain for the channel modulation signal that drives the electrode corresponding with this passage.
2. the method for speech processing for cochlear implant as claimed in claim 1 is characterized in that:
Described pre-treatment step comprises preemphasis and minute frame processing.
3. the method for speech processing for cochlear implant as claimed in claim 1 is characterized in that:
Described pre-treatment step comprises preemphasis, end-point detection and minute frame processing.
4. the method for speech processing for cochlear implant as claimed in claim 1 is characterized in that:
Described diphasic pulse is the diphasic pulse of this channel central frequency.
5. the method for speech processing for cochlear implant as claimed in claim 1 is characterized in that:
The described diphasic pulse pulse corresponding with the mistake zero position of intrinsic mode function corresponding to this passage Hilbert-Huang transform signal that be pulse position.
6. the method for speech processing for cochlear implant as claimed in claim 1 is characterized in that:
In modulation step, the Hilbert-Huang transform signal of each passage is divided into first via signal and the second road signal; Obtain the range signal corresponding with the Hilbert-Huang transform signal of this passage by first via signal; Obtain the frequency signal corresponding with the Hilbert-Huang transform signal of this passage by the second road signal.
7. the method for speech processing for cochlear implant as claimed in claim 6 is characterized in that:
According to the instantaneous amplitude formula, first via signal is carried out range signal and extract, the range signal that then will extract carries out low-pass filtering, thereby obtains the range signal corresponding with the Hilbert-Huang transform signal of this passage.
8. the method for speech processing for cochlear implant as claimed in claim 6 is characterized in that:
According to the instantaneous frequency formula, the second road signal is carried out frequency signal and extract, thereby obtain the frequency signal corresponding with the Hilbert-Huang transform signal of this passage.
9. the method for speech processing for cochlear implant as claimed in claim 8 is characterized in that:
Before with the diphasic pulse pair frequency signal corresponding with the Hilbert-Huang transform signal of this passage, modulating, according to people's ear, the scope of frequency perception is carried out limit bandwidth to the frequency signal that extracts and process.
10. the method for speech processing for cochlear implant as claimed in claim 9 is characterized in that:
After with the diphasic pulse pair frequency signal corresponding with the Hilbert-Huang transform signal of this passage, modulating, further carry out low-pass filtering, obtain the diphasic pulse modulation frequency signal corresponding with the Hilbert-Huang transform signal of this passage.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103288270A CN103393484A (en) | 2013-07-31 | 2013-07-31 | Voice processing method used for electrical cochlea |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103288270A CN103393484A (en) | 2013-07-31 | 2013-07-31 | Voice processing method used for electrical cochlea |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103393484A true CN103393484A (en) | 2013-11-20 |
Family
ID=49557365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013103288270A Pending CN103393484A (en) | 2013-07-31 | 2013-07-31 | Voice processing method used for electrical cochlea |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103393484A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103605291A (en) * | 2013-11-05 | 2014-02-26 | 南车株洲电力机车有限公司 | Semi-physical simulation method of AC traction electric transmission system and system structure |
| CN106108897A (en) * | 2016-07-20 | 2016-11-16 | 西安中科比奇创新科技有限责任公司 | A kind of electromyographic signal filtering method based on empirical mode decomposition |
| CN111768802A (en) * | 2020-09-03 | 2020-10-13 | 江苏爱谛科技研究院有限公司 | Artificial cochlea voice processing method and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6862558B2 (en) * | 2001-02-14 | 2005-03-01 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Empirical mode decomposition for analyzing acoustical signals |
| CN101582260A (en) * | 2009-05-20 | 2009-11-18 | 王卫东 | Non-amplitude modulation speech coding method and device in electrical cochlea |
| CN101637420A (en) * | 2001-08-27 | 2010-02-03 | 加利福尼亚大学董事会 | Method and device for improving audio signal |
| CN103117066A (en) * | 2013-01-17 | 2013-05-22 | 杭州电子科技大学 | Low signal to noise ratio voice endpoint detection method based on time-frequency instaneous energy spectrum |
-
2013
- 2013-07-31 CN CN2013103288270A patent/CN103393484A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6862558B2 (en) * | 2001-02-14 | 2005-03-01 | The United States Of America As Represented By The Administrator Of The National Aeronautics And Space Administration | Empirical mode decomposition for analyzing acoustical signals |
| CN101637420A (en) * | 2001-08-27 | 2010-02-03 | 加利福尼亚大学董事会 | Method and device for improving audio signal |
| CN101582260A (en) * | 2009-05-20 | 2009-11-18 | 王卫东 | Non-amplitude modulation speech coding method and device in electrical cochlea |
| CN103117066A (en) * | 2013-01-17 | 2013-05-22 | 杭州电子科技大学 | Low signal to noise ratio voice endpoint detection method based on time-frequency instaneous energy spectrum |
Non-Patent Citations (2)
| Title |
|---|
| HONGYUN LIU ET AL: "A Novel Speech Coding Algorithm for Cochlear Implants", 《2012 5TH INTERNATIONAL CONFERENCE ON BIOMEDICAL ENGINEERING AND INFORMATICS (BMEI 2012)》, 18 October 2012 (2012-10-18), pages 403 - 406, XP032395331, DOI: doi:10.1109/BMEI.2012.6512918 * |
| 申丽然: "基于Hilbert-Huang变换理论的语音增强初探", 《哈尔滨工程大学学报》, vol. 26, no. 2, 30 April 2005 (2005-04-30), pages 220 - 224 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103605291A (en) * | 2013-11-05 | 2014-02-26 | 南车株洲电力机车有限公司 | Semi-physical simulation method of AC traction electric transmission system and system structure |
| CN103605291B (en) * | 2013-11-05 | 2016-09-07 | 中国铁路总公司 | The Hardware In The Loop Simulation Method of a kind of alternating current traction electric drive system and system architecture |
| CN106108897A (en) * | 2016-07-20 | 2016-11-16 | 西安中科比奇创新科技有限责任公司 | A kind of electromyographic signal filtering method based on empirical mode decomposition |
| CN111768802A (en) * | 2020-09-03 | 2020-10-13 | 江苏爱谛科技研究院有限公司 | Artificial cochlea voice processing method and system |
| WO2022048041A1 (en) * | 2020-09-03 | 2022-03-10 | 江苏爱谛科技研究院有限公司 | Voice processing method and system for cochlear implants |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US4536844A (en) | Method and apparatus for simulating aural response information | |
| CN101642399B (en) | Artificial cochlea speech processing method based on frequency modulation information and artificial cochlea speech processor | |
| CN100469158C (en) | Audio processing method and system implantted cochlea | |
| CN102958561B (en) | The timing of envelope particular stimulation | |
| CN101645267B (en) | Voice processing method applied in electronic cochlear | |
| CN103456312B (en) | A kind of single-channel voice blind separating method based on Computational auditory scene analysis | |
| CN100502819C (en) | Artificial cochlea manufacture method suitable for Chinese voice coding strategy | |
| CN101716108B (en) | System for processing electronic cochlear signal and coding electric stimulation | |
| EP3056022B1 (en) | Arrangement and method for extracting temporal features from spike-like signals | |
| CN103903632A (en) | Voice separating method based on auditory center system under multi-sound-source environment | |
| CN103892939B (en) | Improve language processing device for artificial cochlea and the method for Chinese tone recognition rate | |
| CN102737637B (en) | Method for data transmission by using voice-like modulation and demodulation | |
| CN103393484A (en) | Voice processing method used for electrical cochlea | |
| CN104778948A (en) | Noise-resistant voice recognition method based on warped cepstrum feature | |
| CN101582260B (en) | Non-amplitude modulation speech coding method and device in electrical cochlea | |
| US8670582B2 (en) | N band FM demodulation to aid cochlear hearing impaired persons | |
| CN103559893B (en) | One is target gammachirp cepstrum coefficient aural signature extracting method under water | |
| Li et al. | Harmonic coherent demodulation for improving sound coding in cochlear implants | |
| CN107527625A (en) | Dolphin whistle signal aural signature extracting method based on analog cochlea in bionical auditory system | |
| CN103761969A (en) | Perception domain audio encoding method and system based on Gaussian mixed model | |
| CN111150934B (en) | Evaluation system of Chinese tone coding strategy of cochlear implant | |
| Kumaresan et al. | Encoding bandpass signals using zero/level crossings: A model-based approach | |
| Liu et al. | A novel speech coding algorithm for cochlear implants | |
| Tian et al. | The application and computer simulation of multi-channel cochlear implant based on all phase DFT filter | |
| ESSAID et al. | New method based on single-channel separation algorithm using Gammatone filterbank for cochlear implants |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20131120 |