TAMIL WORDS SPEECH SYNTHESIS IN COCHLEAR IMPLANT USING

TAMIL WORDS SPEECH SYNTHESIS IN COCHLEAR IMPLANT USING ACOUSTIC MODEL GUIDED BY T. JAYASANKAR, ASST. PROFESSOR OF ECE, ANNA UNIVERSITY OF TIRUCHIRAPPALLI. PRESENTED BY C. SENTHILKUMAR, REG. NO: 810011992018, M. E(MBCBS), COM SYSTEM, VI MODULE.

OBJECTIVE A cochlear implant (CI) is a surgically implanted electronic device that provides a sense of sound to a person who is profoundly deaf or severely hard of hearing. Ø The main objective of this work is to develop the system that reproduces the incoming sound/speech signals as naturally as possible Ø

LITERATURE SURVEY S. NO TITLE AUTHORS YEAR & PUBLICATION CONCEPT 1 Estimation of Vowel Recognition With Cochlear Implant Simulations Chuping Liu and Qian-Jie Fu IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 1, JANUARY 2007 In this paper, Mel-frequency cepstrum coefficients (MFCCs) were used to estimate the acoustic vowel space for vowel stimuli processed by the CI simulations. 2 Improving Speech Intelligibility in Cochlear Implants using Acoustic Models P. VIJAYALAKSHMI, T. NAGARAJAN and PREETHI MAHADEVAN ISSN: 1790 -5052 Issue 4, Volume 7, October 2011 In this paper to improve the perceptual quality of the speech generated by a CI model, system specific parameters are analyzed by developing uniform bandwidth filterbank-based acoustic CI models 3 MIMICKING THE HUMAN EAR Philipos C. Loizon IEEE SIGNAL PROCESSING MAGAZINE 1053 -5888/98/$10. 000 1998 IEEE An overview of Signal- Processing Strategies for converting sound into Electrical sgnals in cochlear implants

SYSTEM DESIGN Speech Data Collection Ø Tamil words are recorded from a male speaker at a sampling frequency of 16 k. Hz with a head mounted carbon microphone of frequency range 20 Hz – 20 k. Hz using s PRAAT tool

CHANNEL VOCODER BASED ACOUSTIC MODEL General block Diagram ACOUSTIC MODEL INPUT SPEECH ANALYZER SYNTHESIZER SYNTHETI C SPEECH

CHANNEL VOCODER Uniform bandwidth filter bank method Ø Critical bandwidth filter bank method Ø

CHANNEL VOCODER ANALYZER(UNIFORM BANDWIDTH)

CONT… Acoustic model parameters Ø Sampling Frequency : 16000 Hz Ø Frequency Range : 0 -8200 Hz Ø Filter Type : IIR – Chebyshev type-2 Ø No. of Channels : 21 (1 LPF +20 BPF) Ø Bandwidth : 400 Hz Ø Order of filter : 5

CONT. . Filter order 1 2 3 4 5 6 7 8 9 10 Mean squared difference 0. 0033 0. 0032 0. 0031 0, 0031 0. 0030 1. 4861 e+238 4. 3540 e+300 1. 6547 e+299 2. 7285 e+296 3. 3515 e+299

CHANNEL VOCODER SYNTHESIZER(UNIFORM BANDWIDTH)

WAVEFORM OF THE TAMIL WORD /����� /

FILTERED SIGNAL & ITS ENVELOPE

TRAIN OF IMPULSE Pitch period =0. 0063 sec

MODULATED AND SYNTHESIZED FILTER OUTPUT

ORIGINAL & SYNTHESIZED SPEECH SIGNAL

CRITICAL BANDWIDTH FILTER BANK BASED ACOUSTIC CI MODEL Ø Critical band is the smallest band of frequencies that activate the same part of basilar membrane and human ear can able to discriminate two tones that differ in critical bands.

DESIGN OF CI MODEL BASED ON CRITICAL BANDS Ø Ø Ø Filter bank is designed based on critical bands of the human auditory system. The critical band of each auditory band-pass filter is computed using equivalent rectangular bandwidth (ERB). If the center frequencies (fc) of filters are known, then the corresponding ERBs are calculated using the following formula, ERB=24. 7((0. 00437*fc) +1) (1)

WAVEFORM OF INPUT AND SYNTHESIZED SPEECH FOR THE TAMIL WORD /����� /

MEAN SQUARE DIFFERENCE BETWEEN UNIFORM BANDWIDTH FILTER-BASED CI MODEL AND AUDITORY CI MODEL Mean square Difference 0, 005 0, 004 0, 0035 ����� 0, 003 ����� 0, 0025 ����� 0, 002 ����� 0, 0015 ���� 0, 001 0, 0005 0 Uniform Bandwidh Model Critical Bandwidth Model

MEAN OPINION SCORE(MOS) FOR UBW & CBW SYSTEM MEAN OPINION SCORE 4, 55 4, 45 4, 35 4, 25 MEAN OPINION SCORE 4, 15 4, 05 3, 95 UBW CBW

CONCLUSION The Critical band CI model is performed well when compared with the Uniform bandwidth filter bank method based on the mean square difference & Mean opinion score.

REFERENCES Ø Ø Ø P. Vijayalakshmi , T. Nagarajan and Preethi Mahadevan, (2011), “ Improving Speech Intelligibility in Cochlear Implants using Acoustic Models’’, WSEAS TRANSACTIONS on SIGNAL PROCESSING, Issue 4, Volume 7, October 2011, pp. 131 – 144. Gladston, A. R. ; Vijayalakshmi, P. ; Thangavelu, N. , "Improving speech intelligibility in cochlear implants using vocoder-centric acoustic models, " Recent Trends In Information Technology (ICRTIT), 2012 International Conference on , vol. , no. , pp. 66, 71, 19 -21 April 2012. D. K. Eddington, W. M. Rabinowitz, and L. Dellzome, “Sound Processing for Cochlear Implants”, in Proceedings of International IEEE EMBC, 2001, pp. 34493452. B. Gold and N. Morgan, “Speech and audio signal processing - processing and perception of speech and music”. John Wiley and Sons. Inc. , 2000. P. C. Loizou, “Speech processing in vocoder-centric cochlear implants” Cochlear and Brainstem Implants. Adv Otorhinolaryngol. Basel, Karger, vol 64, pp 109– 143, 2006. P. C. Loizou, ”Mimicking the human ear” IEEE Signal Processing magazine, vol. 15, no. 5, Sep. 1998, pp. 101 -130

Thank You