Dissertation > Excellent graduate degree dissertation topics show

Study on Matching Pursuit Based Low-bit Rate Speech Coding

Author: ZhangWenYao
Tutor: WangYuGuo
School: Institute of Software
Course: Applied Computer Technology
Keywords: Matching tracking Speech coding Speech Enhancement Sinusoidal modeling Pitch estimation Psychoacoustic model
CLC: TN912.3
Type: PhD thesis
Year: 2002
Downloads: 430
Quote: 39
Read: Download Dissertation

Abstract


Have been able to produce a very high quality of the reconstructed speech speech coding technology in high rate and in the rate, but the low rate even very low bit rate, high quality speech coding is still a challenging forefront of theoretical significance and potential practical value issues, prompted many researchers to explore new techniques and methods, such as new sinusoidal modeling technology, the new parameter quantitative methods to achieve low high quality voice coding. This article is the direction sinusoidal modeling sine analysis, matching tracking technology, combined with psycho-acoustic model, new modeling method and model parameters quantization coding, a useful exploration of low-bit-rate speech coding and related issues and as innovative research results: 1. Matching tracking technology processing speech signal enhancement problem, given a match tracking signal enhancement process coherent than the threshold method for determining, in the case of unknown statistical characteristics of signal and noise, to significantly enhance the signal within a large range of purposes. 2. Based on the to match tracking sinusoidal modeling, dynamic masking threshold perception of the concept of gradient, as well as perception gradient sinusoidal modeling algorithm. Perceptual gradient sinusoidal modeling better use psychoacoustic model maximum increase in the modeling process the synthesized signal awareness information to improve the modeling efficiency. Even in the case of model accuracy is not high, the method can be obtained relatively good quality speech synthesis. 3. Sinusoidal model parameters quantization coding the amplitude parameter vector quantization frequency parameter quantized difference methods, and to explore the the frequency box quantitative models and random-phase and zero-phase model. These methods are effective to reduce the coding bit rate. 4. Around the reduction of the coding bit rate and voice quality improved, and studied a series of compression coding scheme to stepwise fine layers of progressive manner and, ultimately, to propose a bit rate in the coding scheme of 1.5 ~ 2.4kbps. For various modeling methods and parameters of quantitative techniques, this paper discusses the tracking sinusoidal modeling based on ordinary match coding, sinusoidal modeling perceptual gradient compression coding, coding based on dynamic dictionary matching tracking, classification of dynamic dictionary compression coding, as well as combined with perceived gradient sinusoidal modeling and classification of dynamic dictionary coding scheme. The results found that match tracking the sinusoidal modeling has great potential in the low-bit-rate speech coding, low high quality speech coding explore a new technology roadmap. Finally, the proposed coding scheme more consideration psychoacoustic factors, the integration of the classification process dynamic dictionary and perceptual gradient modeling idea, is better than existing international coding methods and standards in the encoding bit rate and the quality of synthetic speech . 5. The proposed CAMDF function, and estimation algorithm based on the classification of voice and pitch of CAMDF compression coding scheme can be utilized. To overcome the lack of traditional AMDF function due CAMDF new pitch detection algorithm not only effectively reduce the false positive rate, but also simplifies the process of pitch detection, to improve the accuracy of the estimate. The classification of CAMDF voice achieved a satisfactory result. Finally, concludes the paper analyzes the current research work place should be further improved, and pointed out the direction of the next phase of the study, as well as some of the outlook.

Related Dissertations

  1. Compensation Methods of Different Speech Coding for Speaker Recognition,TN912.34
  2. The Studt of ILBC Speech Coding’s Key Technology and Its DSP Design and Optimization,TN912.3
  3. Research on Key Technologies of Digital Hearing Aids Based on Auditory Masking,TN912.3
  4. Medical Images Denoising Based on Sparse Decomposition,TP391.41
  5. Single channel speech enhancement algorithm,TN912.35
  6. Multi- directional tube array speech enhancement system development,TN912.35
  7. Based on auditory lateral inhibition filtering algorithms,TN911.72
  8. Noisy environments Microphone Array Speech Enhancement Technology,TN912.35
  9. Complex environment speech enhancement method,TN912.35
  10. Research on Very Low Bit Rate Speech Coding,TN912.3
  11. Research of Low-bit-rate Speech Coding Algorithms Based on MELP,TN912.3
  12. The Rearch on the Technology of MELPe Speech Coder,TN912.3
  13. Research on Subspace-Based Speech Enhancement,TN912.35
  14. Research on Speech Enhancement Method at Low Signal-to-Noise Ratio,TN912.35
  15. Realization and Optimization of Speech Enhancement Based on DM6446,TN912.35
  16. The Study of Embedded Speech Recognition System Based on Arm,TN912.34
  17. Research and Implementation of Adaptive Low-Rate Speech Coding,TN912.3
  18. Research of Speech Enhancement Algorithm under Non-stationary Environments,TN912.35
  19. Research on Microphone Array Speech Enhancement,TN912.35
  20. Research on Efficient Parameter Quantization Algorithms for Low-bit-rate Speech Coding,TN912.3
  21. Research on Audio Classification Technology,TN912.3

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Electro-acoustic technology and speech signal processing > Speech Signal Processing
© 2012 www.DissertationTopic.Net  Mobile