Dissertation > Excellent graduate degree dissertation topics show

Research and Implementation of Chinese continuous speech recognition system

Author: ZhangLiPing
Tutor: FengHongWei
School: Northwestern University
Course: Computer Software and Theory
Keywords: Speech recognition Endpoint detection Mel frequency standard cepstral coefficients Dynamic Time Warping
CLC: TN912.34
Type: Master's thesis
Year: 2010
Downloads: 286
Quote: 2
Read: Download Dissertation

Abstract


Speech recognition is the use of computer processing to the human voice , the voice signal is converted to a technique for text symbols . Chinese speech recognition at home and abroad has been a history of nearly 60 years , has made great progress , but there are still a lot of problems . Existing speech recognition technology has not yet reached less than a target man and machine through natural language interaction , large vocabulary continuous speech recognition of non - specific human remains is the difficulty of speech recognition research focus . This paper studies the key Chinese continuous speech recognition technology . First introduced the voice recognition principle , the composition of the speech recognition system as well as basic knowledge of Chinese speech . Then speech recognition preprocessing, feature extraction, pattern matching and post-processing phase function and its key technologies and improvement program for the existing problems in the traditional method . The main work of this paper are : 1 ) in the PC platform , using Microsoft Visual C MATLAB, Microsoft SQL Server and other tools to achieve a medium- vocabulary , non- specific Chinese continuous speech recognition system , and the system is experimental . The system selects the sound vowels as identification primitive characteristic parameters using the Mel frequency standard cepstral coefficients , recognition model selection dynamic time warping model . 2) identify the primitive segmentation accuracy impact on the identification performance of the system is large, the existing sound vowel segmentation method is divided in a non - continuous speech with high accuracy , but significantly reduced in continuous speech segmentation accuracy . To solve this problem , this paper combined the characteristics of the Chinese continuous speech , the use of the vowel formants energy entropy and Chinese design of a new acoustic vowel segmentation method to improve the accuracy of the acoustic vowel split . 3 ) using conventional dynamic time warping techniques speech recognition system in the identification , large amount of calculation , the system response time is long . To solve this problem , we propose algorithm and vector threshold -based speech features to be tested , improved template threshold - based DTW DTW improved algorithm effectively reduces the amount of calculation and improve the real-time nature of the system .

Related Dissertations

  1. Multiple ANN/HMM Hybrid Used in Speech Recognition,TN912.34
  2. The Design of a DSP-Based Robot Speech Command Recgnition System,TN912.34
  3. The Design and Research of Health Management Based on Smartphone Environment,TN929.53
  4. Power spectrum estimation in the broadband ADCP Signal Detection Research and Application,TN911.23
  5. Signature Verification Based on Video,TP391.41
  6. Mobile robot voice recognition control simulation system design and implementation,TN912.34
  7. Fast Time Series Similarity Matching and Its Application Research in Molten Iron Silicon Content Modeling,TF513
  8. Notation of Speaking Face Based on Video and Text Infomation,TP391.41
  9. Research on Stroke Distance Based Handwriting Document Retrieval Algorithm,TP391.43
  10. Phone-level Based Mispronunciation Automatic Detection and Its Application,TN912.34
  11. A Design of Intelligent Terminal of Voice Control Based on ARM9,TN912.3
  12. Application of Speech Recognition in the Testing System of Electromagnetic Valve,TN912.34
  13. The Application of Similarity Query Based on DTW in Well-completion Depth Calculation,TE257
  14. Design & Implementation of Medolic-Based Music Retrieval System,TP391.3
  15. Research of Emotion Recognition Based on Combined Speech Feature,TN912.3
  16. Research and Implementation of Speech Recognition System for Mobile Robot,TN912.34
  17. The Chinese vowel length adjustment based speech recognition,TN912.34
  18. Gait acceleration signal based authentication method,TN911.7
  19. The Study of Gait Recognition Based on Image Sequence and Pressure,TP391.41
  20. The Design and Research of Level-Testing System of the Disyllable of Mandarin Chinese,TN912.34
  21. Research on 3G Mobile Voice Control of a Multimodal Health Information Web Portal,TN929.53

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