تحديد مكان وهوية المتكلم باستخدام تقنية توجيه حزم الاشارة المحسنة
Author name:
علي يعكوب يوسف
Supervisor name:
حسين عطية لفتة الخالدي
General topic:
Computer Science
Specific topic:
Computer Science
Degree:
Doctorate
University:
University of Babylon - College Of Science - Department Of Computer Science
Language:
English
University location:
Babylon
First pages:
28T766 - p.pdf
Abstract:
Automatic speech recognition (ASR) systems perform well when using a close - talking microphone, However, many environments (Hands - free) where the use of such microphones is undesirable for reasons of convenience. In a hands - free environment, the noise and reverberation degrade the accuracy of recognition. An enhanced approach using microphone array for speaker localization and enhancement of speech signal input to an automatic speech identification system was proposed. The proposed system using enhanced beamforming technique based on Minimum Variance Distortionless Response (MVDR) for speaker localization with multi - microphone arrays. The strongest output beam signal corresponding to selected microphone array, used for the speaker identification. The identification phase based on using Mel Frequency Cepstrum Coefficient MFCC for feature extraction and enhance LBG algorithm for speaker modeling.Speaker identification accuracy in using proposed method were compared with conventional beamforming method, It was found that the higher recognition accuracy than previous approaches, and in experiments using speech signals that were artificially corrupted by additive noise. The proposed system provided a consistent, improvement in recognition accuracy for several experiments in simulation environments. It is also showing the benefit of usingmicrophone array processing. The localization phase evaluated using SNR, showing enhanced ratio after applying enhanced beamforming that estimate speaker location. The results showing localization accuracy 96.8% and 98.1% recognition accuracy where achieved.
