09-12-2009, 06:37 PM
DESIGN OF DEAF MANâ„¢S TELEPHONE USING DSP (Digital Signal Processor TMS320C31)
Abstract
The idea of the DEAF MANâ„¢S TELEPHONE is a new way of designing a interactive telephone answering system. This paper is a development of our project done in our college. Herein we introduce an interactive telephone answering system. This is particularly designed for Deaf man who cannot hear what the other person is speaking through the telephone. The processor converts the speech into text as readable to the deaf man. Speech recognition performance suffers from the involved nonlinear operations so that the greatly improved SNR often contrasts with only slight improvements in the recognition rate. Time-Frequency masking has emerged as a powerful technique for source separation of noisy and convolved speech mixtures. It has also been applied successfully for noisy speech recognition. Here, another approach is suggested for source separation, which estimates sufficient statistics for speech features in the preprocessing (e.g. STFT-)domain, propagates these statistics through the transforms from the spectrum to e.g. the MFCCâ„¢s of a speech recognition system and uses the estimated statistics for missing data speech recognition. With this approach, significant gains can be achieved in speech recognition rates, and in this context, time-frequency masking yields recognition rate improvements of more than 35% when compared to TF-masking based source separation. This paper describes a single chip speech recognition system. We use 10th order cepstrum coefficients and log-scaled energy as the feature vectors and Discrete Hidden Markov Model (DHMM) as recognition tool.
For more information, just post here.
Abstract
The idea of the DEAF MANâ„¢S TELEPHONE is a new way of designing a interactive telephone answering system. This paper is a development of our project done in our college. Herein we introduce an interactive telephone answering system. This is particularly designed for Deaf man who cannot hear what the other person is speaking through the telephone. The processor converts the speech into text as readable to the deaf man. Speech recognition performance suffers from the involved nonlinear operations so that the greatly improved SNR often contrasts with only slight improvements in the recognition rate. Time-Frequency masking has emerged as a powerful technique for source separation of noisy and convolved speech mixtures. It has also been applied successfully for noisy speech recognition. Here, another approach is suggested for source separation, which estimates sufficient statistics for speech features in the preprocessing (e.g. STFT-)domain, propagates these statistics through the transforms from the spectrum to e.g. the MFCCâ„¢s of a speech recognition system and uses the estimated statistics for missing data speech recognition. With this approach, significant gains can be achieved in speech recognition rates, and in this context, time-frequency masking yields recognition rate improvements of more than 35% when compared to TF-masking based source separation. This paper describes a single chip speech recognition system. We use 10th order cepstrum coefficients and log-scaled energy as the feature vectors and Discrete Hidden Markov Model (DHMM) as recognition tool.
For more information, just post here.