Neural predictive coding for speech signal
In this paper, we present a new speech coding named NPC (Neural Predictive Coding). It is obtained thanks to a MLP (Multi Layer Perceptron) used in prediction. The system is designed to predict the samples of a signal window from previous ones. The goal of this coding is to extract the signal window characteristics relative to the database which it is extracted. After a precise description of our coding, we compare results obtained by our coding with the ones obtained by classic coding (MFCC, FFT, LAR, LPC and LPCC) on phoneme recognition. The NPC coding allows an improvement of the recognition rate in respect of the other coding.
Unable to display preview. Download preview PDF.
- A. Biem, S. Katagiri “FILTER BANK DESIGN BASED ON DISCRIMINATIVE FEATURE EXTRACTION”, ICASSP'94, I-485–488.Google Scholar
- J. Thyssen, H. Nielsen, S. Duus Hansen “NON-LINEAR SHORT-TERM PREDICTION IN SPEECH CODING”, ICASSP'94, I-185–188.Google Scholar
- F. diaz-de-Maria, A. R. Figueras-Vidal “NONLINEAR PREDICTION FOR SPEECH CODING USING RADIAL BASIS FUNCTIONS”, ICASSP'95, I-788–791.Google Scholar
- B. Townshend “NONLINEAR PREDICTION OF SPEECH”, ICASSP'91, I-425–428.Google Scholar
- C. Jankowski, K. Ashok, S. Basson, J. Spitz “NTIMIT: A PHONETICALLY BALANCED, CONTINUOUS SPEECH, TELEPHONE BANDWITH SPEECH DATABASE” ICASSP'90.Google Scholar