A digitized speech signal is divided into sections and each section is analyzed by the linear prediction method to determine the coefficients of a sound formation model, a sound volume parameter, information concerning voiced or unvoiced excitation and the period of the vocal band base frequency. In order to improve the quality of speech without increasing the data rate, redundance reducing coding of the speech parameters is effected. The coding of the speech parameters is performed in blocks of two or three adjacent speech sections. The parameters of the first speech section are coded in a complete form, and those of the other speech sections in a differential form or in part not at all. The average number of bits required per speech section is reduced to compensate for the increased section rate, so that the overall data rate is not increased.

Digital speech processing system having reduced encoding bit requirements

The article looks at the biomedical applications of speech technology. The techniques of speech synthesis and recognition are described first, and then the range of available products is reviewed. Current biomedical applications of the new technology are described and problem areas are highlighted. Finally, a prediction of the likely future role of speech technology in medicine is attempted.

Speech technology—implications for biomedical engineering

The introduction of multi-pulse excitation for LPC coders has increased the quality achievable for digitally coded speech at bit rates in the 9.6 Kbs range. A simplified multi-pulse analysis is proposed here with particular emphasis on the speech model developed as a result of a two-pass method. In the first pass, estimated LPC parameters generated by conventional covariance analysis are used to generate the forward prediction error; the multi-pulse sequence is then detected by thresholding the residual. The second pass generates the final LPC parameters by a covariance analysis incorporating knowledge of the estimated pulse locations and amplitudes along with perceptual synthesis error weighting. Experimental results are presented to demonstrate the method.

Efficient algorithm for multi-pulse LPC analysis of speech

It is difficult to produce natural-sounding speech using LPC voice coding techniques (vocoder). The main problem is the simplistic model of the excitation source in terms of the pitch pulses and white noise used in the synthesis. The LPC vocoder requires robust and accurate pitch determination, which is a problem that still has not been completely solved. Recently, a multi-pulse approach for modeling the excitation source has been proposed by Atal. However, this approach requires a substantial increase in the bit rate over conventional LPC. We describe in this paper a novel pulse source model for improving the quality of LPC speech. The binary pulse source excitation is obtained from the speech signal using a nonlinear filtering procedure. Our approach is able to maintain good speech quality even at bit rates of 4.8 kbps or lower.

A novel LPC synthesis model using a binary pulse source excitation

A speech coding and decoding apparatus used for linear predictive coding of a speech signal and for transmitting it in a digital signal. In coding operation, a linear predic­tive residual signal is extracted from an input speech signal subjected to linear predictive analysis. The strength of the correlativity between the pitch periods of the waveform of the extracted linear predictive residual signal is obtained for every plural blocks. The time axis of the linear predictive residual signal for every two adjacent pitch period sections for the block in which the strength of correlativity of the waveform is not less than a predetermined threshold value and is larger than that of another block is compressed into a residual signal for one pitch period section repeatedly. Quantization allotting bits are preferentially allotted to the portion of the linear predictive residual signal subjected to time-axis compression, and the quantized signal is transmitted. In decoding operation, the partially compressed and quantized linear predictive residual signal is separated from the transmitted signal, and the separated linear predictive residual signal is inversely quantized. The partially compressed portion of the inversely quantized linear predictive residual signal for every one pitch period section is partially expanded to the residual signal for two pitch period sections repeatedly so as to restore the reproduced residual signal which has the same length of original residual signal. The time axis of the linear predictive residual signal is compressed and expanded only at the portion which has a large waveform correlative strength, thereby ensuring good synthesized speech quality. And time-axis of linear predictive residual signal is compressed and expanded within analysis frame, thereby enhancing the proof to transmission error.

Speech coding and decoding apparatus

This paper presents improvements on the classical model which produces speech that is significantly better than currently available systems. The first major improvement results from treating speech as a two source phenomenon that can be separated for parallel but independent analysis/ synthesis. This two component decomposition is accomplished by making use of the quasi-periodic nature of 'voiced' speech. The second major improvement in bit compression and robustness of operation results from an efficient encoding of the prediction residuals of the two components. The key step is to encode the residual of the periodic component by picking out and transmitting the essential information for only one cycle (pitch period) of the residual.

S. Maitra

Papers

Improvements in the classical model for better speech quality