This paper discusses the possibility of an input-dependent orthogonal transform for low bit rate speech coding. An orthogonal transform, based on an autoregressive model of short-time input speech, which is applicable to low bit rate coding, is presented. Its performance is evaluated in comparison with that for conventional transforms, KLT and DCT. The experiments confirm that the input-dependent orthogonal transform improves the average distortion versus average information rate performance over that for the input-independent transform DCT.

This paper proposes a new low bit-rate speech coding algorithm based on a multi-pulse excitation technique. In the algorithm, an excitation signal in a frame, which includes several pitch periods, is effectively represented by pulses during only one pitch period. Excitation signal for other pitch periods in the frame is reproduced by interpolating the pulses. From several experiments to evaluate the new coder, it is found that the new coder produces natural-sounding speech at low bit rates. Subjective evaluation results show that the new coder at 4.8kb/s attains good speech quality which is almost equivalent to that for 6bit/sample/µ-law PCM.

This paper describes multi-pulse speech coding with pitch prediction. Two realization approaches to the coding, which differ in regard to pitch predictor, are described. Coding performances for the two approaches are evaluated objectively and subjectively. The introduction of pitch prediction to a multipulse coder shows 2-3 dB SNR (Signal to Noise Ratio) improvement, and perceptually annoying noise is reduced over a conventional coder without pitch prediction, at bit rates above 8 kbps.

This paper describes and compares several kinds of pulse search methods for multipulse excited speech coder realization. These pulse search methods are derived from minimization criterion for errorpower between original speech and synthetic speech, but their performances and required computation amounts are different. Objective and subjective evaluations are carried out to compare the performances. Further, the relation between speech quality and pulse search method complexity is described. Based on these results, pulse search methods suitable for realizing a high-quality multipulse coder using current VLSI technology are discussed.

A full-duplex 16 or 9.6 k bits/sec speech codec based on the multi-pulse excited LPC coding technique has been developed. Cross-correlation analysis was used for the pulse search method. The codec was implemented by using five NEC 7720 Digital Signal Processor LSI chips ; four DSPs for the coder and one DSP for the decoder. A general purpose microprocessor was also used for system control. Reconstructed speech was quite natural and almost undiscernible from original speech at 16 k bits/sec and still quite good at 9.6 k bits/sec. Signal to noise power ratio for a sine wave signal was 30 dB at 16 k bits/sec.

A compact 16kbps speech codec for digital radio communications was developed using three NEC signal processor LSI chips. Adaptive Predictive Coding (APC) with variable length coding for residual signal was employed in the codec. Additional functions for communications such as error correction and frame synchronization were also included. Actual board size for the half duplex codec is 9 × 15 cm. A 20.5 dB segmental SN ratio was obtained.

This paper discusses a compact adaptive echo canceller with digital signal processor (SP) chips on the basis of a cascadable structure. By using this cascadable structure, tap number expansibility and program modification for satisfying various requirements can be easily obtained. Two kinds of echo cancellers were implemented. One for teleconference use, which has 600 taps with 12 SPs, has accomplished more than 20 dB echo cancellation in a conference room. It provides subjective high performance. The other, for telephone line use, with up to 300 taps, attains more than 30 dB echo return loss enhancement (ERLE). Each of these echo cancellers was implemented on a single board.

An ADPCM codec, that can provide toll quality speech at a 32 kb/s transmission rate, has been implemented on a single chip signal processor. Maximum effort has been paid to design a robust adaptation scheme for a quantizer and a predictor to withstand transmission bit errors. The codec employs a simplified robust quantizer and also employs a new backward adaptive predictor. The decoder, including the new adaptive predictor, has a structure having fixed poles and adaptive zeros, attaining both high prediction capability and robustness. The performance of a developed codec, which has analog interface capability through a PCM codec chip, satisfies the standard 64 kb/s PCM performance specification in CCITT recommendation G.712.

Performance improvement of the nearly instantaneous companding PCM (or block companding PCM) is discussed. A simplified algorithm is proposed. The design and performance of a channel bank adopting the proposed algorithm is described. The channel bank can transmit 44 voiceband signals on a standard T1 line. It provides commercially acceptable speech quality as well as sufficient transparency for voiceband data up to 4800 bits/s.

The development of a microprocessor-controlled 96/48 DSI terminal is described. The terminal is intended for domestic use, and is designed so as to maximize compatibility with existing digital carrier systems, it merges four T1 signals into two T1 compatible bit streams for transmission. The terminal is given sufficient excess capacity to accommodate voiceband data, while retaining toll-grade transmission quality for speech. Implementation approaches and considerations in both calculated and measured system characteristics are described.

An adaptive echo canceler with two echo path models is proposed to overcome the false adaptation problem for double-talking. The echo canceler possesses two separate echo path models (EPMs), one (background EPM) for adaptively identifying echo path transfer characteristics and the other (foreground EPM) for synthesizing an echo replica to cancel out echo. The parameter values of the foreground EPM are refreshed by those of the background EPM, according to a transfer control logic, when the logic determines that the background EPM is giving a better approximation of echo path transfer characteristics than the foreground EPM. Completely digital hardware implementation is described. Using the hardware, it is shown that virtually complete double-talking protection is actually realizable by the new method.