Chair: Jean-Pierre Adoul, University of Sherbrooke (CANADA)
Masahiro Serizawa, NEC Corporation (JAPAN)
Kazunori Ozawa, NEC Corporation (JAPAN)
This paper proposes a new pitch prediction method for 4kbps CELP coding. In the conventional CELP speech coding, synthetic speech quality deteriorates rapidly at 4kbps, especially for females and children's with short pitch period. The important reason is that when the pitch period is shorter than the subframe length, simple repetition of the past excitation signal based on the estimated lag, not the true pitch prediction, is usually used in the adaptive codebook operation. The proposed pitch prediction method can carry out the true pitch prediction by utilizing the current subframe excitation codevector signal, when the pitch prediction parameters are determined. For further improvement, a split weighting method and a low-complexity harmonic and spectral perceptually- weighting method have also been developed. The informal listening test result shows that the 4kbps coder with 20 msec frame, utilizing all of the proposed improvements, achieves 0.2 MOS higher results than the coder without them.
Peter Kroon, AT&T Bell Laboratories (USA)
Michael Recchione, AT&T Bell Laboratories (USA)
With the deployment of digital cellular systems such as 13 kb/s GSM RPE-LTP, 8 kb/s fixed rate IS54 VSELP and 8 kb/s variable rate IS96 QCELP it is becoming clear that these coders are not as robust as originally expected. As a result there has been renewed interest in higher bit rate coders that provide toll quality performance with regard to various input signals and multiple encodings. This paper describes the implementation of a toll quality 13 kb/s CELP coder. This coder can also operate at a variable bit rate mode (13, 6.5 and 0.8 kb/s), and has potential applications for a revised CDMA cellular system and PCS systems based on GSM.
Juin-Hwey Chen, AT&T Bell Laboratories (USA)
In this paper, we present a 16 kb/s CELP coder with a complexity as low as 3 MIPS. The main thrust is to reduce the complexity as much as possible while maintaining toll-quality. This Low-Complexity CELP (LC-CELP) coder has the following features: (1) fast LPC quantization, (2) 3-tap pitch prediction with efficient open-loop pitch search and predictor tap quantization, (3) backward-adaptive excitation gain, and (4) a trained excitation codebook with a small vector dimension and a small codebook size. Most CELP coders require one full DSP or even two DSP chips to implement in real-time. In contrast, 3 to 6 full-duplex LC-CELP coders can fit into a single DSP chip, since each takes only around 3 MIPS to implement. This coder achieved slightly higher mean opinion scores (MOS) than the CCITT 32 kb/s ADPCM. It also exhibits good performance when tandemed with itself or transcoded with other coders.
Andrei Popescu, Telecom Paris
Nicolas Moreau, Telecom Paris
Claude Lamblin, CNET/LAA/TSS/CMC (FRANCE)
We describe a systematic procedure to replace vector quantization (VQ) with trellis-coded vector quantization (TCVQ) in existing CELP coders. Following this procedure, we design an 8/16 kbit/s TCVQ CELP coder. We analyze the performance of this method in terms of quality and complexity. Our results show that a CELP coder using TCVQ produces significantly better quality than the same coder using VQ, with reasonable complexity. By modifying the TCVQ CELP coder parameters one can favorably trade coding quality against complexity and/or delay.
Per Hedelin, Chalmers University of Technology (SWEDEN)
Thomas Eriksson, Chalmers University of Technology (SWEDEN)
A procedure is presented where the conventional innovation codebook approach of CELP coding is replaced by an interpolative scheme. A generalized LTP-codebook constitutes the basis for the interpolation. A pre-requisite to efficient interpolation concerns establishing the neighboring vectors of a given LTP entry. We discuss several working approximations for establishing a suitable neighborhood concept. By simulations we have found that the interpolative scheme leads to 17-20 bits for the excitation coding of one block. In addition to this, some 3-4 bits are required for block gain. A standard CELP typically spends 25 bits or more for one block of excitation coding. The subjective quality of our proposed coding scheme compares favorably with standard CELP. In particular, the interpolation improves the pitch-related properties giving a less noisy subjective impression.
Cheung-Fat Chan, City University of Hong Kong (HONG KONG)
A fast codebook search method for code excited linear predictive (CELP) coding of speech is described. The method relies on using a vector-sum codebook where the crosscorrelation of any pair of basis vectors is odd-symmetric. Due to this odd-symmetric crosscorrelation (OSC) property the energy term of the cost function for codebook search is a constant with respect to the search, and a simple sign detection procedure is used to locate the optimum codeword with a complexity almost independent of the codebook size. An algorithm for generating a set of OSC, the basis vectors will be described. Simulation results show that by replacing the standard VSELP codebooks with the OSC codebooks, same coder performance can be achieved but with a much lower complexity. An OSC- CELP coder was implemented and demonstrated to achieve good quality speech at rates below 4.8 kbps.
Torbjorn Wigren, Ericsson Radio Systems AB (SWEDEN)
Anders Bergstro, Ericsson Radio Systems AB (SWEDEN)
Susanne Harrysson, Ericsson Radio Systems AB (SWEDEN)
Fredrik Jansson, Ericsson Radio Systems AB (SWEDEN)
Hans Nilsson, Ericsson Radio Systems AB (SWEDEN)
A proper coding of background sounds has proven to be important in digital cellular telephone systems. It is therefore of considerable interest to find methods that improve also background sound coding, in particular in medium and low bitrate speech coders. In many coders operating around 8 kbps, one significant effect has been a swirling distortion when coding for example smooth car noise. The paper discusses the origin of this problem, and presents an algorithm that reduces swirling, giving a significant overall perceptual improvement, as shown by field tests in commercial D-AMPS systems in the US. A related new algorithm for error concealment is also presented. This method makes it possible to perceptually remove the effect of fading dips persisting for a few seconds, during nonspeech periods.
Akitoshi Kataoka, NTT Human Interface Labs
Sachiko Hosaka, NTT Human Interface Labs
Jotaro Ikedo, NTT Wireless Systems Laboratories
Takehiro Moriya, NTT Human Interface Labs. (JAPAN)
Shinji Hayashi, NTT Human Interface Labs. (JAPAN)
A high-quality 8-kbit/s speech coder based on Conjugate Structure CELP (CS-CELP) is proposed that uses a trained sparse conjugate codebook. The trained sparse conjugate codebook improves speech quality for noisy speech. This codebook consists of two sub-codebooks and each sub-codebook consists of a random component and a trained component. Each component has excitation vectors consisting of a few pulses. In the random component, pulse position and amplitude are determined randomly. The trained component is determined by training. Subjective tests (Differential Mean Opinion Score, DMOS and Mean Opinion Score, MOS) indicated that this codebook improves speech quality compared with the conventional trained codebook for noisy speech. The MOS showed that the quality of improved CS-CELP is equivalent to that of the 32-kbit/s ADPCM for clean speech.
Kazuhito Koishida, Tokyo Institute of Technology (JAPAN)
Keiichi Tokuda, Tokyo Institute of Technology (JAPAN)
Takao Kobayashi, Tokyo Institute of Technology (JAPAN)
Satoshi Imai, Tokyo Institute of Technology (JAPAN)
In this paper, we propose a CELP coder based on mel-cepstral analysis. In the coder, since the transfer functions of perceptual weighting and postfiltering are defined through mel-cepstral coefficients, the effects of perceptual weighting and postfiltering should fit with the characteristics of the human auditory sensation. We use a basic CELP structure without adaptive codebook, and subjective speech quality of the proposed coder in terms of the opinion equivalent Q is measured and compared with that of the conventional CELP coder. It is shown that the improvement of more than 1.8dB is achieved by the proposed coder over the conventional CELP coder.
Shude Zhang, University of Leeds (UK)
Gordon Lockhart, University of Leeds (UK)
The feasibility and performance of an embedded RPE (ERPE) scheme based on multistage coding is investigated. The coding efficiency of second and subsequent stages depends on the spectral envelope difference between the original speech and the error signal at each stage whereas re-use of LPC parameters derived from the original speech depends on the corresponding LPC spectral difference. Suitable measures of spectral difference are defined and simulation shows that both decrease with the perceptual weighting factor. The ERPE system requires little extra coding complexity and can be simplified further by using a partial phase adaptation procedure with marginal loss of SNR performance. The simulated ERPE system shows graceful reduction of reconstructed speech quality for bit rates from 14.8 to 6.4 kb/s in 4.2 kb/s steps.