Multipulse Coding Research Articles

융복합 시스템의 8kbps에 있어서 APC-MPC에 관한 연구

In a MPC(Multi-Pulse Coding) using excitation source of voiced and unvoiced, it would be a distortion of voice waveform. This is caused by normalization of synthesis speech waveform of voiced in the process of restoration. To solve this problem, this paper present APC-MPC of amplitude-position compensation in a multi-pulses each pitch interval in order to reduce distortion of synthesis waveform. Also, I was implemented that the APC-MPC in coding system. And I evaluate the SNRseg of APC-MPC in 8kbps coding condition of convergence system. As a result, SNRseg of APC-MPC was 13.9dB for female voice and 14.3dB for male voice respectively. And so, I expect to be able to this method for cellular phone and smart phone using excitation source of low bit rate.

A method of coding LSP residual signals using wavelets for speech synthesis

AbstractThis paper presents a method of using wavelet analysis for speech coding and synthesis by rule. It is a coding system where LSP residual signal is transformed into wave‐let coefficients. As wavelet analysis is implemented effectively by filter banks, our method requires less computation than multipulse coding and others where complicated prediction procedures are essential. To achieve good‐quality speech at low bit rates, we verified allocation of different bits onto the wavelet coefficients, with more bits in lower frequencies, and less in higher. The synthesized speech by Haar wavelet with 16.538 kbit/s has nearly the same perceptual quality with 6 bits µlog PCM (66.15 kbit/s). We are convinced that the coding method of LSP residual signals using wavelet analysis is an effective approach to synthesize speech. © 2004 Wiley Periodicals, Inc. Electr Eng Jpn, 148(3): 54–61, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/eej.10326

Coding Method of LSP Residual Signals Using Wavelets for Speech Synthesis

This paper presents a method to use wavelet analysis for speech coding and synthesis by rule. It is a coding system where LSP residual signal is transformed into wavelet coefficients. As wavelet analysis is implemented effectively by filter banks, our method is featured to require less computation than multipulse coding and others where complicated prediction procedures are essential. To achieve good quality speech at low bit rates, we verified to allocate the different bits onto the wavelet coefficients, with more bits in lower frequencies, and less in higher. The synthesized speech by Haar wavelet with 16.538kbits/s has nearly same perceptual quality with 6 bits μlog PCM (66.15kbits/s). We are convinced that coding method of LSP residual signals using wavelet analysis is an effective approach to synthesize speech.

Pulse-by-pulse reoptimization of the synthesis filter in pulse-based coders

Two iterative algorithms for pulse-based linear predictive (PB-LP) analysis are developed. At the kth step, one of the algorithms assumes that the pulse locations of k pulses are known and calculates the LP coefficients jointly with k pulse amplitudes. In contrast, the other algorithm assumes that the previous k-1 pulse amplitudes are known and jointly estimates the LP filter coefficients and the kth pulse amplitude. These algorithms are proposed for synthesis filter reoptimization in pulse-based LP coders, and their effectiveness in multipulse coding is demonstrated through several simulations.

Selection of excitation vectors for the CELP coders

The authors investigate several algorithms that construct the input for the synthesis filter in the CELP coder, they present them under the same formalism, and compare their performances. They model the excitation vector by a linear combination of K signals, which are issued from K codebooks and multiplied by K associated gains. They demonstrate that this generalized form incorporates several particular coders such as code excited linear predictive coders, multipulse coders, self excited vocoders, etc. The least squares minimization problem is presented afterwards. In the case of orthogonal codebooks, they show that the optimal solution of this least squares problem is equivalent to orthogonal transform coding. They use the Karhunen-Loeve transform to design the corresponding orthogonal codebooks. In the case of nonorthogonal codebooks, they are restricted to suboptimal iterative algorithms for index selection and gain computation. They present some new algorithms based on orthogonalization procedures and QR factorizations that attempt to reduce this suboptimality. In a particular case, when the excitation is modeled using one gain coefficient (for example, ternary excitation or concatenation of short codebook vectors), an iterative angle minimization algorithm is proposed for index selection. The different extraction algorithms are compared with regard to the resulting coder complexity and synthetic speech quality. They find a particularly attractive method that consists of modeling the excitation with one unique gain. >

Means for improving the speech quality in multi‐pulse excited linear predictive coding

A technique that reconciles the differences between the estimator and the filter of a multi-pulse linear predictive voice encoder achieves a higher quality in the output speech. The technique simultaneously solves for the pulse amplitudes and pitch tap gain to minimize the estimator bias in the multi-pulse excitation and thereby improves, performance of the system. The increased signal-to-noise ratio is accomplished by first modifying the pitch predictor such that the pitch synthesis filter accurately reflects the estimation procedure used to find the pitch tap gain and, second, improving the excitation analysis technique such that the pitch predictor tap gain and pulse amplitudes are solved for simultaneously, rather than sequentially. Neither of these modifications results in an increased transmission rate and they do not significantly increase the complexity of the multi-pulse coding algorithm.

Method for protecting multi-pulse coders from fading and random pattern bit errors

A low-overhead method of protecting multi-pulse speech coders from the effects of severe random or fading pattern bit errors combines a standard error correcting code (convolutional rate 1/2 coding and Viterbi trellis decoding) for protection in random errors with cyclic redundancy code (CRC) error detection for fading errors. Compensation for detected fading errors takes place within the speech coder. Protection is applied only to the perceptually significant bits in the transmitted frame.

Multi-pulse coding apparatus with a reduced bit rate

First Page

Hybrid switched multi-pulse/stochastic speech coding technique

First Page

Adaptively controlled 4.8 kbit/s speech coders with low distortion

AbstractAs a low bit‐rate speech coding method to realize a satisfactory quality of reconstructed speech, this paper proposes a 4.8 kbit/s pitch‐controlled multipulse coding method, which adaptively controls the speech pitch prediction interval and represents the multipulse positions for the excitation signals with a small number of bits. As a method to control the pitch adaptively, an algorithm is derived and proposed where the pitch prediction interval is adjusted adaptively for each frame so that the pulse position is determined to minimize the perceptual error power representing subjectively important difference.In addition to the forementioned method, another method also is proposed to improve the speech quality, where the pitch‐controlled multipulse coding is proposed to control adaptively the coefficients of the synthesis filter. An algorithm is proposed for the adaptive control of the coefficients of the synthesis filter, to calculate anew the optimal filter coefficients for excitation pulses derived from the adaptive control of the pitch.A simulation is executed for the two methods, and the following results are obtained as an evaluation of SNRseg. It is 7.5 dB in the conventional MPC algorithm, 8.6 dB in the proposed pitch‐controlled MPC algorithm, and 9.8 dB in the pitch‐controlled MPC algorithm including the adaptive control of the synthesis filter coefficients. In other words, the improvement in the latter two is 1.1 and 2.3 dB, respectively. The performance is satisfactory also from the viewpoint of the cepstral distance. The improvement in the reconstructed speech quality also is verified for the proposed methods by the subjective evaluation.

A New Approach to Multipulse Speech Coding

This paper introduces a new approach to classical Multipulse Speech Coding. Based on spiky deconvolution techniques, which take advantage of the sparse character of the multipulse sequence, a new accurate scheme for generating the excitation sequence in multipulse coders is presented and analyzed. Using different quantization procedures, we propose a multipulse coder that can be adjusted to work within bit rates ranging from 4.8 to 9.1 kbit/s. The performance and computational burden of the new coder are also evaluated.

Method and apparatus for multi‐pulse speech coding

A multi-pulse speech coder uses synthetic filters for generating crosscorrelated signals without pitch prediction and autocorrelated signals with pitch prediction. These signals are used as a basis for calculations to detect the correlations between the signals with pitch predictions and input speech signals.

A reduced complexity multipulse compression system

A reduced complexity multipulse coder is proposed in this paper. A 33% complexity reduction affects the overall speech output quality only slightly as is demonstrated in both informal listening and SNR computation. On the other hand, it makes possible the implementation of the coder with cheap commercial signal processing chips. The TMS32020 has been used for real time implementation of the proposed algorithm at 16 kbit/s with only fixed point computations. The method studied here can be considered as an alternative to the regular pulse excitation technique.

Source coding of speech and video signals

Some digital source coding techniques for speech and video are reviewed. Predictive coding of speech, multipulse and code-excited coders and frequency-domain coders are discussed and compared for the coding of speech signals, and intraframe and still image coding and interframe coding are examined for the coding of image and video signals. The emphasis is on those algorithms that offer high compression while maintaining the perceptual quality of the source signals are discussed. Some algorithms that are general waveform coding algorithms and do not strictly depend on the input source are included. >

Improved excitation prediction and quantization in optimal amplitude multipulse coders

Improvements to the excitation prediction and quantization procedure in optimal amplitude multipulse coders are described. The new procedure successively reoptimizes the pulse amplitudes and predictor gain of the long‐delay correlation filter as each new pulse is found. Furthermore, the amplitudes and gain are requantized at each step so that the new pulse amplitudes and locations can correct for the quantization errors in the existing excitation. To perform the joint amplitude‐predictor reoptimization, nonrecursive pitch prediction structure is used in our analysis. The predictor is implemented as an adaptive vector quantizer whose codebook is populated with past excitation sequences. This structure allows for a computationally efficient closed‐loop analysis procedure. The results showed that at 9.6 kbps, the new technique achieved an average segmental SNR of over 18.5 dB. Informal subjective tests indicated that the reconstructed speech is toll quality.

Amplitude optimization and pitch prediction in multipulse coders

Although the multipulse model is conceptually simple, the problem of locating the pulses is computationally complex. The authors discuss the basic multipulse model and describe a procedure to compute the excitation with optimally adjusted amplitudes. The algorithm provides a framework for computing multipulse excitation with varying degrees of optimization and computational complexity. The authors find that speech quality depends on the pulse rate. They also find that for the same quality, female speech requires a higher pulse rate than male speech. The pitch dependence can be reduced and speech quality improved for high-pitched speakers by incorporating long delay prediction in the multipulse model. >

Perceptual evaluation of two speech coding methods by native and non-native speakers of English

The segmental intelligibility of two speech coding methods was evaluated using the modified rhyme test (MRT). In order to investigate how language knowledge and experience affects the perception of coded speech, we used both native and non-native speakers of English as listeners. We used 8 kb/s pitch predictive multi-pulse coding (MPC) and 50 kb/s μ-law PCM as coding methods, along with 120 kb/s unprocessed speech as a control condition. The results indicated that for native listeners, the intelligibility of unprocessed speech was best, followed by PCM and then MPC speech. However, for non-native listeners, the intelligibility of the two types of coded speech was much worse than the unprocessed speech when compared with the performance of the native English listeners. The difference in error rates between MPC and PCM was not significant for non-native listeners. In contrast, for native listeners, the error rate for PCM was significantly less than for MPC speech. Non-native listeners also had a tendency to confuse stops and fricatives more than native listeners, especially in the coded speech. These results suggest that language knowledge and experience play a much more important role in the perception of coded speech than in the perception of unprocessed speech. Furthermore, non-native listeners appear to be much more affected by some of the unique characteristics of coded speech than native listeners. The results have implications for the design and improvement of low bit-rate speech coding methods and for the development of new techniques for perceptual evaluation.

Impulse train equivalent excitation signals for high‐quality speech synthesis

AbstractIn the speech synthesis using filters, other than the impulse train, some special pulse trains such as an excitation signal are used, especially the pulse train by Barker sequence and the pulse train obtained by averaging in time the energy of the signal. However, they do not produce a high‐quality synthesized speech. By contrast, the possibility exists that a high‐quality speech synthesis can be achieved by utilizing the residual speech as the excitation signal. In this case, a certain data compression must be applied to the residual signal. Several useful methods are known for coding the residual speech, such as base‐band signal coding and multipulse coding. For efficient speech data compression or realization of high‐quality speech synthesis by a rule‐based speech synthesis, the lack of rules or formulation for the excitation source is a serious problem. To realize a high‐quality speech synthesis by a rule‐based excitation, a high‐quality impulse train equivalent excitation signal is needed to replace the usual impulse train. From such a viewpoint, this paper first performed the very‐short‐term analysis of the residual speech, confirming that the spectral envelope of the voiced sound interval varies with time in quite a characteristic way. A method of generating the impulse train equivalent signal (FTSITES) with the spectral envelope and fine‐time structure is proposed according to the result of analysis. It is shown that a high‐quality speech synthesis is achieved by utilizing FTSITES as the excitation signal.

Regular-pulse excitation--A novel approach to effective and efficient multipulse coding of speech

This paper describes an effective and efficient time domain speech encoding technique that has an appealing low complexity, and produces toll quality speech at rates below 16 kbits/s. The proposed coder uses linear predictive techniques to remove the short-time correlation in the speech signal. The remaining (residual) information is then modeled by a low bit rate reduced excitation sequence that, when applied to the time-varying model filter, produces a signal that is close to the reference speech signal. The procedure for finding the optimal constrained excitation signal incorporates the solution of a few strongly coupled sets of linear equations and is of moderate complexity compared to competing coding systems such as adaptive transform coding and multipulse excitation coding. The paper describes the novel coding idea and the procedure for finding the excitation sequence. We then show that the coding procedure can be considered as an optimized baseband coder with spectral folding as high-frequency regeneration technique. The effect of various analysis parameters on the quality of the reconstructed speech is investigated using both objective and subjective tests. Further, modifications of the basic algorithm, and their impact on both the quality of the reconstructed speech signal and the complexity of the encoding algorithm, are discussed. Using the generalized baseband coder formulation, we demonstrate that under reasonable assumptions concerning the weighting filter, an attractive low-complexity/high-quality coder can be obtained.

A Study on Pulse Search Algorithms for Multipulse Excited Speech Coder Realization

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.