Split Vector Quantization Research Articles

Split vector quantization for sinusoidal amplitude and frequency

In this paper, we suggest applying tree structure on the sinusoidal parameters. The suggested sinusoidal coder is targeted to find the coded sinusoidal parameters obtained by minimizing a likelihood function in a least square (LS) sense. From a rate-distortion standpoint, we address the problem of how to allocate available bits among different frequency bands to code sinusoids at each frame. For further analyzing the quantization behavior of the proposed method, we assess the quantization performance with respect to other methods: the short-time Fourier transform (STFT) based coder commonly used for speech enhancement or separation, and the line spectral frequency (LSF) coder used in speech coding. Through extensive simulations, we show that the proposed quantizer leads to less spectral distortion as well as higher perceived quality for the re-synthesized signals based on the coded parameters in a model-based approach with respect to previous STFT-based methods. The proposed method lowers the complexity, and, due to its tree-structure, leads to a rapid search capability. It provides flexibility for use in many speaker-independent applications by finding the most likely frequency vectors selected from a list of frequency candidates. Therefore, the proposed quantizer can be considered an attractive candidate for model-based speech applications in both speaker-dependent and speaker-independent scenarios.

적응적인 확장된 코드북을 이용한 분할 벡터 양자화기 구조의 ISF 양자화기 개선

본 논문에서는 ISF 계수의 순서화 성질을 이용하여 분할구조 벡터양자화기의 단점을 보완하여 ISF 계수 양자화의 성능을 높이는 알고리듬을 제안하고, 이를 이용한 광대역 음성 부호화기용 ISF 계수 양자화기를 설계한다. 16차 이상의 광대역 코덱의 ISF 계수는 계산량과 메모리 사용을 줄이기 위해서 분할구조의 벡터 양자화기를 사용한다. 분할구조 양자화기는 ISF 계수간의 상관도를 충분히 활용하지 못하는 단점이 발생한다. 제안하는 알고리듬은 이러한 단점을 극복하기 위하여 ISF 계수의 순서화 성질을 이용한다. ISF 계수의 순서화 성질을 이용하여 각 서브벡터의 불필요한 코드북 (Codebook Redundancy)을 검색할 수 있다. 이러한 불필요한 코드북은 ISF 계수의 순서화 성질, ISF 계수 예측과정과 기존 코드북의 보간법 (Interpolation)을 통해 적응적인 확장된 코드북으로 교체되어 양자화기의 성능을 향상시킨다. 제안된 알고리듬은 기존의 분할구조 양자화기에서 사용되지 못했던 17 %가량의 불필요한 코드북 인덱스를 적응적인 확장된 코드북에 할당하여, 표준화된 코덱인 AMR-WB의 ISF 계수 양자화기에 비해서 주파수 왜곡 관점에서 약 2 bit 가량의 이득을 보는 결과를 얻었다. This paper presents a method for improving the performance of ISF coefficients quantizer through compensating the defect of the split structure vector quantization using the ordering property of ISF coefficients. And design the ISF coefficients quantizer for wideband speech codec using proposed method. The wideband speech codec uses split structure vector quantizer which could not use the correlation between ISF coefficients fully to reduce complexity and the size of codebook. The proposed algorithm uses the ordering property of ISF coefficients to overcome the defect. Using the ordering property, the codebook redundancy could be figured out. The codebook redundancy is replaced by the adaptive-extended codebook to improve the performance of the quantizer through using the ordering property, ISF coefficient prediction and interpolation of existing codebook. As a result, the proposed algorithm shows that the adaptive-extended codebook algorithm could get about 2 bit gains in comparison with the existing split structure ISF quantizer of AMR-WB (G.722.2) in the points of spectral distortion.

Optimized Subvector Processing in Split Vector Quantization

Split vector quantization (SVQ) is efficient but suboptimal. Here a renormalization process is proposed for intraframe splitting and joining of subvectors, which integrates gracefully with trained interframe prediction. Renormalization increases the availability of codevectors for the quantization of each subvector in ordered vectors such as the line spectral frequency (LSF) vectors. For 16-dimensional LSF vectors from wideband speech, renormalized SVQ (RSVQ) is shown to achieve a savings of 4 bit/frame over standard SVQ, reaching transparent coding at 42 bit/frame. Further, predictive RSVQ saves an additional 4 bit/frame for transparent coding down to 38 bit/frame.

Entropy coding of compressed feature parameters for distributed speech recognition

In this paper, we propose several entropy coding methods to further compress quantized mel-frequency cepstral coefficients (MFCCs) used for distributed speech recognition (DSR). As a standard DSR front-end, the European Telecommunications Standards Institute (ETSI) published an extended front-end that includes the split-vector quantization of MFCCs and voicing class information. By exploring entropy variances of compressed MFCCs according to the voicing class of the analysis frame and the amount of the entropy due to MFCC subvector indices, voicing class-dependent and subvector-wise Huffman coding methods are proposed. In addition, differential Huffman coding is then applied to further enhance the coding gain against class-dependent and subvector-wise Huffman codings. Subsequent experiments show that the average bit-rate of the subvector-wise differential Huffman coding is measured at 33.93 bits/frame, which is the smallest among the proposed Huffman coding methods, whereas that of a traditional Huffman coding that does not consider voicing class and encodes with a single Huffman coding tree for all the subvectors is measured at 42.22 bits/frame for the TIMIT database. In addition, we evaluate the performance of the proposed Huffman coding methods applied to speech in noise by using the Aurora 4 database, a standard speech database for DSR. As a result, it is shown that the subvector-wise differential Huffman coding method provides the smallest average bit-rate.

Speech Coding and Recognition

This paper investigates the performance of a speech recognizer in an interactive voice response system for various coded speech signals, coded by using a vector quantization technique namely Multi Switched Split Vector Quantization Technique. The process of recognizing the coded output can be used in Voice banking application. The recognition technique used for the recognition of the coded speech signals is the Hidden Markov Model technique. The spectral distortion performance, computational complexity, and memory requirements of Multi Switched Split Vector Quantization Technique and the performance of the speech recognizer at various bit rates have been computed. From results it is found that the speech recognizer is showing better performance at 24 bits/frame and it is found that the percentage of recognition is being varied from 100% to 93.33% for various bit rates. Keywords—Linear predictive coding, Speech Recognition, Voice banking, Multi Switched Split Vector Quantization, Hidden Markov Model, Linear Predictive Coefficients.

Constrained Voicing-Based Codebook in Low-Rate Wideband CELP Coding

In this work we propose an efficient technique to quantize the immitance spectral frequency (ISF) in an algebraic code-excited linear prediction (ACELP) wideband codec. The Constrained Voicing-Based Vector Quantization (C-VBVQ) presented in this paper improves substantially the performance of the unconstrained-VBVQ approach for both clean and noisy speech. Both techniques reduce the codebook search time by almost one third. However, in the C-VBVQ training phase, the three codebooks that are individually designed for voiced, unvoiced and transition speech, are jointly reoptimized to impose a structural configuration. The proposed technique reduces the processing delay since it restricts the quantization of an input vector to only one smaller but optimal codebook. For each speech frame, one codebook is selected from the set of three codebooks based on the interframe correlation of the spectral information. The C-VBVQ was successfully implemented in an ACELP wideband coder. The objective and subjective performance are not only superior to that of the combination of the split vector quantization and multistage vector quantization but also to the unconstrained VBVQ.

Low-Complexity Wideband LSF Quantization Using Algebraic Trellis VQ

In this paper an algebraic trellis vector quantization (ATVQ) that introduces algebraic codebooks into trellis coded vector quantization (TCVQ) structure is presented. Low encoding complexity and minimum memory storage requirements are achieved using the proposed approach. It exploits advantages of both the TCVQ and the algebraic codebooks to know the delayed decision, the codebook widening, the low computational complexity and the no storage of codebook. This novel vector quantization scheme is used to encode the wideband speech line spectral frequencies (LSF) parameters. Experimental results on wideband speech have shown that ATVQ yields the same performance as the traditional split vector quantization (SVQ) and the TCVQ in terms of spectral distortion (SD). It can achieve a transparent quality at 47bits/frame with a considerable reduction of memory storage and computation complexity when compared to SVQ and TCVQ.

Reduced complexity two stage vector quantization

We address the issue of complexity for vector quantization (VQ) of wide-band speech LSF (line spectrum frequency) parameters. The recently proposed switched split VQ (SSVQ) method provides better rate-distortion (R/D) performance than the traditional split VQ (SVQ) method, even at the requirement of lower computational complexity, but at the expense of much higher memory. We develop the two stage SVQ (TsSVQ) method, by which we gain both the memory and computational advantages and still retain good R/D performance. The proposed TsSVQ method uses a full dimensional quantizer in its first stage for exploiting all the higher dimensional coding advantages and then, uses an SVQ method for quantizing the residual vector in the second stage so as to reduce the complexity. We also develop a transform domain residual coding method in this two stage architecture such that it further reduces the computational complexity. To design an effective residual codebook in the second stage, variance normalization of Voronoi regions is carried out which leads to the design of two new methods, referred to as normalized two stage SVQ (NTsSVQ) and normalized two stage transform domain SVQ (NTsTrSVQ). These two new methods have complimentary strengths and hence, they are combined in a switched VQ mode which leads to the further improvement in R/D performance, but retaining the low complexity requirement. We evaluate the performances of new methods for wide-band speech LSF parameter quantization and show their advantages over established SVQ and SSVQ methods.

Optimum switched split vector quantization of LSF parameters

We address the issue of rate–distortion (R/D) performance optimality of the recently proposed switched split vector quantization (SSVQ) method. The distribution of the source is modeled using Gaussian mixture density and thus, the non-parametric SSVQ is analyzed in a parametric model based framework for achieving optimum R/D performance. Using high rate quantization theory, we derive the optimum bit allocation formulae for the intra-cluster split vector quantizer (SVQ) and the inter-cluster switching. For the wide-band speech line spectrum frequency (LSF) parameter quantization, it is shown that the Gaussian mixture model (GMM) based optimum parametric SSVQ method provides 1 bit/vector advantage over the non-parametric SSVQ method.

Switched Conditional PDF-Based Split VQ Using Gaussian Mixture Model

In this letter, we develop switched conditional PDF-based split vector quantization (SCSVQ) method using the recently proposed conditional PDF-based split vector quantizer (CSVQ). The use of CSVQ allows us to alleviate the coding loss by exploiting the correlation between subvectors, in each switching region. Using the Gaussian mixture model (GMM)-based parametric framework, we also address the rate-distortion (R/D) performance optimality of the proposed SCSVQ method by allocating the bits optimally among the switching regions. For the wideband speech line spectrum frequency (LSF) parameter quantization, it is shown that the optimum parametric SCSVQ method provides nearly 2 bits/vector advantage over the recently proposed nonparametric switched split vector quantization (SSVQ) method.

Optimum Transform Domain Split VQ

In this letter, we develop an optimum transform domain split vector quantization (TrSVQ) method. We address both the issues of achieving best rate-distortion (R/D) performance and less complexity. For quantizing a multivariate Gaussian source, we derive the mean-square error (MSE) performance expression for the TrSVQ method using high rate theory and optimum bit allocation. Also, to reduce the complexity, we develop a binary split- based iterative algorithm and use the algorithm in a tree structured manner to find the optimum subvectors' dimensions (i.e., optimum splits).

Analysis of Conditional PDF-Based Split VQ

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> The split vector quantization (SVQ) method results in a “coding loss” due to the independent quantization of split subvectors. To recover the coding loss, conditional pdf-based split vector quantization methods were proposed recently. </para>

Conditional PDF-Based Split Vector Quantization of Wideband LSF Parameters

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> The commonly used split vector quantization (SVQ) method is inferior to unconstrained quantization due to independent coding of the split subvectors, resulting in a coding loss. In this paper, we propose a conditional pdf-based split vector quantization (CSVQ) method to recover the coding loss. The CSVQ method is developed assuming the line spectrum frequency source distribution as a multivariate Gaussian and the subvectors are quantized sequentially to exploit the correlation between the subvectors. The new CSVQ method is evaluated for wideband speech LSF quantization; CSVQ is shown to outperform traditional SVQ and provide comparable performance to the recently proposed switched split vector quantization (SSVQ) method. In addition, the transform domain SVQ method is also realized to show that its performance is limited by the distance measure used in the transform domain. </para>

Applying a Speaker-Dependent Speech Compression Technique to Concatenative TTS Synthesizers

This paper proposes a new speaker-dependent coding algorithm to efficiently compress a large speech database for corpus-based concatenative text-to-speech (TTS) engines while maintaining high fidelity. To achieve a high compression ratio and meet the fundamental requirements of concatenative TTS synthesizers, such as partial segment decoding and random access capability, we adopt a nonpredictive analysis-by-synthesis scheme for speaker-dependent parameter estimation and quantization. The spectral coefficients are quantized by using a memoryless split vector quantization (VQ) approach that does not use frame correlation. Considering that excitation signals of a specific speaker show low intra-variation especially in the voiced regions, the conventional adaptive codebook for pitch prediction is replaced by a speaker-dependent pitch-pulse codebook trained by a corpus of single-speaker speech signals. To further improve the coding efficiency, the proposed coder flexibly combines nonpredictive and predictive type method considering the structure of the TTS system. By applying the proposed algorithm to a Korean TTS system, we could obtain comparable quality to the G.729 speech coder and satisfy all the requirements that TTS system needs. The results are verified by both objective and subjective quality measurements. In addition, the decoding complexity of the proposed coder is around 55% lower than that of G.729 annex A

An Efficient Image Representation Technique Using Vector Quantization in Multiple Transform Domains

A novel technique for the efficient representation of still images is presented which employs vector quantization in multiple transform domains of the image signal. The system projects the subimages, obtained by partitioning the images used for training, into multiple transform domains. Energy-based split vector quantization is used to form code books in each of these domains. An adaptive algorithm to further optimize the accuracy of the code books in each transform domain is also developed. Simulations using sample imges show that this scheme provides improved reconstruction quality over existing methods for the same compression ratios, or equivalently, employing the proposed technique, fewer bits per pixel are used for the same reconstruction quality. This is achieved at the expense of increased computation at the encoder. The benefits from the improved representation efficiency of ten outweigh the increased computational complexity.

Quantization using frequency and mean compensated frequency input data for robust speech recogntion

A speech recognition system utilizes multiple quantizers to process frequency parameters and mean compensated frequency parameters derived from an input signal. The quantizers may be matrix and vector quantizer pairs, and such quantizer pairs may also function as front ends to a second stage speech classifiers such as hidden Markov models (HMMs) and/or utilizes neural network postprocessing to, for example, improve speech recognition performance. Mean compensating the frequency parameters can remove noise frequency components that remain approximately constant during the duration of the input signal. HMM initial state and state transition probabilities derived from common quantizer types and the same input signal may be consolidated to improve recognition system performance and efficiency. Matrix quantization exploits the “evolution” of the speech short-term spectral envelopes as well as frequency domain information, and vector quantization (VQ) primarily operates on frequency domain information. Time domain information may be substantially limited which may introduce error into the matrix quantization, and the VQ may provide error compensation. The matrix and vector quantizers may split spectral subbands to target selected frequencies for enhanced processing and may use fuzzy associations to develop fuzzy observation sequence data. A mixer may provide a variety of input data to the neural network for classification determination. Fuzzy operators may be utilized to reduce quantization error. Multiple codebooks may also be combined to form single respective codebooks for split matrix and split vector quantization to reduce processing resources demand.

Multicodebook split vector quantization of LSF parameters

A multicodebook quantization method for improving the performance of the split vector quantization (SVQ) is described. In the proposed method, multiple codebooks having different sizes are trained, and the minimal-size codebook and its codeword satisfying the condition based on the spectral distortion are determined. Several clustering techniques for reducing total bit rate and the analysis of the computational complexity are also presented. Experimental results have shown that the proposed method reduces the number of outliers significantly and achieves a better rate-distortion performance compared with the fixed-bit-rate SVQ, multistage VQ, and variable-bit-rate SVQ.

Matrix quantization with vector quantization error compensation and neural network postprocessing for robust speech recognition

A speech recognition system utilizes both matrix and vector quantizers as front ends to a second stage speech classifier such as hidden Markov models (HMMs) and utilizes neural network postprocessing to, for example, improve speech recognition performance. Matrix quantization exploits the “evolution” of the speech short-term spectral envelopes as well as frequency domain information, and vector quantization (VQ) primarily operates on frequency domain information. Time domain information may be substantially limited which may introduce error into the matrix quantization, and the VQ may provide error compensation. The matrix and vector quantizers may split spectral subbands to target selected frequencies for enhanced processing and may use fuzzy associations to develop fuzzy observation sequence data. A mixer provides a variety of input data to the neural network for classification determination. The neural network's ability to analyze the input data generally enhances recognition accuracy. Fuzzy operators may be utilized to reduce quantization error. Multiple codebooks may also be combined to form single respective codebooks for split matrix and split vector quantization to reduce processing resources demand.

Speech spectral quantizers for wideband speech coding

AbstractIn this treatise a range of Line Spectrum Frequency (LSF) Vector Quantization (VQ) schemes were studied comparatively, which were designed for wideband speech codecs. Both predictive arrangements and memoryless schemes were investigated. Specifically, both memoryless Split Vector Quantization (SVQ) and Classified Vector Quantization (CVQ) were studied. These techniques exhibit a low complexity and high channel error resilience, but require high bit rates for maintaining high speech quality. By contrast, Predictive Vector Quantizers (PVQ) offer an enhanced Spectral Distortion (SD) performance, although they are sensitive to channel error propagation. It is shown that the family of so—called Safety—Net Vector Quantization (SNVQ) schemes offers a good design compromise, providing an extension to memory—based PVQ, and thereby improving the performances both over noisy and noiseless channels.

Energy-Based Split Vector Quantizer Employing Signal Representation in Multiple Transform Domains

Mikhael, W., and Krishnan, V., Energy-Based Split Vector Quantizer Employing Signal Representation in Multiple Transform Domains, Digital Signal Processing11 (2001) 359–370Vector quantization schemes are widely used for waveform coding of one- and multidimensional signals. In this contribution, a novel energy-based, split vector quantization technique is presented, which represents digital signals efficiently as measured by the number of bits per sample for a predetermined signal reconstruction quality. In this approach, each signal vector is projected into multiple transform domains. In the learning mode, for a given transform domain representation, the transformed vector is split into subvectors (subbands) of equal average energy estimated from the transformed training vector ensemble. An equal number of bits is assigned to each subvector. A codebook is then designed for each equal energy subband of each transform domain representation. In the running mode, the coder selects codes from the domain that best represents the signal vector. The proposed multiple transform, split vector quantizer is developed and its performance is evaluated for both single-stage and multistage implementations. Several single transform vector quantizers for waveform coding exist, some of which employ energy-based bit allocation. Sample results using one-dimensional speech signals confirm the superior performance of the proposed scheme over existing single transform vector quantizers for waveform coding.