Transmission of convolutionally encoded multiple descriptions over noisy channels can benefit from the use of iterative source-channel decoding. The authors first modified the BCJR algorithm in a way that symbol a posteriori probabilities can be derived and used as extrinsic information to improve the iterative decoding between the source and channel decoders. The authors also formulate a recursive implementation for the source decoder that processes reliability information received on different channels and combines them with inter-description correlation to estimate the transmitted quantiser index. Simulation results are presented for two-channel scalar quantisation of Gauss–Markov sources which demonstrate the error-resilience capabilities of symbol-based iterative decoding.

In this work, we attempt to capitalize more fully on the source residual redundancy and then develop an MD-ISCD scheme which permits to exchange between its two constituent decoders the whole symbol extrinsic information. The first step toward realization is to derive the modified BCJR algorithm based on sectionalized code trellises that provides reliability information on each transmitted symbol rather than on index-bits. To further reduce the computation, we also apply the concept of Jacobian logarithm to formulate the algorithm in the logarithmic domain.

This paper studies the application of voice conversion to hearing-impaired Mandarin speech enhancement. The system is based on the combined use of a sinusoidal analysis-synthesis model and a priori knowledge about Mandarin syllable phonetic structures. We propose a time-scale modification algorithm that finds accurate alignments between hearing-impaired and normal utterances. Using the alignments, spectral conversion is performed by a continuous probabilistic transform based on a Gaussian mixture model. Simulation results indicate that the proposed system can improve the intelligibility of hearing-impaired Mandarin speech.

This study focuses on two issues: parametric modeling of the channel and index assignment of codevectors, to design a vector quantizer that achieves high robustness against channel errors. We first formulate the design of a robust zero-redundancy vector quantizer as a combinatorial optimization problem leading to a genetic search for a minimum-distortion index assignment. The performance is further enhanced by the use of the Fritchman (1967) channel model that more closely characterizes the statistical dependencies between error sequences. This study also presents an index assignment algorithm based on the Fritchman model with parameter values estimated using a real-coded genetic algorithm. Simulation results indicate that the global explorative properties of genetic algorithms make them very effective in estimating Fritchman model parameters, and use of this model can match index assignment to expected channel conditions.

Speech repairs occur often in spontaneous spoken dialogues. The ability to detect and correct those repairs is necessary for every spoken language system. We present a framework to detect and correct speech repairs where all relevant levels of information, i.e., acoustics, lexic, syntax and semantics could be integrated. The basic idea is to reduce the search space for repairs as soon as possible by cascading filters that involve more and more features. At first an acoustic module generates hypotheses about the existence of a repair. Second a stochastic model suggests a correction for every hypothesis. Well scored corrections are inserted as new paths in the word lattice. A lattice parser then makes the final decision about accepting the repair. 1. INTRODUCTION Human utterances often contain erroneous parts where speakers correct previous words in their speech. The erroneous portions are not part of the intended word sequence. A speech system must be able to deal with such phenomena ...

In this paper we develop hidden Markov model (HMM) based approaches to identify Chinese dialects spoken in Taiwan. This task can be aided by exploiting various characteristic features of Chinese spoken languages. The baseline system performs phonotactic analysis after the speech utterance is tokenized into a sequence of five broad phonetic classes. The sequential statistics of the resulting symbols are then used to distinguish one dialect from another. The second approach we tested is to incorporate dialect-dependent phonotactic constraints into the phonetic tokenization rather than applying these constraints after the broad phonetic classification is complete. These algorithms were evaluated using a multispeaker speech corpus of text-independent spontaneous speech data. Simulation results indicate that the acousticphonotactic approach to dialect identification yields better performance with an average identification rate of 89.6%, compared to 70% for the baseline system.

This paper presents quality enhancement of sinusoidal transform coders (STC) via the development of new parametric models. First explored are the benefits of the Bark spectrum for use in the design of perceptual coding of the sine-wave amplitudes. According to our results, the proposed approach provides a uniform perceptual fit across the spectrum. To enhance the accuracy of phase representation, noncausal all-pole modeling of the vocal system is also discussed. Experimental results indicate that the use of new parametric models allows the STC to improve the phase accuracy as well as the synthetic speech quality.

Most LPC-based audio coders employ simplistic noise-shaping operations to perform psychoacoustic control of quantization noise. In this paper, we report on new approaches to exploiting perceptual masking in the design of adaptive quantization of LPC excitation parameters. Due to its localized spectral sensitivity, sinusoidal excitation representation is preferred to spectrally flat signals for use in excitation modeling. Simulation results indicate that the proposed multisinusoid excited coder can deliver high quality audio reproduction at the rate of 72 kb/s.

Errors encountered in digital mobile radio channels are not independent but exhibit varying degrees of statistical dependencies. In many mobile applications, Gilbert's (1960) finite-state Markov chain model has been shown to adequately characterize the bursty-noise binary channels. An estimation method based on the simulated annealing algorithm is proposed for computing the model. Parameters from the sample error sequences. Experimental results indicate that the simulated annealing method yields more accurate identification at a faster convergence rate for comparison with the gradient iterative method.

Most LPC-based audio coders improve the reproduction quality by using predictor coefficients to embody perceptual masking in noise spectral shaping. Since the predictor coefficients were originally derived to characterize sound production models, they cannot precisely describe the human ear's nonlinear responses to frequency and loudness. We report on new approaches to exploiting the masking threshold in the design of a perceptual noise-weighting filter for excitation searches. To track the nonstationary evolution of a masking threshold, an autoregressive spectral analysis with finite order has been shown to be capable of providing sufficient accuracy. In seeking a faster response, an artificial neural network was also trained to extract autoregressive modeling parameters of the masking threshold from typical audio signals via mapping. Furthermore, we propose the concept of sinusoidal excitation representation to better track the intrinsic characteristics of prediction error signals. Simulation results indicate that the combined use of a multisinusoid excitation model and a masking-threshold-adapted weighting filter allows the implementation of an LPC-based audio coder that delivers near transparent quality at the rate of 96 kb/s.

Interpolative differential pulse code modulation (IDPCM) is known to outperform DPCM at rate 1 b/sample for several synthetic source models. The authors use minimum-mean-squared error (MMSE) fixed-lag smoothing with DPCM to develop a code generator using delayed decoding. This smoothed DPCM (SDPCM) code generator is compared to DPCM and IDPCM code generators at rates 1 and 2 b/sample for tree coding several synthetic sources and to a DPCM code generator at 2 b/sample for speech sources.>

The authors present both forward and backward adaptive speech coders that operate at 9.6, 12, and 16 kb/s using integer and fractional rate trees, weighted squared error distortion measures, the (M,L) tree search algorithm, and incremental path map symbol release. They introduce the concept of multitree source codes and illustrate how the multitree structure allows scalar quantizer-based codes and scalar adaptation rules to be used for fractional rate tree coding. With a frequency weighted distortion measure, the forward and backward adaptive multitree coders produce near toll quality speech at 16 kb/s, while the backward adaptive 9.6 kb/s multitree coder substantially outperforms adaptive predictive coding and has an encoding delay of less than 2 ms. Performance results are present in terms of unweighted and weighted signal-to-noise ratio and segmental signal-to-noise ratio, sound spectrograms, and subjective listening tests.>

Simulation results are presented comparing the performances of four backward adaptive lattice algorithms for updating the short-term predictors in adaptive predictive coding (APC) and differential pulse code modulation (DPCM) code generators for low-delay tree coding of speech at 16 and 9.6 kb/s. The algorithms studied are the least-squares lattice, the exponential window lattice, the signal-driven lattice, and the residual-driven lattice. Ideal channels are emphasized, and comparisons are based primarily upon frequency-weighted signal-to-noise ratio and subjective listening tests.>

To design a tree coder for source coding with a fidelity criterion, one must choose a suitable code generator, an efficient tree search algorithm, an appropriate distortion measure, and a path map symbol release rule. The performance of several path map symbol release rules when used with exhaustive searching of the exponential metric tree is investigated. The average single-letter distortion of fixed-length symbol release rules and two variable-length symbol release rules are derived for shallow search depths and compared to simulation results. The incremental or single-symbol release rule is shown to yield the best performance.>