The temporal characteristics of human inductive strength judgment process have not been previously investigated, although some preliminary spatial localization results have been reported. In the present study, some Ch...
详细信息
The temporal characteristics of human inductive strength judgment process have not been previously investigated, although some preliminary spatial localization results have been reported. In the present study, some Chinese verbal inductive reasoning tasks are used in ERP (event related potentials) experiments to explore the time course of inductive strength judgment. The experimental results confirm our expectation. Inductive strength judgment after the presentation of a conclusion mainly contains three stages: visual encoding, semantic information integration, and strength evaluation. It can be tentatively concluded that visual encoding may be observed at the frontal P200 and the posterior N200, and the frontal LNC and the posterior LPC may reflect semantic information integration, and the slow waves after about 650ms may relate to strength evaluation process.
This paper presents the audio noise classification using Bark scale features and K-NN technique. This paper uses audio noise signal from NOISEX-92 (12 types). We determine the transfer functions from linearpredictive...
详细信息
This paper presents the audio noise classification using Bark scale features and K-NN technique. This paper uses audio noise signal from NOISEX-92 (12 types). We determine the transfer functions from linear predictive coding (LPC) coefficient of noise signal on Bark scale and use K-NN technique to classify them. The results will be used for optimization of speech recognition model in the presence of noise. The highest average accuracy for audio noise classification is obtained when K=3 and median over 5 consecutive frames.
We present a new speech coding algorithm, based on an all-pole model of the vocal tract. Whereas current autoregressive (AR) based modeling techniques (e.g. CELP, LPC-10) minimize a prediction error, which is consider...
详细信息
We present a new speech coding algorithm, based on an all-pole model of the vocal tract. Whereas current autoregressive (AR) based modeling techniques (e.g. CELP, LPC-10) minimize a prediction error, which is considered to be the input to the all-pole model, our approach determines the closest (in L/sub 2/ norm) signal, which exactly satisfies an all-pole model. Each frame is then encoded by storing the parameters of the complex damped exponentials deduced from the all-pole model and its initial conditions. Decoding is performed by adding the complex damped exponentials based on the transmitted parameters. The new algorithm is demonstrated on a speech signal. The quality is compared with that of a standard coding algorithm at comparable compression ratios, by using the segmental signal-to-noise ratio (SNR).
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 repre...
详细信息
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.
Here we consider the problem of providing near optimal performance for a large set of possible models. We adopt the LQR framework in the single-input single-output (SISO) setting, and prove that given a compact set of...
详细信息
Here we consider the problem of providing near optimal performance for a large set of possible models. We adopt the LQR framework in the single-input single-output (SISO) setting, and prove that given a compact set of controllable and observable plant models of a fixed order, we can construct a single linear periodic controller (LPC) which provides near optimal LQR performance.
Traditionally, linear Prediction is used to predict future values of a signal using past values. The goal is to minimize prediction errors. In this paper, we propose a novel method of utilizing prediction errors to ex...
详细信息
Traditionally, linear Prediction is used to predict future values of a signal using past values. The goal is to minimize prediction errors. In this paper, we propose a novel method of utilizing prediction errors to extract edges of images. In this method, smooth prediction errors are minimized while steep changes (larger errors) are amplified. Therefore, when applied to image edge detection, edge information can be accurately extracted. The proposed method is compared with predominant methods such as Sobel and Canny methods. While there is no mathematical proof that the proposed method outperforms predominant methods, however, examples presented in this paper may suggest that the proposed method may perform better for certain applications.
This paper describes our enhanced mixed excitation linear prediction (MELP) speech coder which is a candidate for the new U.S. Federal Standard at 2.4 kbits/s. The new coder is based on the MELP model, and it uses a n...
详细信息
This paper describes our enhanced mixed excitation linear prediction (MELP) speech coder which is a candidate for the new U.S. Federal Standard at 2.4 kbits/s. The new coder is based on the MELP model, and it uses a number of enhancements as well as efficient quantization algorithms to improve performance while maintaining a low bit rate. In addition, the coder has been optimized for performance in acoustic background noise and in channel errors, as well as for efficient real-time implementation. Listening tests confirm that the enhanced 2.4 kbit/s MELP coder performs as well as the higher bit rate 4.8 kbit/s FS1016 CELP standard.
This paper describes an implementation of MELP (mixed excitation linear prediction) vocoder. Subband division required for implementing the MELP vocoder was performed by the lifting wavelet transform. A new method to ...
详细信息
This paper describes an implementation of MELP (mixed excitation linear prediction) vocoder. Subband division required for implementing the MELP vocoder was performed by the lifting wavelet transform. A new method to generate an appropriate glottal waveform was devised. In addition, three kinds of fluctuations observed in the steady parts of voiced speech were incorporated to enhance the naturalness of synthesized speech.
The authors describe a low-complexity coding technique that combines multipulse and stochastic excitation. The system, known as hybrid multipulse coding (HMC), provides good quality at 4.8 and 7.2 kb/s. HMC uses effic...
详细信息
The authors describe a low-complexity coding technique that combines multipulse and stochastic excitation. The system, known as hybrid multipulse coding (HMC), provides good quality at 4.8 and 7.2 kb/s. HMC uses efficient pulse excitation for voiced speech and stochastic excitation for unvoiced speech. For the best speech quality, HMC uses an optimization algorithm for simultaneous solution of pitch predictor and excitation parameters, producing a higher signal-to-noise ratio than CELP (code-excited linear prediction) in the 4.8-kb/s configuration. At 7.2 kb/sec, CELP and HMC give diagnostic acceptability measure (DAM) scores close enough so that their standard error limits overlap. In all configurations, HMC requires from 5 to 14 times fewer multiply/accumulate operations than CELP for similarly sized codebooks.< >
A set of experiments in which the LVQ2 (learning vector quantization) algorithm of T. Kohonen et al (Proc. 1988 IEE Int. Conf. on Neural Networks, p.I-61-68, 1988) is used to generate vector codebooks for a discrete-o...
详细信息
A set of experiments in which the LVQ2 (learning vector quantization) algorithm of T. Kohonen et al (Proc. 1988 IEE Int. Conf. on Neural Networks, p.I-61-68, 1988) is used to generate vector codebooks for a discrete-observation hidden Markov model (HMM) classifier is described. Input feature vectors consist of single-frame linear predictive coding (LPC)-based cepstra and/or differenced cepstra. Classification accuracies using conventional k-means, class-specific k-means, and LVQ2 codebooks are compared for a 16-way speaker-independent vowel classification task. In contrast to speaker-dependent phonetic classification results previously published, no significant performance advantages are observed with LVQ2. These conflicting results are discussed relative to differences in the recognition tasks and the feature sets used. It is also argued that the single-observation Bayesian decision boundaries approximated by LVQ2 are nonoptimal for HMM-based classification involving multiple observations.< >
暂无评论