Changchun Bao

A harmonic bandwidth extension based on Gaussian mixture model

For conventional bandwidth extension, the spectral patching methods, such as spectral folding, spectral translation and non-linear processing, are employed to reconstruct high frequency signal, yet it leads to the spectral shifting between reconstructed and original signal, and does not retain the original harmonic relations. In this paper, a blind harmonic bandwidth extension method from wideband to super-wideband was proposed by estimating the energy of high frequency spectral envelope with Gaussian mixture model (GMM). Both the objective and subjective test results show that proposed algorithm performs better than conventional blind bandwidth extension algorithms.

Pitch detection method for noisy speech signals based on pre-filter and weighted wavelet coefficients

Most of the current pitch detection algorithms can not work well under the high noise environment. For this reason, a pitch detection algorithm for noisy speech signal based on pre-filtering and weighted wavelet coefficients is proposed. Firstly, the noisy speech signals are pre-filtered. Secondly, the speech pre-filtered is decomposed by the quadratic spline wavelet. Thirdly, the wavelet coefficients of three consecutive scales are weighted to emphasize the sharp change points. Fourthly, three candidate pitch periods are extracted from the weighted signals. Finally, the pitch period is calculated by autocorrelation function. Experiments show that this algorithm can increase the performance of pitch detection in noisy environment and decreases computational complexity compared with DWT-NCCF method.

A voice activity detection using cyclic statistics based on sinusoidal speech model

In this paper, by using the cyclostationary properties of speech signal, a voice activity detection (VAD) algorithm based on cyclic cumulant is proposed. The proposed scheme employs the third-order cyclic cumulant of the LPC residual of a speech signal. Analytical expressions for the third-order cyclic cumulant of the LPC residual of short-term speech are derived assuming a sinusoidal model. Matrix pencil method (MP) are adopted to estimate the frequencies of harmonic signal contained in LPC residual of speech signal, which are used as the cyclic frequencies of cyclic cumulant. Then the third-order cyclic cumulant is defined and used to construct the VAD detection variation. The test results show that the proposed algorithm gives better results than G.729B VAD.

Projective non-negative matrix factorization with Bregman divergence for musical instrument classification

In this paper, the projective non-negative matrix factorization (PNMF) with Bregman divergence is applied into the musical instrument classification. A novel supervised learning algorithm for automatic classification of individual musical instrument sounds is addressed inspiring from PNMF with several versions of Bregman divergence. Moreover, the orthogonality of basis matrices between PNMF and conventional non-negative matrix factorization (NMF) is compared. In addition, three classifiers based on nearest neighbors (NN), Gaussian mixture model (GMM) and radial basis function (RBF) are added to evaluate the performance of PNMF classifier. The results indicate that the classification accuracy of the proposed PNMF classifier outperforms the classifiers derived from conventional NMF and machine learning.

1kb/s waveform interpolation speech coding based on non-negative matrix factorization

A new 1kb/s waveform interpolation (WI) speech coding algorithm based on non-negative matrix factorization (NMF) is proposed and implemented in this paper. Multi-frame parameter joint vector or matrix quantization, parameter prediction and discrete cosine transform (DCT) are used to reduce the bit rate and to improve the quality of speech. The results of informal subjective listening test shows that the intelligibility and articulation of the algorithm are close to that of 2kb/s NMF-based WI coder (called ldquoNMF-WIrdquo as convenience).

An improvement of frame erasure concealment for G.722.1 coding algorithm

This paper proposes an efficient method for frame erasure concealment in G.722.1 coding algorithm, which can mitigate the adverse impact of frame erasure on the reconstruction quality. The lost frame is reconstructed in the modulated lapped transform (MLT) domain in terms of magnitude and sign of the coefficients. The method of interpolation is employed for the magnitude estimation. Some sign information is recovered by extra information from encoder. Listening test results indicate that improved quality is produced by the proposed algorithm.

A novel embedded speech and audio codec based on ITU-T Recommendation G.722.1

This paper describes an embedded speech and audio codec which is based on ITU-T Recommendation G.722.1; it can process 7 kHz bandwidth speech and audio signal at scalable bit rates. Based on the G.722.1 of ITU-T, this algorithm adds two modules: the energy ordering of sub-band and the processing of bit-stream truncation. Furthermore, it does some modification on the categorization and noise-fill modules. It makes sure that the codec could produce embedded bit-stream, so this codec had more robustness in the transmission. The test results by ITU-T PESQ show that this codec has good performance as G.722.1 at the same bit-rates.