Vector quantization of harmonic magnitudes for low-rate speech coders

Lupini, P.; Cuperman, V.

doi:10.1109/glocom.1994.512716

Cited by 15 publications

(4 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the literature, some studies address the quantization of variable dimension vectors and propose some solutions adapted to each case of study. A non square transform vector quantization (NSTVQ) is proposed in [38], [39] to code harmonic amplitudes of the excitation in a LP codec: a non square linear transform is applied to the variable dimension vectors in order to obtain fixed length vectors which are then submitted to VQ. Another solution used for coding variable dimension harmonic amplitude vectors, and called Variable Dimension Vector Quantization (VDVQ), consists in designing a single universal fixed length codebook and using a binary selector vector that points on the non zero components of the harmonic amplitude vectors [40], [41].…”

Section: B Dimension Variability Of the Remaining Lt-dcm Vectorsmentioning

confidence: 99%

Low Bit-Rate Speech Codec Based on a Long-Term Harmonic Plus Noise Model

Ali¹,

Djaziri-Larbi²,

Girin³

2016

J. Audio Eng. Soc.

View full text Add to dashboard Cite

The long-term harmonic plus noise model (LT-HNM) for speech shows an interesting data compression, since it exploits the smooth evolution of the time trajectories of the short-term harmonic plus noise model parameters, by applying a discrete cosine model (DCM). In this paper, we extend the LT-HNM to a complete low bit-rate speech coder. A Normalized Split Vector Quantization (NSVQ) is proposed to quantize the variable dimension LT-DCM vectors. The NSVQ is designed according to the properties of the DCM vectors obtained from a standard speech database. The obtained LT-HNM coder reaches an average bit-rate of 2.7kbps for wideband speech. The proposed coder is evaluated in terms of modeling and coding errors, bit-rate, listening quality and intelligibility.

show abstract

Section: B Dimension Variability Of the Remaining Lt-dcm Vectorsmentioning

confidence: 99%

Low Bit-Rate Speech Codec Based on a Long-Term Harmonic Plus Noise Model

Ali¹,

Djaziri-Larbi²,

Girin³

2016

J. Audio Eng. Soc.

View full text Add to dashboard Cite

show abstract

“…The basic idea of this technique adopted from [4] is to first represent In order to find the optimal number of OCT coeffi cients per vector, both objective and subjective tests were carried out. The obj ective tests indicated that no number of coefficients can be considered optimal in the objective sense because increasing the number of coefficients lead to higher signal-to-noise ratio (SNR) at all bit rates in cluded in the test.…”

Section: Smoothed Magnitude Spectrum Representationmentioning

confidence: 99%

A novel quantization scheme for the noise-like component in waveform interpolation speech coding

Nurminen

Heikkinen

Saarinen

2002

IEEE International Conference on Acoustics Speech and Signal Processing

View full text Add to dashboard Cite

This paper presents a novel and efficient coding scheme for the noise-like component in wavefonn interpolation speech coding. The techniques employed in the task are described in detail and the advantages gained by using them are discussed. These techniques are shown to enable perceptually transparent coding of the magnitude spectra of the noise-like component with only ten bits per frame and to offer speech quality improvements at higher bit rates. The most significant features of the proposed scheme are smoothed representation and matrix quantiza tion of the magnitude spectra. The proposed coding tech niques can be used for enhancement of speech coders based on the wavefonn interpolation approach.

show abstract

“…where R means the set of code vectors in MBCB, composed of 2M code vectors, and the linear-scale code vector x, can be obtained from the mel-scale code vector c as (6). The optimal gain gc of the optimal code vector can be obtained as …”

Section: Predictive Quantizationmentioning

confidence: 99%

“…In variable dimension vector quantization (VDVQ), the spectral vector is quantized directly using a universal codebook of a fixeddimension, in which each element of the spectral vector is mapped onto a code vector using a selector [5]. In nonsquared transform vector quantization (NSTVQ), the input vector is transformed into fixed-dimension using a linear transform matrix [6]. By the way, the performance of the quantization can be improved perceptually by incorporating perceptual weighting to its distortion measure [7] [8].…”

Section: Introductionmentioning

confidence: 99%

Predictive and mel-scale binary vector quantization of variable dimension spectral magnitude

Cho

Kim²,

Kondoz³

2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100)

View full text Add to dashboard Cite

In sinusoidal speech coding, LP-spectral envelope is limited in its spectral accuracy if the order of LP-model is not high enough. Thus the quantization of residual spectrum of low order LP-model may be desirable for good quality speech reconstruction. From the investigation of the magnitude of the LP-residual spectrum, it is found that predictive coding scheme is useful to remove coding redundancy considerably. The problem of having variable number of harmonics due to pitch changes can be alleviated by a length warping technique. Subsequently, the residual spectrum of the predictive coding is represented by mel-scale binary vector quantizer (MBVQ), which quantizes the residual spectrum by splitting harmonic bands of variable dimension into fixed dimension, based on me1 scale, and representing each element of the code vector as binary value. The optimal code vector for the MBVQ can be derived by minimizing an error measure, defined as the weighted square-sum of the difference between original and synthesized spectral envelopes. From the performance evaluation, it is shown that the predictive-coded MBVQ with low order LP can obtain the effect of considerably high order LP-model. Additionally, the proposed method can be implemented with very low computational complexity in time and space.

show abstract

Vector quantization of harmonic magnitudes for low-rate speech coders

Cited by 15 publications

References 8 publications

Low Bit-Rate Speech Codec Based on a Long-Term Harmonic Plus Noise Model

Low Bit-Rate Speech Codec Based on a Long-Term Harmonic Plus Noise Model

A novel quantization scheme for the noise-like component in waveform interpolation speech coding

Predictive and mel-scale binary vector quantization of variable dimension spectral magnitude

Contact Info

Product

Resources

About