Optimal Coding of Generalized-Gaussian-Distributed Frequency Spectra for Low-Delay Audio Coder With Powered All-Pole Spectrum Estimation

Sugiura, Ryo; Kamamoto, Yutaka; Harada, Noboru; Kameoka, Hirokazu; Moriya, Takehiro

doi:10.1109/taslp.2015.2431851

Cited by 9 publications

(9 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our previous work on GGDs [39], [40] revealed that, with appropriate spectral envelope estimation, we can improve compression efficiency by switching shape parameter α frame by frame depending on the acoustic features of the input signals. That work presented some signal models based on the assumptions of GGDs and there was no reasonable choice to combine communication models other than arithmetic coding.…”

Section: B Generalized Gaussian Distributionsmentioning

confidence: 99%

Shape Control of Discrete Generalized Gaussian Distributions for Frequency-Domain Audio Coding

Sugiura

Kamamoto

Moriya

2019

IEEE/ACM Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

Entropy coding, which is an essential part of audio compression, is always required to manage the tradeoffs between compression efficiency and computational complexity, and the strategy to achieve them highly depends on the distributions of inputs. In this paper, we present a method of controlling them for enhancing the compression efficiency of Golomb-Rice (GR) encoding, one of the simplest entropy coding methods optimal for Laplacian distributions. We will show that the proposed invertible and low-complexity mapping of integers enables the GR encoding to assign nearly the optimal code length for a wider range of distributions, generalized Gaussian distributions, maintaining low computational cost. A simulation by random numbers reveals that the proposed coder based on this scheme works about 6 times faster than the state-of-the-art arithmetic coder for Gaussian-distributed integers maintaining the increase in relative redundancy around 2.6%, which is much lower than that of a conventional GR coder. Additionally, an application to a practical speech and audio coding scheme is presented, and an objective evaluation for real speech and audio signals confirms the advantages of the proposed method in compression. The method is expected to widen the capability of low-complexity entropy coding, providing us with more flexible codec designs.

show abstract

Section: B Generalized Gaussian Distributionsmentioning

confidence: 99%

Shape Control of Discrete Generalized Gaussian Distributions for Frequency-Domain Audio Coding

Sugiura

Kamamoto

Moriya

2019

IEEE/ACM Trans. Audio Speech Lang. Process.

Self Cite

View full text Add to dashboard Cite

show abstract

“…To guarantee the optimality of the envelope with low computational costs, we extended LPC in [11] using a powered model as…”

Section: I P a P S E F O R T C Xmentioning

confidence: 99%

“…To guarantee the optimality of the envelope with low computational costs, we extended LPC in [11] using a powered model as which leads to the code length of a frame of quantized MDCT coefficients as written in Itakura–Saito (IS) divergence . The minimization of this code length can be solved just as LPC by regarding α-th-powered spectra as power spectra since the algorithm gives coefficients that minimize the IS divergence between the squared LPC envelope and power spectra.…”

Section: Papse For Tcxmentioning

confidence: 99%

“…The second is frequency-resolution warping of the LPC envelope [8,9]. The third is powered all-pole spectrum estimation (PAPSE) [10,11], for LPC modeling of the envelope, which takes the distribution of quantized MDCT coefficients into account. All tools are intended to enhance the efficiency of entropy coding of quantized MDCT coefficients.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Progress in LPC-based frequency-domain audio coding

Moriya

Sugiura

Kamamoto

et al. 2016

SIP

Self Cite

View full text Add to dashboard Cite

show abstract

“…For example, in coding of the short-time spectra of speech signals, the Laplacian has been shown to be both accurate, but also computationally simple to implement [5,6]. Similarly, compensating for the mismatch between Gaussian and Laplacian distributions can be used to improve performance in speaker recognition [7].…”

Section: Introductionmentioning

confidence: 99%

Estimation of the Probability Distribution of Spectral Fine Structure in the Speech Source

Bäckström

2017

Interspeech 2017

View full text Add to dashboard Cite

The efficiency of many speech processing methods rely on accurate modeling of the distribution of the signal spectrum and a majority of prior works suggest that the spectral components follow the Laplace distribution. To improve the probability distribution models based on our knowledge of speech source modeling, we argue that the model should in fact be a multiplicative mixture model, including terms for voiced and unvoiced utterances. While prior works have applied Gaussian mixture models, we demonstrate that a mixture of generalized Gaussian models more accurately follows the observations. The proposed estimation method is based on measuring the ratio of Lp-norms between spectral bands. Such ratios follow the Beta-distribution when the input signal is generalized Gaussian, whereby the estimated parameters can be used to determine the underlying parameters of the mixture of generalized Gaussian distributions.

show abstract

Optimal Coding of Generalized-Gaussian-Distributed Frequency Spectra for Low-Delay Audio Coder With Powered All-Pole Spectrum Estimation

Cited by 9 publications

References 29 publications

Shape Control of Discrete Generalized Gaussian Distributions for Frequency-Domain Audio Coding

Shape Control of Discrete Generalized Gaussian Distributions for Frequency-Domain Audio Coding

Progress in LPC-based frequency-domain audio coding

Estimation of the Probability Distribution of Spectral Fine Structure in the Speech Source

Contact Info

Product

Resources

About