Instrumental Estimation of E-Model Parameters for Wideband Speech Codecs

Möller, Sebastian; Côté, Nathalie; Gautier-Turbin, Valérie; Kitawaki, Nobuhiko; Takahashi, Akihiro

doi:10.1155/2010/782731

Cited by 10 publications

(5 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These will be further discussed in Section 12. The E-model can determine the effect of packet loss on voice quality if the RTP/UTP voice-over-Internet protocol (VoIP) is used in digital networks [85] Table 1.…”

Section: Evaluation Criteria For Speech Codingmentioning

confidence: 99%

Review of methods for coding of speech signals

O’Shaughnessy

2023

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

Speech is the most common form of human communication, and many conversations use digital communication links. For efficient transmission, acoustic speech waveforms are usually converted to digital form, with reduced bit rates, while maintaining decoded speech quality. This paper reviews the history of speech coding techniques, from early mu-law logarithmic compression to recent neural-network methods. The techniques are examined in terms of output quality, algorithmic complexity, delay, and cost. Focus is on which aspects of speech can be exploited for high-quality transmission. The choices made to code speech are motivated by efficiency, the needs of applications, and access to information in the speech signal that is useful for both intelligibility and naturalness in the reconstructed speech at the decoder.

show abstract

Section: Evaluation Criteria For Speech Codingmentioning

confidence: 99%

Review of methods for coding of speech signals

O’Shaughnessy

2023

J AUDIO SPEECH MUSIC PROC.

View full text Add to dashboard Cite

show abstract

“…Currently, there are only ITU standard narrowband codecs, along with a few non-ITU narrowband codecs that have been widely tested and their impairment factors standardized. Additionally, the wideband extension of E-Model [21] requires that the packet-loss robustness factor (Bpl) corresponds to the codec, packet size and packet loss concealment (PLC) used. For our purposes, as the PLC technique can vary in the Opus codec, we used the previously detailed method of inserting an empty packet.…”

Section: Opus Mos Fittingmentioning

confidence: 99%

Determination of bit-rate adaptation thresholds for the Opus codec for VoIP services

Han¹,

Magoni²,

McDonagh³

et al. 2014

2014 IEEE Symposium on Computers and Communications (ISCC)

View full text Add to dashboard Cite

In this paper, we present an experimental evaluation of the recently standardized Opus codec used in a VoIP context. Opus operates in both narrow and wideband modes, similar to Adaptive Multi-Rate (AMR). Through the use of the Wideband Perceptual Evaluation of Speech Quality (WB-PESQ) metric, we have conducted an extensive set of experiments using multiple audio samples encoded at different bit-rates, to investigate the impact of packet loss on resulting speech quality. Using these results, fitting functions for each bit-rate were computed to provide a straightforward manner of evaluating speech quality when given a specified packet loss rate. Using ns-2, a simulation analysis was conducted to evaluate the effect of background traffic on transmitted Opus streams. We observed that, when using different levels of background traffic, the observed packet loss rates varied heavily depending on the stream bit-rate. By correlating this information with the fitting functions derived previously, we were able to define switching thresholds. These are points where the speech quality of a lower bit-rate stream is greater than that of a higher bit-rate stream for the same levels of link bandwidth saturation.

show abstract

“…In a recent study [9], it has been shown that I e -values can be predicted with signal-based models achieving a high correlation (e.g., with WB-PESQ, a correlation of r = 0.85 is achievable). Then, provided that I bw is given, I res can be determined via Eq.…”

Section: Instrumental Estimation Of the Residual Componentmentioning

confidence: 99%

Towards a new E-Model impairment factor for linear distortion of narrowband and wideband speech transmission

Wältermann

Raake

2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

The E-Model, a tool for network planning recommended by the ITU-T, suffers from the lack of predicting linear distortions as they may occur from channel filtering, codecs, and user interfaces. In order to face this deficiency, a new impairment factor is introduced in this paper which is estimated on the basis of two simple parameters from the linear portion of a system. Therewith, conventional Equipment Impairment Factors of narrowband and wideband speech codecs are decomposed into a linear and a residual part, allowing to quantify both magnitudes separately on the so-called R-scale. By extending the concept of distortion classes in the E-Model, the proposed scheme provides a plausible picture of the linear effect of transmission systems. The advantage of wideband causing a 36% quality gain is well reflected. Further, the decomposition leads to a reduction of error when impairment factors are added on the R-scale. Examples for instrumentally estimating the residual impairment are given.

show abstract

Instrumental Estimation of E-Model Parameters for Wideband Speech Codecs

Cited by 10 publications

References 3 publications

Review of methods for coding of speech signals

Review of methods for coding of speech signals

Determination of bit-rate adaptation thresholds for the Opus codec for VoIP services

Towards a new E-Model impairment factor for linear distortion of narrowband and wideband speech transmission

Contact Info

Product

Resources

About