In this paper, we propose a new approach for the subjective assessment of the loudness of complex audio signals such as speech or music. This two-stage approach makes it possible to study the influence on loudness of the frequency bandwidth and of different kinds of codecs. In the first stage, the individual loudness function of each subject is estimated using a specific 100-point response scale. In the second stage, the subject evaluates the loudness of each processed sample, by filtering or coding/decoding, using the same scale. The loudness obtained in terms of points is then converted in loudness levels in terms of phons using the estimated individual loudness function. Results show that loudness increases with the bandwidth extension up to super-wideband. Similar behavior is observed when codecs are applied.

For subjective assessment of speech quality in codecs, a reference impairment system is required to introduce controlled degradations to calibrate subjective evaluation. A reference system provides a convenient means for making meaningful comparisons between subjective test results across laboratories and can be viewed as a scale on which mean opinion scores are projected, this scale being supposed to cover the whole range of quality. Nowadays, standardized anchor systems do not fit any more the degradations brought by the present codecs. This paper aims at offering new reference signals simulating the defaults of codecs currently used on telecommunication networks. Twenty wideband codecs are compared through dissimilarity tests. A multidimensional scaling technique allows us to define a four-dimensional perceptive space that appears stable for male and female talkers. A verbalization task suggests qualifying the degradations perceived by the listeners with the following attributes: muffle, background noise, noise on speech, and hiss, each conveyed by one dimension. These dimensions are correlated with objective measures such as spectral centroid, energy in the silent part in the high frequency sub-band, ratio of brightness between deterministic part and residual part of the signal and spectral correlation coefficient. New reference signals are produced and a phase of validation suggests a perceptive space quite coherent with the original one.

The introduction of packet switched (PS) networks (e.g., IMS) creates a need to evaluate speech transmission quality when using the 3GPP default speech codecs. 3GPP SA4 has created a work item in 3GPP Release 6 on "Performance characterization of default codecs for PS conversational application". France Telecom R&D and Siemens proposed a test framework consisting of a UMTS simulator for the air interface and an IP network simulator. ITU-T recommendation P.800 (1996) is used for the quality estimation of the transmission. The real-time conversational test results show that the AMR-NB (adaptive multirate narrow band) and AMR-WB (AMR wide band) speech codecs are well suited for PS conversational applications. Furthermore, the results clearly show a higher understanding when using AMR-WB rather than AMR-NB.

This paper describes an adaptive multi-rate wideband (AMR-WB) speech codec proposed for the GSM system and also for the evolving third generation (3G) mobile speech services. The speech codec is based on SB-CELP (splitband-code-excited linear prediction) with five modes operating bit rates from 24 kbit/s down to 9.1 kbit/s. The respective channel coding schemes are based on RSC (recursive systematic code) and UEP (unequal error protection). Both, source and channel codec are designed as homogenous as possible to guarantee robust transmission on current and future mobile radio channels.

This paper describes a combined adaptive transform codec (ATC) and code-excited linear prediction (CELP) algorithm, called ATCELP, for the compression of wideband (7 kHz) signals. The CELP algorithm applies mainly to speech, whereas the ATC mode is selected for music and noise signals. We propose a switching scheme between CELP and ATC mode and describe a frame erasure concealment technique. Subjective listening tests have shown that the ATCELP codec at bit rates of 16, 24 and 32 kbit/s achieved performances close to those of the CCITT G.722 at 48, 56 and 64 kbit/s, respectively, at most operating conditions.

This paper emphasizes the importance of the question of word-order variation in connectionist language modeling. More precisely, it investigates whether recurrent networks can integrate various linguistic constraints to process variable word-order languages. This paper reports three experiments on the understanding of spoken French that suggest that recurrent architectures could apply to the understanding of variable spoken languages.

In this paper, an algorithm for coding 20 Hz -15 kHz speech signals at 64 kbit/s with a very low delay (frame of 0.16 ms) is presented.To achieve a quality near to transparency, we propose adapting the Low-Delay CELP coder [1] to the 15 kHz bandwidth and suggest a new noise shaping method based on a psychoacoustic model.In this way we take advantage of linear predictive coding and masking properties of the human perception system.Finally, an algebraic codebook is proposed, allowing an important reduction of coder computational complexity, without decreasing the perceived quality of signals.