Speech is easier to understand in continuous noise than in noise which is switched on at the beginning of the speech and off at the end. It is suggested that this is due to some adaptation process. In order to test this hypothesis a series of experiments have been performed in which the intelligibility of plosives in isolated syllables was measured as a function of the duration of the preceding noise. The spectrum of the noise was also varied. It was found that the adaptation, as measured by the mean increase in intelligibility, increased as the duration of the noise preceding the syllable was lengthened. It was also found that the adaptation varied with the centre frequency of the spectrum of the noise. The amount of adaptation was negatively correlated with the threshold of hearing.

Long-term envelope modulations ( 100Hz) influence the identification of speech in noise. It is not clear, however, whether this influence only takes place at the level of acoustic-phonetic analysis (phonetic identification) or if envelope fluctuations may also help in auditory figureground segregation (e.g. separation of speech from concurrent backgrounds). An experiment is presented in which the influence of long-term envelope modulations was investigated using signals mixed with either stationary or temporally modulated noise. The better performance observed when processing speech in modulated background may be related to the listeners’ ability to use envelope information in trying to follow concurrent signals independently. It is therefore predicted that, if longterm envelope modulations help to segregate speech from noisy backgrounds, this effect should be stronger when envelope information is fully available.

After having given the reasons why it is legitimate to dedicate a special attention to the oral code with regard to the written code, we propose an approach to the automatic generation of French phonetic inflected forms. We shall examine it first inside the general context of automatic phoneticization, with a special attention to the phoneticization of loanwords. We shall then give the methodology we adopted to build up our generation system.

An improvement of the speech spectrogram based on the method of reassignment is presented. This method consists of moving each point of the spectrogram to a new point that represents the distribution of the energy in the time-frequency window more accurately. Examples of natural speech show an improvement of the energy localization in both time and frequency domains. This allows a better description of speech features.

A model that is able to predict human performance in a simultaneous glide recognition task is described. The model combines a primitive, F0 guided, segregation stage and a schema driven stage with a heuristic that models whether listeners perceive a single or two simultaneous sounds. Introduction Previous studies [1,2,3] suggest that human listeners use simple cues, such as signal harmonicity, speaker location or segmental onset and offset to aid in the segregation of simultaneous sounds. These cues are called ‘primitive’ grouping cues because they can be applied without prior knowledge. The only heuristic is that segments in an ‘auditory scene’ that share the same features are likely to be produced by the same speaker. In addition to the primitive segregation process human listeners use high-level knowledge, schemata, to deal with mixtures of sounds [1]. One of the most intensively studied primitive grouping cues is harmonicity. Figure 1 shows human performance for a recognition task involving simultaneous vowels. Each of the panels shows the percentage of pairs that listeners correctly recognise. The stimuli were pairs of the French long vowels /#,G,K,Q,W,[/. One of the vowels always had a fundamental frequency (F0) of 100Hz, the fundamental frequency of the second vowel is plotted along the x-axis. The only primitive segregation cue is the vowel fundamental frequency. The three panels show subject performance for vowels of 200ms, 100ms and 50ms duration. For signals of at least 100ms duration subject performance improves significantly as the frequency difference between the vowels increases. If signals are only 50ms long no improvement in performance is seen. The perceptual data is surprising considering the dynamic nature of speech sounds where stationary segments of more than 100ms duration are very rare. Another important feature that emerges from the data is that humans are able to recognise both constituents of a pair in around 65% of all cases independent of the signal duration and without any grouping cues. 100106112 126 2nd vowel F0 40 50 60 70 80 90 % p ai rs c or re ct 204.8ms 100106112 126 2nd vowel F0 40 50 60 70 80 90 Human Performance

In this paper, we an interested in the comparison of several idnds of methods for fundamental frrquency estimation and GCI (Glottal Closure Instant) detection.These methods operatc in various doperformances have been compand for both fundamental frequency estimation and voicing decision tasks as well as GCI detection, when applicable.This comparison was designed to be as unbiased BS possible, SO as to nRect the infrinsic propaties of each method.A "d based on a "Bom-Jordan" kernel bilinear ti-frequency representation of speech signals achieves the best performance in terms of GQ. deuction accuracybut is not as robust to inter-speaker variability as the SIFT algorithm.An auditory model, which has been applied on the same data in a previous smdy, has been shown to compare favourably to other methods (such as SIFT) in adverse noisy conditions only.

In many speech analysis/synthesis schemes, the source for excitation for voiced speech is a train of impulses. However, the quality of speech that has been attained due to the introduction of a dynamically varying source, e.g., parametric source model, multipulse excitation has been found to be better than that using impulse excitation. The authors describe a pitch-synchronous glottal autoregressive moving average analysis/synthesis scheme in which a parametric voice source model is used in jointly estimating the source and vocal tract parameters from the speech signal. This method is then compared with closed-phase linear predictive coding (LPC), wherein covariance analysis is required to be carried out in the closed glottis interval, and with robust LPC, in which the analysis frame is insensitive to glottal closure. The superiority of the proposed scheme over the latter two methods is shown in terms of better formant/bandwidth tracking capability and efficiency of resynthesis.>