1,721,051 research outputs found
Speech perception in a sparse domain
No full textEnvironmental statistics are known to be important factors shaping our perceptual system. The visual and auditory systems have evolved to be efficient for processing natural images or speech. The common characteristics between natural images and speech are that they are both highly structured, therefore having much redundancy. Our perceptual system may use redundancy reduction and sparse coding strategies to deal with complex stimuli every day. Both redundancy reduction and sparse coding theory emphasise the importance of high order statistics signals.This thesis includes psycho-acoustical experiments designed to investigate how higher order statistics affect our speech perception. Sparseness can be defined by the fourth order statistics, kurtosis, and it is hypothesised that greater kurtosis should be reflected by better speech recognition performance in noise. Based on a corpus of speech material, kurtosis was found to be significantly correlated to the glimpsing area of noisy speech, an established measure that predicts speech recognition. Kurtosis was also found to be a good predictor of speech recognition and an algorithm based on increasing kurtosis was also found to improve speech recognition score in noise. The listening experiment for the first time showed that higher order statistics are important for speech perception in noise.It is known the hearing impaired listeners have difficulty understanding speech in noise. Increasing kurtosis of noisy speech may be particularly helpful for them to achieve better performance. Currently, neither hearing aids nor cochlear implants help hearing impaired users greatly in adverse listening enviroments, partly due to having a reduced dynamic range of hearing. Thus there is an information bottleneck, whereby these devices must transform acoustical sounds with a large dynamic range into the smaller range of hearing impaired listeners. The limited dynamic range problem can be thought of as a communication channel with limited capacity. Information could be more efficiently encoded for such a communication channel if redundant information could be reduced. For cochlear implant users, unwanted channel interaction could also contribute lower speech recognition scores in noisy conditions.This thesis proposes a solution to these problems for cochlear implant users by reducing signal redundancy and making signals more sparse. A novel speech processing algorithm, SPARSE, was developed and implemented. This algorithm aims to reduce redundant information and transform signals input into more sparse stimulation sequences. It is hypothesised that sparse firing patterns of neurons will be achieved, which should be more biological efficient based on sparse coding theory. Listening experiments were conducted with ten cochlear implant users who listened to speech signals in modulated and speech babble noises, either using the conventional coding strategy or the new SPARSE algorithm. Results showed that the SPARSE algorithm can help them to improve speech understanding in noise, particularly for those with low baseline performance. It is concluded that signal processing algorithms for cochlear implants, and possibly also for hearing aids, that increase signal sparseness may deliver benefits for speech recognition in noise. A patent based on the algorithm has been applied for
Sparseness and speech perception in noise
No full textCan we model speech recognition in noise by exploring higher order statistics of the combined signal? How will changes in these statistics affect speech perception in noise? This study addresses these questions in two experiments. One investigated the relationship between an established "glimpsing" model and the fourth order statistic, kurtosis. The glimpsing model [1] proposes that listeners can explore the local speech-to-noise ratio (SNR) in short time segments (glimpses) and focus on areas where SNR is high. Results showed that there is a very high correlation between percentages of glimpsing area and kurtosis (r = 0.99;p < 0.01), suggesting that kurtosis can serve as a simpler index for measuring glimpsing. The experiment also examined the association between kurtosis and recognition of nonsense words (vowel-consonant-vowel, VCV) in babble modulated noise, also showing very high correlation (r = 0.97;p < 0.01). Another separate study focused on the relationship of sparseness to speech recognition score for VCV words in natural babble noise made of 100 people talking simultaneously [2]. Results show that there is also high correlation between kurtosis and speech recognition score with this noise. Logistic regression analysis to obtain the kurtosis for 50% correct showed this was achieved at a kurtosis of approximately 1.
Independent component analysis, a new framework for speech processing of cochlear implant?
No full textSparse stimuli for cochlear implants
Full text linkPublication in the conference proceedings of EUSIPCO, Lausanne, Switzerland, 200
Application of a sparse coding strategy to enhance speech perception for hearing aid users
No full textRelationship between speech recognition in noise and sparseness
No full textEstablished methods for predicting speech recognition in noise require knowledge of clean speech signals, placing limitations on their application. The study evaluates an alternative approach based on characteristics of noisy speech, specifically its sparseness as represented by the statistic kurtosis. Design: Experiments 1 and 2 involved acoustic analysis of vowel-consonant-vowel (VCV) syllables in babble noise, comparing kurtosis, glimpsing areas, and extended speech intelligibility index (ESII) of noisy speech signals with one another and with pre-existing speech recognition scores. Experiment 3 manipulated kurtosis of VCV syllables and investigated effects on speech recognition scores in normal-hearing listeners. Study sample: Pre-existing speech recognition data for Experiments 1 and 2; seven normal-hearing participants for Experiment 3. Results: Experiments 1 and 2 demonstrated that kurtosis calculated in the time-domain from noisy speech is highly correlated (r > 0.98) with established prediction models: glimpsing and ESII. All three measures predicted speech recognition scores well. The final experiment showed a clear monotonic relationship between speech recognition scores and kurtosis. Conclusions: Speech recognition performance in noise is closely related to the sparseness (kurtosis) of the noisy speech signal, at least for the types of speech and noise used here and for listeners with normal hearing<br/
- …
