<p>The acoustic echo problem arises whenever there is acoustic coupling between a loudspeaker and a microphone, such as in a teleconference system. This problem is traditionally solved by using an acoustic echo canceler (AEC), which models the echo path with adaptive filters. Long adaptive filters are necessary for satisfactory echo cancellation, which makes AEC highly computationally complex. Recently, a low-complexity echo suppression scheme was presented, the perceptual acoustic echo suppressor (PAES). Spectral modification is used to suppress the echoes, and the complexity is reduced by incorporating perceptual theories. However, under ideal conditions AEC performs better than PAES. </p><p>This thesis considers a hybrid system, which combines AEC and PAES. AEC is used to cancel low-frequency echo components, while PAES suppresses high-frequency echo components. The hybrid system is simulated and assessed, both through subjective listening tests and objective evaluations. The hybrid scheme is shown to have virtually the same perceived quality as a full-band AEC, while having a significantly lower complexity and a higher degree of robustness.</p>

Central auditory processing is complex and can not be evaluated by a single method. This thesis focuses on assessment of some aspects of central auditory functions by the use of dichotic speech tests and cognitive tests that tax functions important for speech processing. Paper A deals with the cognitive effects in dichotic speech testing in elderly hearing-impaired subjects. It was found that different listening tasks in the dichotic tests put different demands on cognitive ability, shown by a varying degree of correlation between cognitive functions and dichotic test parameters. Age-related cognitive decline was strongly connected with problems to perceive stimuli presented to the left ear. Paper B presents a new cognitive test battery sensitive for functions important for speech processing and understanding, performed in text, auditory and audiovisual modalities. The test battery was evaluated in four groups, differing in age and hearing status, and has proven to be useful in assessing the relative contribution of different input-modalities and the effect of age, hearingimpairment and visual contribution on functions important for speech processing. In Paper C the test battery developed in Paper B was used to study listening situations with different kinds of background noise. Interfering noise at +10 dB signal-to-noise ratio has significant negative effects on performance in speech processing tasks and on the effort perceived. Hearing-impaired subjects showed poorer results in noise with temporal variations, and elderly subjects were more distracted by noise with temporal variations, especially by noise with meaningful content. In noise, all subjects, particularly those with impaired hearing, were more dependent upon visual cues than in the quiet condition. Hearing aid benefit in speech processing with and without background noise was studied in Paper D. The test battery developed in Paper B was used together with a standard measure of speech recognition. With hearing aids, speech recognition was improved in the background condition without noise and in the background condition of ordinary speech. Significantly less effort was perceived in the cognitive tests when hearing aids were used, although only minor benefits of hearing aid amplification were seen. This underlines the importance of considering perceived effort as a dimension when evaluating hearing aid benefit, in further research as well as in clinical practice. The results from the studies contribute to the knowledge about speech processing but also to the search for more specific evaluation of speech understanding, incorporating both sensory and cognitive factors. / The ISBN 91-85297-49-6 in the printed verison is incorrect. The correct ISBN is 91-85297-93-3.

The acoustic echo problem arises whenever there is acoustic coupling between a loudspeaker and a microphone, such as in a teleconference system. This problem is traditionally solved by using an acoustic echo canceler (AEC), which models the echo path with adaptive filters. Long adaptive filters are necessary for satisfactory echo cancellation, which makes AEC highly computationally complex. Recently, a low-complexity echo suppression scheme was presented, the perceptual acoustic echo suppressor (PAES). Spectral modification is used to suppress the echoes, and the complexity is reduced by incorporating perceptual theories. However, under ideal conditions AEC performs better than PAES. This thesis considers a hybrid system, which combines AEC and PAES. AEC is used to cancel low-frequency echo components, while PAES suppresses high-frequency echo components. The hybrid system is simulated and assessed, both through subjective listening tests and objective evaluations. The hybrid scheme is shown to have virtually the same perceived quality as a full-band AEC, while having a significantly lower complexity and a higher degree of robustness.

Realizou-se um inventário quantitativo e qualitativo das características demográficas e dos erros cometidos nos itens dicóticos do Teste de Dissílabos Dicóticos Alternados, também conhecido como SSW em português, com as informações de 722 sujeitos que completaram este teste nos anos de 1994 a 2001. Os resultados revelaram que a amostra estudada foi composta em sua maioria de sujeitos do sexo masculino, cursando as séries iniciais do ensino fundamental, relatando dificuldades de fala, escria e leitura e neurologicamente normais, com desordem de processamento auditivo do tipo decodificação, de grau leve/moderado. Uma tendência de erro em alguns itens dicóticos do teste foi observada. / A total of 722 subjects provided the information for the quantitative and qualitative inventory of demographic features and the errors observed in the dichotic items of the Portuguese version of the SSW Test. Results revealed that, collectively, these subjects were mostly males, in elementary school, referring speech, reading and writing difficulties, and not presenting neurological disorders. They also presented a decoding-type auditory processing disorder of mild/moderate severity. Error and performance profiles were characterized by multivariate statistics.

Le processus du traitement de la parole est latéralisé au niveau cortical. En effet la théorie de l’échantillonnage asymétrique suggère que le signal acoustique est segmenté sous forme d’unités discrètes, et ensuite traité par les deux cortex auditifs non primaires en utilisant des fenêtres d’intégrations respectivement adaptées aux traitements des modulations lentes et rapides du signal de la parole. L’objectif de cette thèse était d’étudier les concepts de cette théorie au moyen de deux approches (psychophysique et neurophysiologique), en s’intéressant à l’activation corticale dans les deux hémisphères en fonction de la nature du signal auditif présenté. Les résultats révèlent que les scores d’intelligibilité de ces stimuli sont plus importants quand les modulations rapides arrivent au cortex auditif gauche et les modulations lentes arrivent au cortex auditif droit. Les résultats de l’IRMf démontrent une interaction entre l’enveloppe des stimuli présentés et les rythmes corticaux. Les deux cortex auditifs présentaient des asymétries inter-hémisphériques en réponse à ces stimuli. En outre, l’activité neuronale augmente avec les performances. Dans l'ensemble, les résultats de nos études vont dans le sens des prédictions de la théorie d’échantillonnage asymétrique. Nous supposons que, l’application des effets de la théorie de la latéralisation auditive dans le traitement du signal auditif pourrait améliorer la perception de la parole chez les personnes sourdes profonds en développant les stratégies de codage des Implants cochléaire de manière à les alignées sur les propriétés corticales d’échantillonnages intrinsèques / Speech perception consists of a set of bilateral computations that take continuously varying acoustic waveforms as input and generate discrete representations. Hypothesis of ‘asymmetric sampling in time’, suggests that auditory functional asymmetries can be explained by differences in temporal sampling between the two auditory cortices. We suggest that asymmetry in auditory cortical oscillations could play a role in speech perception by fostering hemispheric triage of information across the two hemispheres. Due to this asymmetry, fast speech temporal modulations, could be best perceived by the left auditory cortex, while slower modulations would be better captured by the right one. The aim of this thesis was to study and to test the validity of the predictions of the AST theory by investigating psychophysical and neurophysiological approach. They focus on the cortical activation in both hemispheres according to the nature of the auditory signal presented to both ears. Our results show that when we provide a different part of the speech envelope to each ear, word recognition is facilitated when the temporal properties of speech match the rhythmic properties of auditory cortices. We further show that the interaction between speech envelope and auditory cortices rhythms translates in their level of neural activity (as measured with fMRI). In the left auditory cortex, the neural activity level related to stimulus/brain rhythm interaction predicts speech perception facilitation. This interaction impacts speech perception performance. We propose that this lateralization effect could have practical implications in the framework of bilateral cochlear implants

yes / This paper offers a brief overview of innovative methods for road and rail traffic noise reduction between source and receiver. These include using new barrier designs, planting of trees, treatments of ground and road surfaces and greening of building façades and roofs using natural materials, like vegetation, soil and other substrates in combination with recycled materials and artificial elements. The abatements are assessed in terms of numerically predicted sound level reductions, perceptual effects and cost–benefit analysis. Useful reductions of noise from urban roads and tramways are predicted for 1-m-high urban noise barriers and these are increased by adding inter-lane barriers. A 3 m wide 0.3 m high lattice ground treatment, a carefully planted 15-m-wide tree belt and replacing 50 m of paved areas by grassland are predicted to give similar reductions. Tree belts are shown to be very cost-effective and combining tall barriers with a row of trees reduces the negative impact of wind. Green roofs may significantly reduce the noise at the quiet side of buildings.