Sound Source Separation refers to separating sound signals according to their sources from a given observed sound. It is efficient to code and very easy to analyze and manipulate sounds from individual sources separately than in a mixture. This thesis deals with the problem of source separation in monaural recordings of harmonic musical instruments. A good amount of literature is surveyed and presented since sound source separation has been tried by many researchers over many decades through various approaches. A prediction driven approach is first presented which is inspired by old-plus-new heuristic used by humans for Auditory Scene Analysis. In this approach, the signals from different sources are predicted using a general model and then these predictions are reconciled with observed sound to get the separated signal. This approach failed for real world sound recordings in which the spectrum of the source signals change very dynamically. Considering the dynamic nature of the spectrums, an approach which uses covariance matrix of amplitudes of harmonics is proposed. The overlapping and non-overlapping harmonics of the notes are first identified with the knowledge of pitch of the notes. The notes are matched on the basis of their covariance profiles. The second order properties of overlapping harmonics of a note are estimated with the use of co-variance matrix of a matching note. The full harmonic is then reconstructed using these second order characteristics. The technique has performed well over sound samples taken from RWC musical Instrument database.

En undersökning om HRTFs effektivitet när spelandet sker med ett öra.

The mature central nervous system (CNS) has the capacity to reorganise when there is a change in sensory input. However, studies using the N1 cortical auditory evoked potential, or its magnetic homolog N1m, have not consistently demonstrated evidence of plasticity in adults with late onset unilateral deafness. In addition, little is known about the time course of experience-related plasticity in adults with unilateral deafness. The aim of the studies described in this thesis was to investigate plasticity in adults with unilateral deafness, using N1 auditory evoked potentials. Deafness occurred as a result of surgery for the removal of an acoustic neuroma. The stimuli were 500-Hz and 4-kHz tones presented monaurally to the intact ear, and the data were analysed using global field power and dipole source analysis. In the first study (Chapter 3), hemispheric asymmetries in the N1 response were measured in a group of 24 normally hearing adults at presentation levels of 40, 60 and 80 dB sensation level (SL). The results revealed that the mean hemispheric asymmetry was greater for the 4-kHz stimulus but there was no significant effect of presentation level. In addition, the results revealed that the magnitude of hemispheric asymmetry depended on the ear of stimulation; a trend for larger asymmetries was observed following stimulation of the left ear. The results of the study provide confidence that the methodology is suitable for measuring hemispheric asymmetries in individuals with unilateral deafness. The effect of stimulus level is important since this will vary in plasticity studies involving individuals with late onset unilateral deafness due to their pure tone sensitivity thresholds. Clarifying the effect of stimulus frequency in normally hearing adults is important since the effect of stimulus frequency on plasticity following unilateral deafness has not been reported previously.In the second study (Chapter 4), N1 responses were measured in 19 adults with unilateral deafness (10 and 9 right- and left-sided deafness respectively). Stimuli were typically presented at 60 dB SL. The results revealed that there was significantly greater mean activity and a shift towards reduced hemispheric asymmetries compared with 19 audiogram-matched controls. Similar changes were apparent after both right- and left-sided deafness, and for both 500-Hz and 4-kHz stimuli. Therefore the results reveal evidence of experience-related plasticity that mirrors the findings reported in animal models. The reduced hemispheric asymmetries were reflected in the dipole source model used in this thesis by changes in dipole strength, location and orientation. These findings may explain the inconsistencies reported in previous studies that have used N1 or N1m, where dipole location and orientation have not always been taken into adequate consideration.In the third study (Chapter 5), longitudinal measurements were made in six adults just prior to the onset of complete unilateral deafness, and at 1-, 3- and 6-months after the onset (4 right-sided and 2 left-sided deafness respectively). The results from the second study were further analysed by splitting the data into two groups: nine participants with <2 years deafness and 10 with ≥2 year's deafness. The results from the longitudinal data revealed that there was a significant difference in mean activity across the four conditions. For both stimulus types an increase in mean activity occurred after the onset of deafness, and hemispheric asymmetries were reduced. The biggest changes occurred within 1-month, although further increases were noted in some individuals with ≥2 year's duration of deafness. Changes that continue over this period of time suggest different physiological mechanisms for plasticity within the human central auditory system.

Abstract is available through the Abstracts of the Association for Research in Otolaryngology.

<p>Studien riktar sig främst till audionomer, och andra med intresse av hörselrehabilitering.</p><p>Auditiv deprivering kan visa sig som en med tiden försämrad taluppfattning i örat som inte fått adekvat stimulering. Vad som karaktäriserar dem som drabbas är inte klarlagt. Inte heller vet man exakt orsak eller tidsförlopp. Genom att hörapparatanpassa det drabbade örat har effekten kunnat reduceras och i vissa fall har fullständig återhämtning konstaterats.</p><p>Litteraturstudiens syfte var att belysa fenomenet och aspekter som bör beaktas i audionomens arbete med insatser av både förebyggande och rehabiliterande karaktär. Metoden var att granska artiklar framtagna genom systematisk och manuell sökning, i databaser tillgängliga genom Örebro universi¬tet.</p><p>Resultatet visar att personer med hörselnedsättning av såväl sensorineural som konduktiv art drabbas. Hörselnedsättningens grad kan troligen påverka, liksom symmetrin mellan öronen. Den initiala taluppfattningen i örat som drabbas kan variera. Detta arbete visar också att en sensitiv metod för att konstatera effekten är den som i Sverige heter maximal taluppfattning, men objektiva mätmetoder välkomnas. Audionomens ställningstagande i val av en eller två hörapparater berörs, med utgångspunkt av auditiv deprivering och binaural interference. Resultat och diskussion lyfter viktiga aspekter att beakta för audionomen i det rehabiliterande arbetet.</p>

Studien riktar sig främst till audionomer, och andra med intresse av hörselrehabilitering. Auditiv deprivering kan visa sig som en med tiden försämrad taluppfattning i örat som inte fått adekvat stimulering. Vad som karaktäriserar dem som drabbas är inte klarlagt. Inte heller vet man exakt orsak eller tidsförlopp. Genom att hörapparatanpassa det drabbade örat har effekten kunnat reduceras och i vissa fall har fullständig återhämtning konstaterats. Litteraturstudiens syfte var att belysa fenomenet och aspekter som bör beaktas i audionomens arbete med insatser av både förebyggande och rehabiliterande karaktär. Metoden var att granska artiklar framtagna genom systematisk och manuell sökning, i databaser tillgängliga genom Örebro universi¬tet. Resultatet visar att personer med hörselnedsättning av såväl sensorineural som konduktiv art drabbas. Hörselnedsättningens grad kan troligen påverka, liksom symmetrin mellan öronen. Den initiala taluppfattningen i örat som drabbas kan variera. Detta arbete visar också att en sensitiv metod för att konstatera effekten är den som i Sverige heter maximal taluppfattning, men objektiva mätmetoder välkomnas. Audionomens ställningstagande i val av en eller två hörapparater berörs, med utgångspunkt av auditiv deprivering och binaural interference. Resultat och diskussion lyfter viktiga aspekter att beakta för audionomen i det rehabiliterande arbetet.

Nous traitons le sujet de l’estimation conjointe des signaux aléatoires dépendant de paramètres déterministes et inconnus. Premièrement, on aborde le sujet du côté applicatif en proposant deux algorithmes de séparation de la parole voisée mono-capteur. Dans le premier, nous utilisons le modèle autorégressif de la parole qui décrit les corrélations court et long termes (quasi-périodique) pour formuler un modèle d’état dépendant de paramètres inconnus. EM-Kalman est ainsi utilisé pour estimer conjointement les sources et les paramètres. Dans le deuxième, nous proposons une méthode fréquentielle pour le même modèle de la parole où les sources et les paramètres sont estimés séparément. Les observations sont découpées à l’aide d’un fenêtrage bien conçu pour assurer une reconstruction parfaite des sources après. Les paramètres (de l’enveloppe spectrale) sont estimés en maximisant le critère du GML exprimé avec la matrice de covariance paramétrée que nous modélisons plus correctement en tenant compte de l’effet du fenêtrage. Le filtre de Wiener est utilisé pour estimer les sources. Deuxièmement, on aborde l’estimation conjointe d’un point de vue plus théorique en s'interrogeant sur les performances relatives de l’estimation conjointe par rapport à l’estimation séparée d’une manière générale. Nous considérons le cas conjointement Gaussien (observations et variables cachées) et trois méthodes itératives d'estimation conjointe: MAP en alternance avec ML, biaisé même asymptotiquement pour les paramètres, EM qui converge asymptotiquement vers ML et VB que nous prouvons converger asymptotiquement vers la solution ML pour les paramètres déterministes. / The thesis is composed of two parts. In the first part, we deal with the monaural speech separation problem. We propose two algorithms. In the first algorithm, we exploit the joint autoregressive model that models short and long (periodic) correlations of Gaussian speech signals to formulate a state space model with unknown parameters. The EM-Kalman algorithm is then used to estimate jointly the sources (involved in the state vector) and the parameters of the model. In the second algorithm, we use the same speech model but this time in the frequency domain (quasi-periodic Gaussian sources with AR spectral envelope). Observation data is sliced using a well-designed window. Parameters are estimated separately from the sources by optimizing the Gaussian ML criterion expressed using the sample and parameterized covariance matrices. Classical frequency domain asymptotic methods replace linear convolution by circulant convolution leading to approximation errors. We show how the introduction of windows can lead to slightly more complex frequency domain techniques, replacing diagonal covariance matrices by banded covariance matrices, but with controlled approximation error. The sources are then estimated using the Wiener filtering. The second part is about the relative performance of joint vs. marginalized parameter estimation. We consider jointly Gaussian latent data and observations. We provide contributions to Cramer-Rao bounds, then, we investigate three iterative joint estimation approaches: Alternating MAP/ML which suffers from inconsistent parameter bias, EM which converges to ML and VB that we prove converges asymptotically to the ML solution for parameter estimation.

No description available.

Binaural beamforming technology, which is based on the auditory perception of both ears, uses a wireless data connection to exchange data between the right-side and the left-side hearing aids. Over the years, several multichannel speech enhancement algorithms have been used in the hearing aid industry. For example, beamforming algorithms work by keeping a target signal undistorted while attenuating the noise fields (such as diffuse noise or white noise) and the interferers from different directions. Fixed and adaptive algorithms of this nature have been under active investigation by the hearing aid industry. Although binaural beamforming hearing aids designs have shown better performance than single-channel based hearing aids or bilateral hearing aids, the performance of binaural beamforming still suffers from errors in the direction of arrival estimates, i.e., errors which occur when the right set of steering vectors is used in a beamformer design but the target signal source is not located at the direction considered in the design. Therefore, this thesis is devoted to find and propose structures showing more robustness to errors in the direction of arrival estimates. The focus is mainly on the Generalized Sidelobe Canceller (GSC) structure and several binaural beamforming algorithms and configurations are proposed in this thesis as alternatives for the fixed beamformer and blocking matrix units of the GSC. The proposed algorithms show promise of providing wider notch and/or wider beam possibilities, as well as providing greater noise reduction and superior adaptive null positioning capabilities. The algorithms proposed in this thesis were simulated in MATLAB using recorded signals and data provided by a hearing aid firm, to assess their utility for improving hearing aid performance. The results demonstrated a superiority over algorithms currently in use in industry.

No description available.