Pour un malentendant, le déficit auditif peut représenter un handicap majeur tant du point de vue physique que du point de vue social. Pour cette raison, la réhabilitation auditive constitue un champ d'application particulièrement important du traitement du signal audio et ses enjeux dépassent le seul contexte technologique. C'est dans ce cadre, et en particulier dans celui de la réhabilitation auditive par port de prothèses externes, que se situe ce travail de thèse. L'audition étant un sens particulièrement performant et complexe, compenser les pertes auditives des malentendants est une tâche difficile et délicate. En outre, les appareils sont conçus à partir d'éléments miniaturisés de pointe, notamment les micro-processeurs, ce qui impose des contraintes de fonctionnement très strictes. Au cours de nos recherches, nous avons travaillé sur la réalisation d'un module de rehaussement du signal de parole utilisant une technique d'atténuation spectrale à court-terme mise au point par Ephraïm et Malah. A l'issue de ce travail, nous avons proposé d'adapter cette méthode en réalisant l'analyse spectrale selon une échelle en fréquence conforme à notre perception auditive, l'échelle ERB (Equivalent Rectangular Bandwith) qui modélise la résolution fréquentielle des filtres auditifs. Dans un deuxième temps, nous avons étudié le problème de retour acoustique entre les transducteurs de la prothèse. Ce retour acoustique est source d'instabilité pour le système et aboutit fréquemment à la production d'un effet Larsen. Ce type de désagrément est extrêmement gênant et contraint l'audioprothésiste à limiter le gain d'amplification de la prothèse. De ce fait, la réhabilitation auditive ne peut pas être optimale. Pour pallier ce problème, nous proposons d'utiliser une méthode de filtrage adaptatif multi-délais basée sur le LMS (Least Mean Square) et réalisée dans le domaine de la transformée de Hartley. Du fait du retour acoustique, une prothèse auditive est un système électro-acoustique en boucle fermée. Les conditions d'application sont donc contraires aux hypothèses de convergence du LMS. Pour assurer la convergence de l'algorithme de descente de gradient, nous proposons une nouvelle méthode de définition du pas d'adaptation. Afin de vérifier le bon fonctionnement de la méthode proposée dans des conditions réalistes, nous avons mis en place une plate-forme de test en temps-réel basée sur un prototype de prothèses.

La transcription automatique de la musique (TAM) s'inscrit dans le champ de l'indexation audio et de la recherche d'informations musicales. Elle vise à extraire l'information relative aux notes jouées - début, fin, hauteur - dans un enregistrement numérique. Cette thèse aborde la TAM dans le cadre du piano, instrument difficile à transcrire par les systèmes actuels et qui occupe une place majeure d'instrument solo dans la musique occidentale. La problématique générale de la TAM étant vaste et complexe, nous exploitons certains aspects spécifiques du piano pour spécialiser la tâche, tout en abordant des thématiques générales comme l'estimation de fréquences fondamentales (F0) ou l'évaluation. Nous étudions tout d'abord la caractérisation spectrale des sons de piano pour la transcription à travers la distribution inharmonique des partiels, la modélisation de l'enveloppe spectrale des notes et celle du bruit. Nous proposons ensuite une méthode d'estimation de F0 isolées performante dans des conditions d'analyse difficiles qui sont réunies dans le cas du piano: les contraintes d'une fenêtre d'analyse courte et d'un registre étendu. L'estimation de F0 multiples et du degré de polyphonie est abordée via une modélisation conjointe paramétrique (autorégressive/à moyenne ajustée). Cette méthode spectrale s'intègre dans un système de TAM pour le piano analysant tout enregistrement avec des limites raisonnables en termes de polyphonie, de vélocité et de tessiture. Enfin, nous proposons de nouvelles métriques d'évaluation des transcriptions, un cadre d'évaluation de la qualité des métriques et une base de données de sons de piano pour l'estimation de F0 multiples et la TAM.

<p>The interest for various multimedia services accessed via the Internet has been growing immensely along with the bandwidth available. A similar development has emerged in the 3G mobile network. The focus of this master thesis is on the speech/audio part of a 3G multimedia application. The purpose has been to implement a traffic generating tool that can measure QoS (Quality of Service) in 3G networks. The application is compliant to the 3G standards, i.e. it uses AMR (Adaptive Multi Rate), SIP (Session Initiation Protocol) and RTP (Real Time Transport Protocol). AMR is a speech compression algorithm with the special feature that it can compress speech into several different bitrates. SIP signalling is used so that different applications can agree on how to communicate. RTP carries the speech frames over the network, in order to provide features that are necessary for media/multimedia applications. Issues like perception of audio and QoS related parameters is also discussed, from the perspective of users and developers.</p>

<p>The purpose with the report is to show that it is possible to design a class-D amplifier, using a programmable FPGA mounted on a developing card from Xess and a H-bridge. The FPGA was programmed in VHDL which is the language the software from Xilinx use to implement a logical function into the FPGA The logical function corresponds to a modeling of the music signal and the modeling can be described as a comparator which compare the music signal with a triangle wave and as a out signal produce a pulse width modulated (PWM) signal. The report is also a review and evaluating of two different modulating technologies, AD- modeling and BD-modeling. A detailed part about how the H-bridge was designed and how it works. The result of this project is a working audio amplifier to a significant lower price than the products in todays market.</p>

<p>This project investigates the importance of availability in the scope of dictation programs. Using speech recognition technology for dictating has not reached the public, and that may very well be a result of poor availability in today’s technical solutions. </p><p>I have constructed a persona character, Johanna, who personalizes the target user. I have also developed a solution that streams audio into a speech recognition server and sends back interpreted text. Johanna affirmed that the solution was successful in theory. </p><p>I then incorporated test users that tried out the solution in practice. Half of them do indeed claim that their usage has been and will continue to be increased thanks to the new level of availability.</p>

<p>Ogg Vorbis is a rather new audio format with some similarities with other more known formats such as MP3 and WMA. It is generally accepted to have a better audio quality than most competing formats and it is in contrast to many of its competitors totally licence and royalty free. </p><p>The goal with this thesis is to port the existing fixed point decoder Tremor, which is written in C, to Motorola's DSP56002. The DSP has a very limited amount of memory so some optimizations has to be made to be able to run Tremor successfully. </p><p>The report presents the necessary steps taken to port Tremor to the DSP and the difficulties of this process. It also describes the memory and CPU usage of the DSP when running Tremor and other results of the port. </p><p>A description as well as examples and workarounds of bugs found in the compiler g56k is attached to this report.</p>

<p>With the intention of surveying the field of research in adaptive audio systems for interactive media, a suggested audio system design for adaptive music control is described, and a prototype implementation of key parts of the system is presented and evaluated. </p><p>Foregoing midi-triggered sound banks, the proposed design uses layered segmented audio files, defined and controlled by XML-scripts. The results demonstrate an inclination of a flexible system, capable of adequate adaptive behaviour of high quality sound. </p><p>The implemented system will serve as an extensive basis for future work contributing to the research of adaptive behaviour to both music and sound effects for interactive media, and also as a preliminary foray into the more experimental field of stand-alone non-linear music playback.</p>

<p>“Reading with Your Ears” is a comparative study of comprehension in reading a text versus listening to an audio book. The text excerpt is from Mark Haddon’s <em>The Curious Incident of the Dog in the Night-Time</em> and has been read or listened to by seventh-grade students. The results show that the readers understand more from the text since they can read at their own speed and see the pictures in the book. The listeners, on the other hand, seem unaccustomed to listening and have troubles focusing.</p>

An investigation of the individual perceptions of selected students, teachers, and parents toward the use of computer-assisted instruction (CAI) in a careers program was undertaken to provide information about the role CAI plays in the career education curriculum and other curricular areas utilizing CAI. A qualitative methodology was selected which utilized the informal conversational interview, allowing the researcher to be responsive to individual differences and situational changes to establish in-depth communication. Data analysis consisted of a constant comparative process approach, which allowed for analysis of data from different sources, within a data triangulation design across the three subject groups. The subjects were from a small rural school district and consisted of five high school students, six of their parents, and their teacher. As a result of the research, several significant factors about CAI, career programs, and computer usage were formed into the following hypothesis: 1) Students who have access to computers on a regular basis and are provided with training will be comfortable and enthusiastic about their use. 2) Computers are an essential information and communication tool whose access and use by students gives them an advantage in school and later in life. 3) Students feel the use of the computer for word processing and publishing is significant in that it enables them to be better and more creative writers. 4) The computer is an educational tool which should be utilized in all areas of the curriculum. 5) Careers curriculum is essential and enhanced by the use of CAI for accessing information and providing search and other functions. / Graduation date: 1995

Aims: The aims of this thesis were to describe and gain an understanding of how people who live in the vicinity of wind turbines are affected by wind turbine noise, and how individual, situational and visual factors, as well as sound properties, moderate the response. Methods: A cross-sectional study was carried out in a flat, mainly rural area in Sweden, with the objective to estimate the prevalence of noise annoyance and to examine the dose-response relationship between A-weighted sound pressure levels (SPLs) and perception of and annoyance with wind turbine noise. Subjective responses were obtained through a questionnaire (n = 513; response rate: 68%) and outdoor, A-weighted SPLs were calculated for each respondent. To gain a deeper understanding of the observed noise annoyance, 15 people living in an area were interviewed using open-ended questions. The interviews were analysed using the comparative method of Grounded Theory (GT). An additional cross-sectional study, mainly exploring the influence of individual and situational factors, was carried out in seven areas in Sweden that differed with regard to terrain (flat or complex) and degree of urbanization (n = 765; response rate: 58%). To further explore the impact of visual factors, data from the two cross-sectional studies were tested with structural equation modelling. A proposed model of the influence of visual attitude on noise annoyance, also comprising the influence of noise level and general attitude, was tested among respondents who could see wind turbines versus respondents who could not see wind turbines from their dwelling, and respondents living in flat versus complex terrain. Results: Dose-response relationships were found both for perception of noise and for noise annoyance in relation to A-weighted SPLs. The risk of annoyance was enhanced among respondents who could see at least one turbine from their dwelling and among those living in a rural in comparison with a suburban area. Noise from wind turbines was appraised as an intrusion of privacy among people who expected quiet and peace in their living environment. Negative experiences that led to feelings of inferiority added to the distress. Sound characteristics describing the amplitude modulated aerodynamic sound were appraised as the most annoying (swishing, whistling and pulsating/throbbing). Wind turbines were judged as environmentally friendly, efficient and necessary, but also as ugly and unnatural. Being negative towards the visual impact of the wind turbines on the landscape scenery, rather than towards wind turbines as such, was strongly associated with annoyance. Self-reported health impairment was not correlated to SPL, while decreased well-being was associated with noise annoyance. Indications of possible hindrance to psycho-physiological restoration were observed. Conclusions: Wind turbine noise is easily perceived and is annoying even at low A-weighted SPLs. This could be due to perceived incongruence between the characteristics of wind turbine noise and the background sound. Wind turbines are furthermore prominent objects whose rotational movement attracts the eye. Multimodal sensory effects or negative aesthetic response could enhance the risk of noise annoyance. Adverse reactions could possibly lead to stress-related symptoms due to prolonged physiological arousal and hindrance to psychophysiological restoration. The observed differences in prevalence of noise annoyance between living environments make it necessary to assess separate dose-response relationships for different types of landscapes.