There is little information about mixing music for 9.1 surround, and the information available are mostly for classical music and the use of ambience in the height channels. The production methods for pop music differs from classical music and the productions does not have to sound realistic. In this study, a solo pop guitar was recorded and placed in the center channel. A 9.1 surround playback environment was created and two short delays were added to increase the perceived width of the guitar and shifted between the front L,R speakers, the back L,R, the front height L,R and back height L,R. A listening test was conducted where subjects compared how they perceived the different delay positions. The results showed that the positions were perceived very differently and depends on the listener, although some tendencies were found. The front position was perceived as broader and had were generally rated highest. The front height position was also described as broad but also bright and distinctive. The back position was generally rated lowest and perceived as thin and dull. The back-height position was perceived as dull and to have more bass.

When recording music for surround audio engineers sometimes face limitations in time, ideal microphone positions or a noisy audience. If this cannot be dealt with at the location, artificial reverbs are often used in the mixing to “fill in the blanks”. In this study, three instruments were recorded separately with two 5.1 surround microphone setups. Two audio engineer students created artificial reverbs that replaced the back channels of each microphone setup. A listening test was conducted where test subjects compared the 5.1 real recording to the two other stimuli with artificial back channels in terms of realism, envelopment and preference. The result showed that the real recording and the artificial back channels were interchangeable, but that the artificial back channels pointed towards more envelopment, and that the real recording pointed towards more realism.

Aujourd'hui, les technologies de captation et de restitution sonore se développent dans le but de diffuser des scènes avec un rendu spatialisé. Avant leur diffusion, les extraits sonores peuvent être évalués en terme de qualité par des méthodes recommandées par I'Union lnternationale des Télécommunications (évaluation des codecs de compression, procédés de prise ou restitution sonore...). Cependant, ces standards d'évaluation montrent certaines faiblesses notamment en ce qui concerne les attributs de qualité à évaluer. La dimension spatiale n'est pas prise en compte spécifiquement. Dans ce travail, une méthodologie dédiée à l'évaluation de la qualité de I'audio spatialisé est mise en place notamment pour répondre aux biais identifiés. De par l'utilisation d'une catégorisation libre et d'une analyse multidimensionnelle, vingt-huit attributs ont été catégorisés en trois familles d'attributs : le Timbre, l'Espace et les Défauts. Ces trois attributs généraux ont été inclus dans un test d'écoute. Celui-ci se déroule en deux phases : l'évaluation de la qualité globale suivie de l'évaluation des trois attributs simultanément sur une même interface. Les tests sont réalisés sans référence explicite, le fichier original constitue une référence cachée. De plus, trois signaux audio, dit ancrages, spécifiques à chacun des trois attributs ont été définis puis superposés pour définir un ancrage unique triplement dégradé. La méthode a été testée à la fois sur un système de restitution au casque avec des contenus binauraux mais également sur un système multicanal 5.1. L'évaluation de stimuli de qualité intermédiaire est préconisée ainsi que des contenus présentant un effet spatial prononcé. L'évaluation multicritère a montré son intérêt dans certaines conditions et permet ainsi d'identifier les caractéristiques qui sont dégradées. Les attributs Défauts et Timbre ont montré un poids influant sur la qualité globale tandis que le poids de I'attribut Espace est plus discutable. / Nowadays, recording and restitution technologies focus on a spatial rendering of sound. Before their broadcast, the quality evaluation of sound excerpts is often necessary. Methods recommended by the international telecommunication union denote some weaknesses about sound attributes to be evaluated.For example, spatial dimension is barely taken into account. A methodology dedicated to the assessment of spatial audio quality is proposed in order to avoid some biases. With a free categorization and a multidimensional scaling, 28 attributes were clustered in three families '. Timbre, Space and Defecfs. These three categories were included in a listening test split into two sessions : first, the assessment of overall quality and then, the evaluation of the three categories presented simultaneously on a same interface.Tests were conducted without explicit reference, but, the original version was considered as a hidden reference. Moreover, three specific anchors, each one associated to dedicated categories, were defined and then were mixed to define a unique anchor impaired in three ways. The method was tested on a 5.1 system and on binaural contents with headphone restitution. lntermediate quality of contents is recommended as well as contents with relevant spatial effects. The interest of a multicriteria assessment is to identify which properties of sound are impaired. Linear regression shows that Defecfs and Timbre attributes have influential weight on overall quality while the weight of Space attribute is more dubious.

The use of conversational agents has increased across the world. From providing automated support for companies to being virtual psychologists they have moved from an academic curiosity to an application with real world relevance. While many researchers have focused on the content of the dialogue and synthetic speech to give the agents a voice, more recently animating these characters has become a topic of interest. An additional use for character animation technology is in the film and video game industry where having characters animated without needing to pay for expensive labour would save tremendous costs. When animating characters there are many aspects to consider, for example the way they walk. However, to truly assist with communication automated animation needs to duplicate the body language used when speaking. In particular conversational agents are often only an animation of the upper parts of the body, so head motion is one of the keys to a believable agent. While certain linguistic features are obvious, such as nodding to indicate agreement, research has shown that head motion also aids understanding of speech. Additionally head motion often contains emotional cues, prosodic information, and other paralinguistic information. In this thesis we will present our research into synthesising head motion using only recorded speech as input. During this research we collected a large dataset of head motion synchronised with speech, examined evaluation methodology, and developed a synthesis system. Our dataset is one of the larger ones available. From it we present some statistics about head motion in general. Including differences between read speech and story telling speech, and differences between speakers. From this we are able to draw some conclusions as to what type of source data will be the most interesting in head motion research, and if speaker-dependent models are needed for synthesis. In our examination of head motion evaluation methodology we introduce Forced Canonical Correlation Analysis (FCCA). FCCA shows the difference between head motion shaped noise and motion capture better than standard methods for objective evaluation used in the literature. We have shown that for subjective testing it is best practice to use a variation of MUltiple Stimuli with Hidden Reference and Anchor (MUSHRA) based testing, adapted for head motion. Through experimentation we have developed guidelines for the implementation of the test, and the constraints on the length. Finally we present a new system for head motion synthesis. We make use of simple templates of motion, automatically extracted from source data, that are warped to suit the speech features. Our system uses clustering to pick the small motion units, and a combined HMM and GMM based approach for determining the values of warping parameters at synthesis time. This results in highly natural looking motion that outperforms other state of the art systems. Our system requires minimal human intervention and produces believable motion. The key innovates were the new methods for segmenting head motion and creating a process similar to language modelling for synthesising head motion.

The focus of this diploma thesis is on methods for the subjective assessment of sound quality according to recommandations given by International Telecommunication Union ITU. The thesis is thematically divided into four parts. The first part is an interpretation of methods based on internationally accepted standards (method of small impairments, MUSHRA, general methods). The second part describes the functional blocks of application developed for this thesis in LabVIEW enviroment. Next part explains its practical application, especially its running possibilities. The last part describes a listening experiment, aim of which was to verify the application's features and to compare the results of the subjective method used in the experiment with the results of the objective method PEAQ.