Rozprawy doktorskie na temat „Parametric coding”

This thesis presents a stereo coding technique used as an extension for the Enhanced Voice Services (EVS) codec [10] [8]. EVS is an audio codec recently standardized by the 3rd Generation Partnership Project (3GPP) for compressing mono signals at chosen rates from 7.2 to 128 kbit/s (for xed bit rate) and around 5.9 kbit/s (for variable bit rate). The main goal of the thesis is to present the architecture of a parametric stereo codec and how the stereo extension of EVS may be built. Parametric stereo coding relies on the transmission of a downmixed signal, sum of left and right channels, and the necessary audible cues to synthesize back the stereo image from it at the decoding end. The codec has been implemented in MATLAB with use of the existing EVS codec. An important part of the thesis is dedicated to the description of the implementation of a robust downmixing technique. The remaining parts present the parametric coding architecture that has been adapted and used to develop the EVS stereo extension at 24.4 and 32 kbit/s and other open researches that have been conducted for more specic situations such as spatial coding for stereo or binaural applications. Whereas the downmixing algorithm quality has been confronted to subjective testing and proven to be more ecient than any other existing techniques, the stereo extension has been tested less extensively. Yet the quality reached with the proposed reconstruction algorithms tends to highlight the potential of the codec that could be revealed by future work.

Résumé : Afin de répondre aux besoins de communication et de divertissement, il ne fait aucun doute que la parole et l’audio doivent être encodés sous forme numérique. En qualité CD, cela nécessite un débit numérique de 1411.2 kb/s pour un signal stéréo-phonique. Une telle quantité de données devient rapidement prohibitive pour le stockage de longues durées d’audio ou pour la transmission sur certains réseaux, particulièrement en temps réel (d’où l’adhésion universelle au format MP3). De plus, ces dernières années, la quantité de productions musicales et cinématographiques disponibles en cinq canaux et plus ne cesse d’augmenter. Afin de maintenir le débit numérique à un niveau acceptable pour une application donnée, il est donc naturel pour un codeur audio à bas débit d’exploiter la redondance entre les canaux et la psychoacoustique binaurale. Le codage perceptuel et plus particulièrement le codage paramétrique permet d’atteindre des débits manifestement inférieurs en exploitant les limites de l’audition humaine (étudiées en psychoacoustique). Cette recherche se concentre donc sur le codage paramétrique à bas débit de plus d’un canal audio. // Abstract : In order to fulfill our communications and entertainment needs, there is no doubt that speech and audio must be encoded in digital format. In"CD" quality, this requires a bit-rate of 1411.2 kb/s for a stereo signal. Such a large amount of data quickly becomes prohibitive for long-term storage of audio or for transmitting on some networks, especially in real-time (leading to a universal adhesion to the MP3 format). Moreover, throughout the course of these last years, the number of musical and cinematographic productions available in five channels or more continually increased.In order to maintain an acceptable bit-rate for any given application, it is obvious that a low bit-rate audio coder must exploit the redundancies between audio channels and binaural psychoacoustics. Perceptual audio coding, and more specifically parametric audio coding, offers the possibility of achieving much lower bit-rates by taking into account the limits of human hearing (psychoacoustics). Therefore, this research concentrates on parametric audio coding of more than one audio channel.

Depuis une dizaine d’années, des techniques de codage joint, exploitant les relations et les redondances entre canaux audios, ont été développées afin de réduire davantage la quantité d’information nécessaire à la représentation des signaux multicanaux. Dans cette thèse, nous étudions plus particulièrement le codage des signaux audio stéréo en l’absence d’informations à priori sur la nature des sources en présences, leur nombre et la manière dont elles sont spatialisées. Cette situation correspond à l’immense majorité des enregistrements commerciaux dans l’industrie de la musique et du multimédia de manière générale. Nous étudions des approches paramétrique et signal de la problématique de codage de ces sources, où les deux sont souvent mêlées. Dans ce contexte, trois types d’approches sont utilisés. L’approche paramétrique spatiale consiste à réduire le nombre de canaux audio de la source à coder et à recréer le nombre de canaux d’origine à partir des canaux réduits et de paramètres spatiaux, extraits des canaux d’origine. L’approche signal conserve le nombre de canaux d’origine, mais encode des canaux construits à partir de ces derniers et présentant moins de redondances. Enfin, l’approche mixte introduite dans MPEG USAC utilise un signal audio et un signal résiduel, issu d’une prédiction, et dont les paramètres sont codés conjointement. Dans cette thèse, nous analysons tout d’abord les caractéristiques d’un signal stéréo issu d’un enregistrement commercial et les techniques de production associées. Cette étude nous mène à une réflexion sur les rapports entre les modèles paramétriques d’émetteur, obtenus en analysant les techniques de production des enregistrements commerciaux, et les modèles de récepteur qui sont au coeur du codage spatial paramétrique. A partir de cette mise en perspective nous présentons et étudions les trois approches évoquées plus haut. Pour l’approche purement paramétrique, nous montrons l’impossibilité d’arriver à la transparence pour la majorité des sources audios, nous menons une réflexion sur les représentations paramétriques et proposons des techniques afin de réduire le débit de leurs paramètres et d’améliorer la qualité audio. Ces améliorations passent par une meilleur segmentation du signal audio, basée sur les transitoires, sur des caractéristiques perceptives de certains indices spatiaux et sur une meilleur estimation des indices spatiaux. L’approche mixte étant récemment standardisée dans MPEG USAC, nous l’étudions en détail, puis nous proposons une nouvelle technique de codage qui exploite au mieux l’allocation du résidu aux bandes fréquentielles, lorsque celui-ci n’est pas utilisé sur l’ensemble de la bande passante du signal. Enfin, nous concluons en évoquant l’avenir du codage audio spatial généraliste et mettons l’accent sur l’importance de développer des techniques de classification et de segmentation audio pour optimiser le rapport qualité/débit
During the last ten years, technics for joint coding exploiting relations and redundancies between channels have been developped in order to further reduce the amount of information needed to represent multichannel audio signals.In this document, we focus on the coding of stereo audio signals where prior informations on the nature of sources in presence, their number or the manner they are spatialized is unknown. Such signals are actually the most representative in commercial records of music industry and in multimedia entertainment in general. To address the coding problematic of these signals, we study parametric and signal approaches, where both of them are often mixed.In this context, three types of approaches are used. The spatial parametric approach reduce the number of audio channels of the signal to encode and recreate the original number of channels from reduced channels and spatial parameters extracted from original channels. The signal approach keep the original number of channels, but encode mono signals, built from the combination of the original ones and containing less redundancies. Finally, the hybrid approach introduced in the MPEG USAC standard keep the two channels of a stereo signal, but one is a mono downmix and the other is a residual signal, resulting from a prediction on the downmix, where prediction parameters are encoded as side information.In this document, we first analyse the characteristics of a stereo audio signal coming from a commercial recording and the associated production techniques. This study lead us to consider the relations between the emitter parametric models, elaborated from our analysis of commercial recording production techniques, and the receiver models which are the basis of spatial parametric coding. In the light of these considerations, we present and study the three approaches mentioned earlier. For the parametric approach, we show that transparency cannot be achieved for most of the stereo audio signals, we have a reflection on parametric representations and we propose techniques to improve the audio quality and further reduce the bitrate of their parameters. These improvements are obtained by applying a better segmentation on the signal, based on the significant transient, by exploiting perceptive characteristics of some spatial cues and by adapting the estimation of spatial cues. As the hybrid approach has been recently standardized in MPEG USAC, we propose a full review of it, then we develop a new coding technique to optimize the allocation of the residual bands when the residual is not used on the whole bandwidth of the signal to encode. In the conclusion, we discuss about the future of the general spatial audio coding and we show the importance of developping new technics of segmentation and classification for audio signals to further adapt the coding to the content of the signal

This Master Thesis report was written by Umeå University Engineering Physics student Richard Tano during his thesis work at Ericsson Luleå. Monitoring network quality is of utmost importance to network providers. This can be done with models evaluating QoS (Quality of Service) and conforming to ITU-T Recommendations. When determining video stream quality there is of more importance to evaluatethe QoE (Quality of Experience) to understand how the user perceives the quality. This isranked in MOS (Mean opinion scores) values. An important aspect of determining the QoEis the video content type, which is correlated to the coding complexity and MOS values ofthe video. In this work the possibilities to improve quality estimation models complying to ITU-T study group 12 (q.14) was investigated. Methods were evaluated and an algorithm was developed that applies time series analysis of packet statistics for determination of videostreams MOS scores. Methods used in the algorithm includes a novel assembling of frequentpattern analysis and regression analysis. A model which incorporates the algorithm for usage from low to high bitrates was dened. The new model resulted in around 20% improvedprecision in MOS score estimation compared to the existing reference model. Furthermore an algorithm using only regression statistics and modeling of related statistical parameters was developed. Improvements in coding estimation was comparable with earlier algorithm but efficiency increased considerably.
Detta examensarbete skrevs av Richard Tano student på Umeå universitet åt Ericsson Luleå. Övervakning av nätets prestanda är av yttersta vikt för nätverksleverantörer. Detta görs med modeller för att utvärdera QoS (Quality of Service) som överensstämmer med ITU-T rekommendationer. Vid bestämning av kvaliten på videoströmmar är det mer meningsfullt att utvärdera QoE (Quality of Experience) för att få insikt i hur användaren uppfattar kvaliten. Detta graderas i värden av MOS (Mean opinion score). En viktig aspekt för att bestämma QoE är typen av videoinnehåll, vilket är korrelerat till videons kodningskomplexitet och MOS värden. I detta arbete undersöktes möjligheterna att förbättra kvalitetsuppskattningsmodellerna under uppfyllande av ITU-T studygroup 12 (q.14). Metoder undersöktes och en algoritm utvecklades som använder tidsserieanalys av paketstatistik för uppskattning av videoströmmars MOS-värden. Metoder som ingår i algoritmen är en nyutvecklad frekventa mönster metod tillsammans med regressions analys. En modell som använder algoritmen från låg till hög bithastighet definierades. Den nya modellen gav omkring 20% förbättrad precision i uppskattning av MOS-värden jämfört med existerande referensmodell. Även en algoritm som enbart använder regressionsstatistik och modellerande av statistiska parametrar utvecklades. Denna algoritm levererade jämförbara resultat med föregående algoritm men gav även kraftigt förbättrad effektivitet.

Cette thèse s’inscrit dans le contexte de l’essor des contenus immersifs. Depuis quelques années, les technologies de captation et de restitution sonore immersive se sont développées de manière importante. Ce nouveau contenu a fait naître le besoin de créer de nouvelles méthodes dédiées à la compression audio spatialisée, notamment dans le domaine de la téléphonie et des services conversationnels. Il existe plusieurs manières de représenter l’audio spatialisé, dans cette thèse nous sommes intéressés à l’ambisonie d’ordre 1. Dans un premier temps, nos travaux ont porté sur la recherche d’une solution pour améliorer le codage multimono. Cette solution consiste en un traitement en amont du codec multimono pour décorréler les signaux des composantes ambisoniques. Une attention particulière a été portée à la garantie de continuité du signal entre les trames et à la quantification des métadonnées spatiales. Dans un second temps, nous avons étudié comment utiliser la connaissance de la répartition de l’énergie du signal dans l’espace, aussi appelée image spatiale, pour créer de nouvelles méthodes de codage. L’utilisation de cette image spatiale a permis d’élaborer deux méthodes de compression. La première approche proposée est basée sur la correction spatiale du signal décodé. Cette correction se base sur la différence entre les images spatiales du signal d’origine et du signal décodés pour atténuer les altérations spatiales. Ce principe a été étendu dans une seconde approche à une méthode de codage paramétrique. Dans une dernière partie de cette thèse, plus exploratoire, nous avons étudié une approche de compression par réseaux de neurones en nous inspirant de modèles de compression d’images par auto-encodeur variationnel
This thesis takes place in the context of the spread of immersive content. For the last couple of years, immersive audio recording and playback technologies have gained momentum and have become more and more popular. New codecs are needed to handle those spatial audio formats, especially for communication applications. There are several ways to represent spatial audio scenes. In this thesis, we focused on First Order Ambisonic. The first part of our research focused on improving multi-monocoding by decorrelated each ambisonic signal component before the multi-mono coding. To guarantee signal continuity between frames, efficient quantization new mechanisms are proposed. In the second part of this thesis, we proposed a new coding concept using a power map to recreate the original spatial image. With this concept, we proposed two compressing methods. The first one is a post-processing focused on limiting the spatial distortion of the decoded signal. The spatial correction is based on the difference between the original and the decoded spatial image. This post-processing is later extended to a parametric coding method. The last part of this thesis presents a more exploratory method. This method studied audio signal compression by neural networks inspired by image compression models using variational autoencoders

Questo lavoro di tesi tratta la simulazione dinamica di sistemi HVAC per veicoli tramite il software Simulink. Obiettivo dell'elabora è quello di fornire uno strumento di previsione delle prestazioni di tali sistemi. Ad oggi nella progettazione si cerca di massimizzare l'efficienza e ridurre i consumi del sistema scegliendo i componenti più adeguati dai cataloghi dei costruttori. Cataloghi che riportano le prestazioni statiche di tali componenti, ovvero prestazioni che si riferiscono al comportamento testato in condizioni di progetto su banco prova. Questi sistemi però hanno un comportamento sempre off-design nel loro funzionamento in dinamico, ovvero una volta assemblati. Non è semplice simulare con un modello a parametri concentrati, come quello presentato in questo elaborato, un funzionamento dinamico così complesso ciò nonostante si è arrivati ad un buon risultato. Il modello è stato ottimizzato per la simulazione di sistemi HVAC montati su mezzi movimentazione terra, nello specifico per un miniescavatore. Grazie all'esperienza di tirocinio svolta presso l'azienda Sea-Italia è stato possibile non solo studiare i vari componenti del sistema scelto ma anche eseguire una prima validazione del modello dinamico grazie ai risultati sperimentali rilevati con un test in camera climatica presso l'azienda stessa.

Uno dei principali problemi affrontati dalla cosmologia moderna è la comprensione dell’energia oscura, la cui presenza è necessaria per spiegare l’espansione accelerata dell’Universo. Euclid è una missione spaziale progettata dall’ESA (European Space Agency) per far luce sulla natura dell’energia oscura e sulla formazione ed evoluzione delle galassie. In questa tesi confrontiamo il codice Python CLOE, sviluppato per l’esecuzione di analisi cosmologiche basate sui dati che verranno raccolti dal telescopio Euclid, con un codice che utilizza le CosmoBolognaLib (CBL), un set di librerie C++/Python sviluppate all’Università di Bologna. Il codice CLOE è ancora in fase di sviluppo e in questa tesi facciamo uso della prima versione, v1.0. Abbiamo effettuato un primo confronto sull’andamento delle funzioni cosmologiche, prendendo in considerazione la distanza di diametro angolare D_A(z), la radice della media quadratica (RMS) della densità delle fluttuazioni lineari nell’unità di 8h^(−1) Mpc σ8(z), il fattore di crescita D(z), la distanza comovente r(z), il tasso di crescita f(z), lo spettro di potenza della materia P_m(k) e i multipoli dello spettro di potenza P_l(k). Nel secondo confronto abbiamo stimato, tramite la catene di Markov Monte Carlo (MCMC), i seguenti parametri cosmologici: il parametro di densità della materia Ω_m, il parametro di densità della materia barionica Ω_b, la costante di Hubble H_0, l’ampiezza scalare delle perturbazioni primordiali A_s, l’indice spettrale n_s e il parametro σ8 a z= 0. I risultati ottenuti evidenziano un buon accordo tra i due codici analizzati, essendo le discrepanze percentuali delle funzioni cosmologiche comprese nell’intervallo tra -0.02%e 2%. Nel caso della stima dei parametri cosmologici attraverso le MCMC si sono evidenziate differenze più significative, le cui cause non sono ancora chiare, saranno necessarie ulteriori analisi per fare luce su questo aspetto.

In questa Tesi di Laurea viene presentata un’interessante esperienza di implementazione numerica: lo sviluppo di un codice agli elementi finiti in grado di calcolare, verificare e ottimizzare edifici industriali in acciaio. Al giorno d’oggi, la tendenza delle imprese nel campo dell’ingegneria strutturale ed in particolare nel campo dell’edilizia industriale, è quella della specializzazione. E’ sempre più frequente, ad esempio nel campo dei capannoni industriali, che le aziende concentrino la loro attività solo su determinate tipologie costruttive, sulle quali ottimizzano il lavoro riducendo al massimo i tempi di progettazione, di costruzione e abbassando il prezzo. Il mondo dei programmi di calcolo, per la maggior parte, sembra aver preso una direzione di sviluppo opposta. Le case di produzione software mettono a disposizione dei progettisti strumenti sempre più raffinati, capaci di modellare dettagliatamente qualsiasi tipo di struttura, materiale, azione statica o dinamica; spesso questi programmi contengono anche un codice integrato CAD per il disegno della struttura e altri tools per lasciare all’utente la più grande libertà di azione possibile. Se da un lato questi strumenti danno al progettista la possibilità di una modellazione sempre più dettagliata, dall’altra parte hanno il limite di essere poco pratici per un tipo di progettazione standardizzato. Spesso quello di cui le imprese hanno bisogno è invece un programma creato ‘ad hoc’ per le loro attività che, grazie all’inserimento di pochi parametri, possa garantire una progettazione rapida e magari gestire non solo la fase di calcolo, ma anche quella di verifica e di ottimizzazione. In quest’ottica si inserisce lo sviluppo del codice eseguito in questa tesi. L’esposizione si articola in quattro parti. La prima, introduttiva, è dedicata alla descrizione delle tipologie di edifici monopiano in acciaio maggiormente diffuse, dei diversi tipi materiale, dei principali aspetti della normativa per queste costruzioni. Viene inoltre descritta la tipologia costruttiva implementata nel codice sviluppato. Il secondo capitolo è dedicato alla descrizione del metodo agli elementi finiti, esponendone i fondamenti teorici e le principali fasi della costruzione di un codice di calcolo numerico per elementi monodimensionali. Nella terza parte è illustrato il codice sviluppato. In particolare vengono dettagliatamente descritti i moduli di generazione del modello, del solutore, del post-processore in grado di eseguire le verifiche secondo le normative vigenti, e quello dedicato all’ottimizzazione strutturale. In fine, nell’ultimo capitolo viene illustrato un esempio progettuale con il quale si è potuta effettuare la validazione del codice confrontando i risultati ottenuti con quelli di riferimento forniti da programmi attualmente in commercio. La presente dissertazione non mira alla “certificazione” di un software che sia in grado di fare calcoli complessi nell’ambito dell’ingegneria strutturale, ma lo scopo è piuttosto quello di affrontare le problematiche e gestire le scelte operative che riguardano la scrittura di un codice di calcolo. Programmatori non si nasce, ma si diventa attraverso anni di esperienza che permettono di acquisire quella sensibilità numerica che è definibile come una vera e propria “arte”. Ed è in questa direzione che si è svolta la Tesi, ovvero comprendere prima di tutto l’atteggiamento da assumere nei confronti di un elaboratore elettronico e, solo successivamente, passare ad un utilizzo consapevole per scopi progettuali.

The first goal of diploma thesis is to study the basic principles of video signal compression. Introduction to techniques used to reduce irrelevancy and redundancy in the video signal. The second goal is, on the basis of information about compression tools, implement the individual compression tools in the programming environment of Matlab and assemble simple model of the video codec. Diploma thesis contains a description of the three basic blocks, namely - interframe coding, intraframe coding and coding with variable length word - according the standard MPEG-2.

En Guadeloupe, la prévalence du diabète est deux fois plus élevée qu'en France hexagonale, avec une prédominance féminine. En dehors des facteurs environnementaux, la vitamine D et certains polymorphismes de gènes impliqués dans son métabolisme seraient associés à un risque de pathologies métaboliques. Les androgènes sont également associés aux anomalies du métabolisme glucidique, soit directement, soit via leur transporteur SHBG (Sex Hormone Binding Protein). Nous avons émis les hypothèses de recherche suivantes: 11 le statut en vitamine D et les polymorphismes des gènes impliqués dans son métabolisme pourraient être associés aux paramètres métaboliques chez des sujets Afro-Caribéens (AC). Ils expliqueraient l'importance des pathologies cardiométaboliques en Guadeloupe. 2/ la SHBG pourrait être associée aux anomalies du métabolisme glucidique, indépendamment des stéroïdes sexuels. Nos travaux sont présentés dans 4 études. Nous avons mis en évidence une prévalence élevée du déficit en vitamine D chez les sujets AC diabétiques de type 2. Nous avons trouvé une association significative entre le statut vitaminique D et le risque cardiométabolique chez ces sujets, mais aussi dans une population de sujets en hémodialyse chronique. Les polymorphismes des gènes impliqués dans le métabolisme de la vitamine D sont aussi associés à ce risque. Chez les sujets dysmétaboliques, une relation entre la SHBG, la graisse intra-hépatique, les hépatokines et les paramètres métaboliques a été mise en évidence, indépendamment des stéroïdes sexuels. En conclusion, la vitamine D et la SHBG pourraient être des marqueurs d'intérêt pour le dépistage des sujets à haut risque cardiométabolique
In Guadeloupe, the prevalence of diabetes is two times higher than in Metropolitan France, with a female predominance. Apart from environmental factors, vitamin D and polymorphisms of genes involved in vitamin D metabolism would be associated with increased risk of metabolic diseases. Androgens are also associated with abnormal glucose metabolism, directly or through the Sex Hormone Binding Protein (SHBG). Our main hypotheses were: 11 Vitamin D status and polymorphisms of genes involved in its metabolism may be associated with metabolic pammeters in Afro-Caribbean (AC). They could explain the importance of cardiometabolic diseases in Guadeloupe. 2/ SHBG may be associated with abnormal glucose metabolism, independently of sex steroids. Our research is presente

碩士
國立中央大學
通訊工程研究所
98
The Parametric Stereo (PS) audio coding is an audio coding object of High Efficiency Advanced Audio Coding version 2 (HE-AAC v2) which was standardized by ISO/MPEG in 2004. Traditional audio codec, e.g. MP3 or AAC, utilize “Psycoaustic Model” and “Masking Effect” to achieve high compression efficiency. However, they mainly process the signal with single channel. Different from traditional audio codec, the PS audio coding incorporates the characteristics of two channels, to extract spatial parameters and to down-mixes stereo signals into a mono signal. The PS can save almost half data size which provides great help in storage and transmission. Nevertheless, the complexity of PS decoder is nearly twice larger than that of PS encoder, which causes a serious problem in implementing PS on portable devices. Therefore, this thesis proposes a modified PS coding scheme to reduce the complexity of decoder. The encoder extracts and transmits the additional residual parameters from the residual signal and the mono signal. On the contrary, the decoder reconstructs the residual signal by the mono signal and the transmitted residual parameters. In addition, we detect the existence of transient signal and measure the artifact of reconstructed residual signal. Finally, “Energy compesated algorithm” is proposed to reduce the artifact produced by the transient signal. The proposed scheme can improve the Objective Difference Grade (ODG) of audio quality measurement “EAQUAL” with 0.6 score. Combining with audio coder AAC, the modified PS coding scheme still maintains a good performance at low coding bitrates.

碩士
國立交通大學
多媒體工程研究所
97
Parametric Stereo Coding (PS) tool in HE-AAC v2 is an efficient tool for low bit-rate stereo signal compression. The PS encoder combines the stereo signal into a monaural signal and extracts the stereo parameters. Decoder can reconstruct the similar stereo signal from the monaural signal with the stereo parameters. This thesis considers the PS design from three aspects: the design of stereo parameter extraction, the method of interpolation in downmix/upmix, and the downmix/upmix method with decorrelated signals. Because the output monaural signal of PS can be encoded by the HE-AAC, these methods are integrated in the NCTU-HEAAC for implementation. Finally according to the experiments result of objective (ODG) and subjective (MUSHRA) quality measurement, the proposed methods can reach a better audio quality than previously

碩士
國立中央大學
通訊工程研究所
99
Nowadays the multimedia applications of the 3D virtual reality are more and more popular. Although most applications focus on 3D video, the combination of 3D video and audio processing can enrich the experience of users. In this thesis, we propose an object-based audio rendering system (OARS) for 3D applications, such as first person shooter (FPS) games. With the proposed system, users are able to locate the objects, whether it is static or in motion. Since the audio objects may be in remote sites that are connected over Internet in many applications, the bitrate reduction is still critical. In this work, the system consists of the audio analysis part and synthesis part. In the audio analysis part, we utilize the parametric coding technique to generate spatial parameters, which include the time difference and the intensity difference for an object and loudspeakers, for rate reduction while keeping the spatial information. In the audio synthesis part, we reconstruct multi-channel audio outputs by integrating an audio signal and the spatial parameters. We evaluate the system performance by analyzing the spectrum of processed audios and subjective listening tests. Based on the modified ITU-R seven-grade (-3 to 3) subjective quality evaluation, our proposed system scores 1.49 on average for static audio objects and 1.31 for moving objects.

碩士
國立交通大學
資訊科學與工程研究所
94
Parametric Stereo Coding (PS) tool is an efficient tool for low bit-rate audio compression. It combines the stereo signal into a monaural downmix signal and extracts the parameters of stereo signal as the stereo image. By above information, the PS decoder can reconstruct the stereo signal. Moreover, the stereo parameters requires only a small overhead, the bit usage is substantially decreased. For the design of PS tool, the thesis will focus on the T/F stereo parameter extraction which controls the delivery of stereo parameters both in time domain and frequency domain, and downmix method for generating the monaural downmix signal. Because the PS tool can be jointly used with the HE-AAC for downmix signal encoding, these methods are integrated in the NCTU-HEAAC for implementation and to verify the coding quality. The extensive experiments are executed for both subjective and objective quality measurement in the end of the thesis.

碩士
國立交通大學
多媒體工程研究所
95
Due to the aliasing-free properties, the complex quadrature mirror filter (QMF) bank has been used in MPEG-4 audio standard on SBR, parametric, and surrounds coding. The high complexity overhead from the complex QMF bank and the complex data processing in the decoder leads to the development of low power decoder which adopts the real QMF bank as the basic building module to reduce the complexity. However the artifacts from the aliasing in the real QMF bank are the major concern. This paper studies the artifacts from the real QMF bank and proposes a novel QMF bank design to achieve both low complexity and high quality. Also, this paper applies the novel QMF bank to develop the high-quality and low-power SBR, parametric, and MPEG surround decoders and shows the merits in complexity and quality.