Mobile telephony is omnipresent in modern society. Currently there are more ac- tive mobile phone subscriptions than inhabitants on the planet. One technological building block facilitating this modern technology is speech coding. These algorithms reduce the amount of data that needs to be transmitted while enabling an effortless communication due to the high intelligibility of the encoded speech signal. This thesis focuses on improving various aspects of speech coding in order to enhance the quality of state-of-the-art algorithms. In addition to the proposed improvements, state of the art speech coding algorithms are modified in order to be applicable for modern scenarios such as Wireless Acoustic Sensor Network (WASN). This thesis depicts a mathematical formalism for the the application of windowing functions to time domain signals, as applied by codecs based on the Code Excited Linear Prediction (CELP) paradigm and modifications of that windowing are deduced. Moreover, it covers a contribution to efficiently enhance the speech signal by including Wiener filtering in a speech coding algorithm, without added computational com- plexity or delay. Furthermore, a hybrid quantization scheme is presented, which can facilitate the muffled sounding nature of the quantization and coding scheme of current state of the art speech coding approaches. In addition to that, we show how to modify current speech coding paradigms to gain from multi-device scenarios. These topics are complemented by a low-complexity, low-power acoustic front end including Voice Activity Detection (VAD) and Keyword Spotter (KS). ; Mobiltelefone sind in der modernen Gesellschaft allgegenwärtig. Momentan gibt es mehr Mobilfunkverträge als Menschen auf unserem Planeten. Ein wichtiger Baustein der diese moderne Technologie ermöglicht sind Sprachcodierverfahren. Diese Algo- rithmen verringern die zu übertragene Datenmenge bei gleichzeitiger Gewährleistung einer mühelosen Kommunikation dank hoher Sprachverständlichkeit der kodierten Signale. Der ...


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Contributions to speech and audio coding for single- and multi-device scenarios ; Beiträge zur Sprach- und Audiocodierung für Einzel- und Mehrgerätszenarien


    Beteiligte:

    Erscheinungsdatum :

    2021-01-01


    Medientyp :

    Hochschulschrift


    Format :

    Elektronische Ressource


    Sprache :

    Englisch


    Schlagwörter :

    Klassifikation :

    DDC:    629





    EINZEL-ENDFÜHRERSTAND-LOKOMOTIVE

    RIESS MANUEL | Europäisches Patentamt | 2024

    Freier Zugriff

    Einzel-Endführerstand-Lokomotive

    RIESS MANUEL | Europäisches Patentamt | 2024

    Freier Zugriff

    Improved Einzel lenses

    Hart, R. K. | NTRS | 1976