VSE knjižnice (vzajemna bibliografsko-kataložna baza podatkov COBIB.SI)
  • Towards a mathematical proof of the speech alignment paradox
    Sündermann, David ; Smrekar, Jaka ; Höge, Harald
    Applying a recently presented text-independent speech alignment technique based on unit selection to the training of a voice conversion system suggested that the more training data was available, the ... less speaker-specific information was learned. This paradoxical effect contradicts the experience we have from other corpus-based applications as speech recognition, synthesis or translation. There, the performance usually gains with increasing amount of data. In this paper, we investigate this paradox by means of objective tests and derive a mathematical model of the underlying stochastic process.
    Vir: Advances in speech technology : proceedings : Thirteenth International Workshop, Maribor, Slovenia, July 6-7, 2006 (9 str.)
    Vrsta gradiva - prispevek na konferenci
    Leto - 2007
    Jezik - angleški
    COBISS.SI-ID - 14369369

Vnos na polico