Christophe d'Alessandro
Voice quality modification using
periodic-aperiodic decomposition
and spectral processing of the voice
source signal
Abstract:Voice quality is currently a key issue in speech synthesis. On the one hand, the
lack of realistic intra-speaker voice quality variation results in poor naturalness in
synthesis methods using small corpora and signal processing (e.g diphone synthesis). On the other hand, voice quality mismatches is one of the main source of
concern for methods based on large corpora and labelling (e.g. word or subword
units concatenation systems,like CHATR). A new method for voice quality modification is designed. It takes advantage of a spectral theory for voice source signal
representation. An algorithm based on periodic-aperiodic decomposition and spectral processing (using the short-term Fourier transform) is described. The use of
adaptive inverse filtering in this framework is also discussed. Applications of this
algorithm may include: pre-processing of speech corpora, modification of voice
quality parameters together with intonation in synthesis, voice transformation.
Some experiments were perfomed, showing convincing voice quality modifications
for various speakers.