Interpolating Expressions in Unit Selection

ended abstract reports on on-going work exploring the addition of a certain degree of control over expressivity in a unit selection context. Rather than merely choosing one unit selection voice database in order to determine the expression contained in the generated speech, we use technology from the voice conversion domain to flexibly interpolate between two voice databases. This provides us with the possibility to generate a continuum of expressive tones between the two extremes defined by the two voice databases. Spectral interpolation algorithm The spectral interpolation method employed here has previously been used to interpolate diphone voices with different vocal effort [1]. The method is based on a linear predictive coding (LPC) paradigm of speech representation, using line spectral frequencies (LSFs) as a representation of LPC coefficients with good interpolation properties [2]. The method works as follows. Two utterances with the same phoneme chain are mapped to each othe...

• Type: Conference
• Year: 2007
• Where: ACII
• Authors: Marc Schröder
• Keywords: ACII 2007, Applied Computing, Spectral Interpolation, Spectral Interpolation Method, Voice Databases