Hierarchical Representation of Prosody for Statistical Speech Synthesis
This work addresses the lack of unified representation schemes for prosodic structure in text-to-speech synthesis, offering an unsupervised approach that could improve speech synthesis systems, though it appears incremental as it builds on existing analysis techniques.
The paper tackled the problem of estimating and modeling prosodic prominences and boundaries in speech synthesis by proposing an unsupervised unified method using scale-space analysis based on continuous wavelet transform. The results showed that this method is comparable to the best supervised annotation methods, as evaluated on the Boston University Radio News corpus.
Prominences and boundaries are the essential constituents of prosodic structure in speech. They provide for means to chunk the speech stream into linguistically relevant units by providing them with relative saliences and demarcating them within coherent utterance structures. Prominences and boundaries have both been widely used in both basic research on prosody as well as in text-to-speech synthesis. However, there are no representation schemes that would provide for both estimating and modelling them in a unified fashion. Here we present an unsupervised unified account for estimating and representing prosodic prominences and boundaries using a scale-space analysis based on continuous wavelet transform. The methods are evaluated and compared to earlier work using the Boston University Radio News corpus. The results show that the proposed method is comparable with the best published supervised annotation methods.