DeepA: A Deep Neural Analyzer For Speech And Singing Vocoding
This work addresses the need for more accurate and generalizable vocoder-like parameters for tasks like speech synthesis and voice conversion, but it is incremental as it builds on existing VAE architectures.
The paper tackles the problem of conventional vocoders lacking generalizability across different audio types, such as speech to singing, by proposing DeepA, a deep neural analyzer that extracts interpretable F0 and timbre/aperiodicity encodings. It shows that DeepA improves F0 estimation over the conventional WORLD vocoder, though specific numerical gains are not provided.
Conventional vocoders are commonly used as analysis tools to provide interpretable features for downstream tasks such as speech synthesis and voice conversion. They are built under certain assumptions about the signals following signal processing principle, therefore, not easily generalizable to different audio, for example, from speech to singing. In this paper, we propose a deep neural analyzer, denoted as DeepA - a neural vocoder that extracts F0 and timbre/aperiodicity encoding from the input speech that emulate those defined in conventional vocoders. Therefore, the resulting parameters are more interpretable than other latent neural representations. At the same time, as the deep neural analyzer is learnable, it is expected to be more accurate for signal reconstruction and manipulation, and generalizable from speech to singing. The proposed neural analyzer is built based on a variational autoencoder (VAE) architecture. We show that DeepA improves F0 estimation over the conventional vocoder (WORLD). To our best knowledge, this is the first study dedicated to the development of a neural framework for extracting learnable vocoder-like parameters.