Sequence-to-Sequence Acoustic Modeling for Voice Conversion

TL;DR

This paper introduces SCENT, a neural network for voice conversion that aligns source and target speech features using attention, enabling simultaneous conversion of acoustic features and durations with improved quality over traditional methods.

Contribution

The paper presents a novel sequence-to-sequence neural network model for voice conversion that effectively aligns and converts speech features and durations in a unified framework.

Findings

Outperforms GMM and DNN baseline models in objective and subjective tests.

Achieves better duration conversion than conventional methods.

Outperforms previous state-of-the-art in Voice Conversion Challenge 2018.

Abstract

In this paper, a neural network named Sequence-to-sequence ConvErsion NeTwork (SCENT) is presented for acoustic modeling in voice conversion. At training stage, a SCENT model is estimated by aligning the feature sequences of source and target speakers implicitly using attention mechanism. At conversion stage, acoustic features and durations of source utterances are converted simultaneously using the unified acoustic model. Mel-scale spectrograms are adopted as acoustic features which contain both excitation and vocal tract descriptions of speech signals. The bottleneck features extracted from source speech using an automatic speech recognition (ASR) model are appended as auxiliary input. A WaveNet vocoder conditioned on Mel-spectrograms is built to reconstruct waveforms from the outputs of the SCENT model. It is worth noting that our proposed method can achieve appropriate duration conversion which is difficult in conventional methods. Experimental results show that our proposed method obtained better objective and subjective performance than the baseline methods using Gaussian mixture models (GMM) and deep neural networks (DNN) as acoustic models. This proposed method also outperformed our previous work which achieved the top rank in Voice Conversion Challenge 2018. Ablation tests further confirmed the effectiveness of several components in our proposed method.