Any-to-Many Voice Conversion with Location-Relative Sequence-to-Sequence Modeling

TL;DR

This paper introduces an any-to-many voice conversion method using a sequence-to-sequence model with location-relative attention, leveraging text supervision and a bottleneck feature extractor for improved naturalness and speaker similarity.

Contribution

It proposes a novel any-to-many voice conversion framework combining a phoneme recognizer, bottleneck features, and a location-relative attention seq2seq model, enabling high-quality non-parallel voice conversion.

Findings

Superior naturalness and speaker similarity in voice conversion results

Effective alignment of long sequences using MoL attention and down-sampling

The approach can be extended to any-to-any voice conversion with high performance

Abstract

This paper proposes an any-to-many location-relative, sequence-to-sequence (seq2seq), non-parallel voice conversion approach, which utilizes text supervision during training. In this approach, we combine a bottle-neck feature extractor (BNE) with a seq2seq synthesis module. During the training stage, an encoder-decoder-based hybrid connectionist-temporal-classification-attention (CTC-attention) phoneme recognizer is trained, whose encoder has a bottle-neck layer. A BNE is obtained from the phoneme recognizer and is utilized to extract speaker-independent, dense and rich spoken content representations from spectral features. Then a multi-speaker location-relative attention based seq2seq synthesis model is trained to reconstruct spectral features from the bottle-neck features, conditioning on speaker representations for speaker identity control in the generated speech. To mitigate the difficulties of using seq2seq models to align long sequences, we down-sample the input spectral feature along the temporal dimension and equip the synthesis model with a discretized mixture of logistic (MoL) attention mechanism. Since the phoneme recognizer is trained with large speech recognition data corpus, the proposed approach can conduct any-to-many voice conversion. Objective and subjective evaluations show that the proposed any-to-many approach has superior voice conversion performance in terms of both naturalness and speaker similarity. Ablation studies are conducted to confirm the effectiveness of feature selection and model design strategies in the proposed approach. The proposed VC approach can readily be extended to support any-to-any VC (also known as one/few-shot VC), and achieve high performance according to objective and subjective evaluations.