Recognition-Synthesis Based Non-Parallel Voice Conversion with Adversarial Learning

TL;DR

This paper introduces an adversarial learning approach for non-parallel voice conversion that separates speaker identity from linguistic content, resulting in more natural and speaker-similar converted speech.

Contribution

It proposes a novel recognition-synthesis framework with adversarial training and GANs to improve speaker independence and feature quality in voice conversion.

Findings

Achieved higher similarity than the top baseline in Voice Conversion Challenge 2018.

Effectively separates speaker characteristics from linguistic content.

Utilizes a pre-training and fine-tuning strategy for model optimization.

Abstract

This paper presents an adversarial learning method for recognition-synthesis based non-parallel voice conversion. A recognizer is used to transform acoustic features into linguistic representations while a synthesizer recovers output features from the recognizer outputs together with the speaker identity. By separating the speaker characteristics from the linguistic representations, voice conversion can be achieved by replacing the speaker identity with the target one. In our proposed method, a speaker adversarial loss is adopted in order to obtain speaker-independent linguistic representations using the recognizer. Furthermore, discriminators are introduced and a generative adversarial network (GAN) loss is used to prevent the predicted features from being over-smoothed. For training model parameters, a strategy of pre-training on a multi-speaker dataset and then fine-tuning on the source-target speaker pair is designed. Our method achieved higher similarity than the baseline model that obtained the best performance in Voice Conversion Challenge 2018.