Multi-SpectroGAN: High-Diversity and High-Fidelity Spectrogram Generation with Adversarial Style Combination for Speech Synthesis

TL;DR

Multi-SpectroGAN introduces a novel adversarial training method for speech synthesis that generates high-diversity, high-fidelity spectrograms by learning style embeddings and combining styles without reconstruction loss.

Contribution

The paper proposes Multi-SpectroGAN, a new GAN-based TTS model that trains solely with adversarial feedback and introduces adversarial style combination for style generalization.

Findings

Achieves high naturalness in spectrogram synthesis comparable to ground-truth.

Generates diverse speech styles by controlling style embeddings.

Outperforms existing models in style generalization and spectrogram quality.

Abstract

While generative adversarial networks (GANs) based neural text-to-speech (TTS) systems have shown significant improvement in neural speech synthesis, there is no TTS system to learn to synthesize speech from text sequences with only adversarial feedback. Because adversarial feedback alone is not sufficient to train the generator, current models still require the reconstruction loss compared with the ground-truth and the generated mel-spectrogram directly. In this paper, we present Multi-SpectroGAN (MSG), which can train the multi-speaker model with only the adversarial feedback by conditioning a self-supervised hidden representation of the generator to a conditional discriminator. This leads to better guidance for generator training. Moreover, we also propose adversarial style combination (ASC) for better generalization in the unseen speaking style and transcript, which can learn latent representations of the combined style embedding from multiple mel-spectrograms. Trained with ASC and feature matching, the MSG synthesizes a high-diversity mel-spectrogram by controlling and mixing the individual speaking styles (e.g., duration, pitch, and energy). The result shows that the MSG synthesizes a high-fidelity mel-spectrogram, which has almost the same naturalness MOS score as the ground-truth mel-spectrogram.