TextrolSpeech: A Text Style Control Speech Corpus With Codec Language Text-to-Speech Models

TL;DR

This work addresses the challenge of controllable text to speech by introducing TextrolSpeech, a large-scale open dataset of speech paired with natural language style prompts across five style factors, and Salle, a codec-based TTS framework that uses discrete acoustic tokens guided by text prompts. Salle employs a two-stage architecture with an autoregressive first layer and parallel second to n layers, enabling in-context style control via a text style prompt language model. Experiments show Salle outperforming a state-of-the-art PromptTTS baseline in style-factor accuracy and achieving competitive MOS scores, though overall audio quality is limited by data scale and task complexity. By providing open datasets and a strong baseline, the work lays groundwork for more robust, text-driven TTS with diverse emotional expression.

Abstract

Recently, there has been a growing interest in the field of controllable Text-to-Speech (TTS). While previous studies have relied on users providing specific style factor values based on acoustic knowledge or selecting reference speeches that meet certain requirements, generating speech solely from natural text prompts has emerged as a new challenge for researchers. This challenge arises due to the scarcity of high-quality speech datasets with natural text style prompt and the absence of advanced text-controllable TTS models. In light of this, 1) we propose TextrolSpeech, which is the first large-scale speech emotion dataset annotated with rich text attributes. The dataset comprises 236,220 pairs of style prompt in natural text descriptions with five style factors and corresponding speech samples. Through iterative experimentation, we introduce a multi-stage prompt programming approach that effectively utilizes the GPT model for generating natural style descriptions in large volumes. 2) Furthermore, to address the need for generating audio with greater style diversity, we propose an efficient architecture called Salle. This architecture treats text controllable TTS as a language model task, utilizing audio codec codes as an intermediate representation to replace the conventional mel-spectrogram. Finally, we successfully demonstrate the ability of the proposed model by showing a comparable performance in the controllable TTS task. Audio samples are available at https://sall-e.github.io/

Figures (2)

• Figure 1: Two sets of word cloud exemplars
• Figure 2: The overall architecture of Salle