# Text-to-speech synthesis from dark data with evaluation-in-the-loop data selection

arXiv: https://arxiv.org/abs/2210.14850 · HTML: https://mlsanity.com/paper/2210.14850/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 26 Oct 2022.

## Built on by (most cited first, 4 of 4)

- [AudioJudge: Understanding What Works in Large Audio Model Based Speech Evaluation](https://mlsanity.com/paper/2507.12705/) (2025)
- [Learning to Speak from Text: Zero-Shot Multilingual Text-to-Speech with Unsupervised Text Pretraining](https://mlsanity.com/paper/2301.12596/) (2023)
- [How Generative Spoken Language Modeling Encodes Noisy Speech: Investigation from Phonetics to Syntactics](https://mlsanity.com/paper/2306.00697/) (2023)
- [J-CHAT: Japanese Large-scale Spoken Dialogue Corpus for Spoken Dialogue Language Modeling](https://mlsanity.com/paper/2407.15828/) (2024)

## Abstract (arXiv)

This paper proposes a method for selecting training data for text-to-speech (TTS) synthesis from dark data. TTS models are typically trained on high-quality speech corpora that cost much time and money for data collection, which makes it very challenging to increase speaker variation. In contrast, there is a large amount of data whose availability is unknown (a.k.a, "dark data"), such as YouTube videos. To utilize data other than TTS corpora, previous studies have selected speech data from the corpora on the basis of acoustic quality. However, considering that TTS models robust to data noise have been proposed, we should select data on the basis of its importance as training data to the given TTS model, not the quality of speech itself. Our method with a loop of training and evaluation selects training data on the basis of the automatically predicted quality of synthetic speech of a given TTS model. Results of evaluations using YouTube data reveal that our method outperforms the conventional acoustic-quality-based method.