Text-to-Speech Dataset

本数据集由aiXplain, Inc.创建，旨在为文本到语音（TTS）模型提供高质量训练数据。数据集包含德语、英语、普通话、意大利语、法语和西班牙语六种语言，每种语言目标生成30小时音频。创建过程包括样本选择、自动化录音和质量保证，确保数据覆盖所有目标语言的音素，并通过自动语音识别（ASR）模型验证录音准确性。该数据集适用于提升TTS技术在辅助技术、内容创作和客户服务等领域的应用。

This dataset was developed by aiXplain, Inc. to provide high-quality training data for text-to-speech (TTS) models. It covers six languages: German, English, Mandarin, Italian, French, and Spanish, with a target of 30 hours of audio per language. The creation pipeline includes sample selection, automated audio recording, and quality assurance procedures, which ensure that the dataset encompasses all phonemes of the target languages and verify the accuracy of the recorded audio via automatic speech recognition (ASR) models. This dataset is applicable to advancing the deployment of TTS technology in fields such as assistive technology, content creation, and customer service.