Vikhrmodels/ToneSpeak
收藏Hugging Face2025-05-23 更新2025-05-31 收录
下载链接:
https://hf-mirror.com/datasets/Vikhrmodels/ToneSpeak
下载链接
链接失效反馈官方服务:
资源简介:
ToneSpeak是一个大型俄语音频数据集,包含详细的语调、音色和情感特征的描述。每个音频片段都包括文本翻译、详细的语调情感描述、声音名称和音频文件的链接。数据集分为训练集和验证集,收集过程中使用了GPT-4.1 mini生成文本和提示,以及GPT-4o mini TTS进行语音合成,共有10种不同的声音。
ToneSpeak is a large Russian audio dataset with detailed descriptions of intonation, timbre, and emotional characteristics. Each audio clip includes a text transcription, detailed intonation and emotion descriptions, voice name, and a link to the audio file. The dataset is split into training and validation sets, and the collection process involved using GPT-4.1 mini to generate texts and prompts, as well as GPT-4o mini TTS for speech synthesis, with a total of 10 different voices.
提供机构:
Vikhrmodels



