humanify/Env-TTS-SD-Corpus

环境感知文本到语音训练语料库。每一行数据包含三个16 kHz单声道FLAC音频片段及其转录文本：一个环境样本（不同说话人，相同声学场景）、一个说话人参考（相同说话人，可选声学增强）和目标语音，因此TTS模型可以学习合成具有指定语音和指定环境的语音。

Environment-aware text-to-speech training corpus. Each row pairs three short 16 kHz mono FLAC clips with a transcript: an environment sample (different speaker, same acoustic scene), a speaker reference (same speaker, optionally with augmented acoustics), the target speech, so a TTS model can learn to synthesise a target utterance with both a specified voice and a specified environment.