ChristianYang/Env-TTS-Clean

环境感知文本到语音训练语料库（清洁版本）。每行数据配对四个短24 kHz单声道FLAC音频剪辑及其对齐的转录：一个环境样本（不同说话者，相同声学场景）、一个说话者参考（与目标话语相同说话者）、说话者参考的增强版本（使用MossFormer2语音增强技术）以及要合成的目标语音。这样，模型可以学习生成具有指定语音和指定环境的话语。此版本取代了早期的Env-TTS-SD-Corpus，具有更丰富的模式、更高的采样率、所有三个上下文的每剪辑ASR以及两个额外的会议来源（AMI和AliMeeting）。数据集包含189,918行数据，总语音时长为1,542,868.6秒（约428.6小时），总音频时长为4,507,421.5秒（约1,252.1小时），来源包括M3SD、AISHELL-4、AliMeeting、AMI、MSDWILD和CHiME-6等语料库。

Environment-aware text-to-speech training corpus (clean release). Each row pairs four short 24 kHz mono FLAC clips with aligned transcripts: an environment sample (different speaker, same acoustic scene), a speaker reference (same speaker as the target utterance), a speaker-enhanced copy of the reference (MossFormer2 speech enhancement), and the target speech to synthesise, so a model can learn to generate an utterance with both a specified voice and a specified environment. This release supersedes the earlier Env-TTS-SD-Corpus with a richer schema, higher sample rate, per-clip ASR for all three contexts, and two additional meeting sources (AMI, AliMeeting).