Pranavz/emilia-s2s-mimi-q8-named

这是一个名为Emilia S2S Mimi Q8 Named Speakers的数据集，它是一个TinyAya codec-tokenized文本到语音（TTS）数据集，从特定路径的parquet文件构建而成。数据行是仅包含Mimi codec的示例，训练文本字段已预置了面向用户的说话者前缀，例如Ira: transcript...。数据集提供了说话者映射关系：SP_SP010对应Ira:，SP_SP061对应Aisha:，SP_SP073对应Siya:，SP_SP079对应Zoya:。此外，参考条件在此数据集被禁用，因此不包含任何ref_*字段。This is a dataset named Emilia S2S Mimi Q8 Named Speakers, which is a TinyAya codec-tokenized text-to-speech (TTS) dataset built from parquet files at a specific path. The rows are Mimi codec-only examples, and the training text field already includes user-facing speaker prefixes, such as Ira: transcript.... The dataset provides speaker mappings: SP_SP010 maps to Ira:, SP_SP061 maps to Aisha:, SP_SP073 maps to Siya:, and SP_SP079 maps to Zoya:. Additionally, reference conditioning is disabled for this artifact, so no ref_* fields are emitted.