Emova-ollm/emova-sft-speech-231k

EMOVA-SFT-Speech-231K是一个用于全模态指令调整和情感口语对话的全面数据集。该数据集通过使用文本到语音（TTS）工具转换现有的文本和视觉指令数据集而创建。EMOVA-SFT-Speech-231K是EMOVA-Datasets集合的一部分，用于EMOVA模型系列的第三阶段——全模态指令调整。该数据集保存了EMOVA语音对话数据的单独副本，因此是EMOVA-SFT-4M数据集的子集。相应的评估数据集在EMOVA-SFT-Speech-Eval数据集中维护。我们使用EMOVA语音分词器提取语音单元。

EMOVA-SFT-Speech-231K is a comprehensive dataset curated for omni-modal instruction tuning and emotional spoken dialogue. This dataset is created by converting existing text and visual instruction datasets via Text-to-Speech (TTS) tools. EMOVA-SFT-Speech-231K is part of the EMOVA-Datasets collection and is used in the Stage 3 - Omni-modal instruction tuning of the EMOVA family of models. This dataset saves a separate copy of the EMOVA speech conversation data, and thus, is a subset of the EMOVA-SFT-4M dataset. The corresponding evaluation dataset is maintained in the EMOVA-SFT-Speech-Eval dataset. We extract speech units using the EMOVA Speech Tokenizer.