potsawee/speecheval-advanced-v1
收藏Hugging Face2025-02-11 更新2025-02-15 收录
下载链接：
https://hf-mirror.com/datasets/potsawee/speecheval-advanced-v1
下载链接
链接失效反馈官方服务：
资源简介：
SpeechIn SpeechOut评估数据集，用于评估端到端音频大型语言模型的先进语音生成能力。数据集中的指令是经过人工策划的，并得到了ChatGPT的辅助。指令音频是通过kokoro TTS获得的，输出音频是通过GPT-4o-Audio生成的，该输出音频作为基线，用于与其他模型进行性能比较。
SpeechIn SpeechOut Evaluation Dataset, designed to assess the advanced speech generation capabilities of end-to-end audio large language models. The instructions in the dataset are curated manually with the assistance of ChatGPT. The instruction audio is obtained from kokoro TTS, and the output audio is generated from GPT-4o-Audio, which serves as a baseline for performance comparison against other models.
提供机构：
potsawee
