FBK-MT/fama-data
收藏Hugging Face2025-09-10 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/FBK-MT/fama-data
下载链接
链接失效反馈官方服务：
资源简介：
FAMA数据集是一个包含英语和意大利语的大型开源语音基础模型训练数据集，用于自动语音识别（ASR）和语音翻译（ST）任务。该数据集由多个子数据集组成，包括CommonVoice、CoVoST2、FLEURS、LibriSpeech、MOSEL、MLS、VoxPopuli-ASR和YouTube-Commons等，涵盖了两种语言之间的相互翻译。数据集的结构包括唯一ID、音频文件名、起始时间、持续时间、说话者ID、源语言ID、识别文本、目标语言ID、翻译文本等字段。
The FAMA dataset is a large-scale open-source speech foundation model training dataset for English and Italian, used for automatic speech recognition (ASR) and speech translation (ST) tasks. It consists of multiple sub-datasets, including CommonVoice, CoVoST2, FLEURS, LibriSpeech, MOSEL, MLS, VoxPopuli-ASR, and YouTube-Commons, covering translations between the two languages. The dataset structure includes fields such as unique ID, audio filename, start time, duration, speaker ID, source language ID, recognized text, target language ID, and translated text.
提供机构：
FBK-MT
