ggfox00000/stt-librispeech-test-en
收藏Hugging Face2026-04-28 更新2026-05-03 收录
下载链接：
https://hf-mirror.com/datasets/ggfox00000/stt-librispeech-test-en
下载链接
链接失效反馈官方服务：
资源简介：
LibriSpeech ASR测试集（clean + other）是openslr/librispeech_asr的镜像数据集，包含两个配置：clean和other。clean配置包含2620个测试用例，用于干净的自动语音识别（ASR）基线测试，主要来自有声读物的朗读。other配置包含2939个测试用例，用于声学条件较差、说话者较难的测试。数据集包含16 kHz单声道的FLAC音频文件和对应的文本参考（WER）。数据来源于Panayotov等人在2015年ICASSP上发表的论文《LibriSpeech: an ASR corpus based on public domain audio books》。
The LibriSpeech ASR test set (clean + other) is a byte-exact mirror of the `test` splits from the `clean` and `other` configurations of `openslr/librispeech_asr` (Panayotov et al. ICASSP 2015). The `clean` configuration contains 2620 utterances for clean ASR baseline testing, derived from audiobook readings. The `other` configuration contains 2939 utterances with more challenging acoustic conditions and speakers. The dataset includes 16 kHz mono FLAC audio files and corresponding reference text (WER). Source: Panayotov, Chen, Povey, Khudanpur. *LibriSpeech: an ASR corpus based on public domain audio books*, ICASSP 2015.
提供机构：
ggfox00000
