Fact · Speech recognition
1000hours
LibriSpeech is a large-scale corpus of 1,000 hours of read English speech, distributed with clean and harder other test sets.
Checked against its source on October 16, 2026
In the source's own words
“Large-scale (1000 hours) corpus of read English speech”
LibriSpeech ASR corpus (SLR12), Open Speech and Language Resources, Vassil Panayotov et al., Open Speech and Language Resources (OpenSLR), 2015, Summary.
Open the original- What it measures
- Built from public-domain audiobooks and the matching book texts. It is the most common benchmark for research on speech recognition.
- What it does not show
- Read speech from books is cleaner and more regular than meetings, calls or dictation.