Fact · Speech recognition
1000hours
LibriSpeech is a corpus of about 1,000 hours of 16 kHz read English speech, and its data is derived from read audiobooks from the LibriVox project.
Checked against its source on October 17, 2026
In the source's own words
“LibriSpeech is a corpus of approximately 1000 hours of 16kHz read English speech, prepared by Vassil Panayotov with the assistance of Daniel Povey. The data is derived from read audiobooks from the LibriVox project”
LibriSpeech ASR corpus (SLR12), Open Speech and Language Resources, Vassil Panayotov et al., Open Speech and Language Resources (OpenSLR), 2015, About this resource.
Open the original- What it measures
- The corpus was prepared from volunteer audiobook recordings and aligned to the text of the books, so each segment has a reference transcript. Its recordings are read aloud, not spontaneous speech.
- What it does not show
- Describes the corpus, not how well any system performs on it. Read audiobook speech is cleaner and more regular than dictation in an office or on a phone.