Fact · Speech recognition
9.85percent word error rate
Vosk's small English model vosk-model-small-en-us-0.15, at 40 MB, had a word error rate of 9.85 percent on LibriSpeech test-clean, and it is listed under the Apache 2.0 license.
Checked against its source on October 17, 2026
In the source's own words
“vosk-model-small-en-us-0.15 40M 9.85 (librispeech test-clean) 10.38 (tedlium) Lightweight wideband model for Android and RPi Apache 2.0”
Vosk models (Alpha Cephei), Alpha Cephei, 2026, Model list table.
Open the original- What it measures
- The Vosk project describes this model as a lightweight wideband model for Android and Raspberry Pi. The score is for LibriSpeech test-clean, measured by the project.
- What it does not show
- The project's own measurement on one read-speech test set. Small models trade accuracy for size, so the score is not a guide to how the model handles noisy or conversational audio.