Dataset · fsicoli
Common Voice 22.0 multilingual speech corpus
fsicoli/common_voice_22_0
Unofficial export of the 22nd version of Mozilla's open speech dataset for speech recognition tasks.
What's inside
Contains speech audio recordings and corresponding text transcripts for multiple languages, including Abkhaz, Bashkir, Belarusian, Chuvash, Arabic, Chinese, and others. Data downloaded from the official Mozilla Common Voice project site and converted for convenient use.
What you can do with it
- Training and fine-tuning ASR models for low-resource languages
- Evaluating the quality of multilingual speech recognition systems
- Tasks
- Automatic speech recognition
- Languages
- Abkhazian, Afrikaans, Amharic, Arabic, Assamese, Asturian, Azerbaijani, Bashkir and 4 more
- Size
- 100B–1T rows
- Downloads this month
- 31.7K
- Likes
- 20
Models trained on it
- MihaiPopa-1/F5-TTS-RomanianText → speech 360
- AleksTv/whisper-medium-ky-mergedAutomatic speech recognition 138
- yuriyvnv/Qwen3-ASR-1.7B-NLAutomatic speech recognition 81
- yuriyvnv/Qwen3-ASR-1.7B-PTAutomatic speech recognition 56
- yuriyvnv/Qwen3-ASR-0.6B-PTAutomatic speech recognition 56
- teckedd/gha-whisper-small-twi-v6Automatic speech recognition 47
- yuriyvnv/Qwen3-ASR-0.6B-NLAutomatic speech recognition 42
- akimbabananan/whisper-tiny-ru-en-onlyAutomatic speech recognition 40
How to load it
from datasets import load_dataset
ds = load_dataset("fsicoli/common_voice_22_0")