README.md · Yehor/whisper-small-ukrainian at main

metadata

language:
  - uk
license: mit
tags:
  - whisper-event
datasets:
  - Yehor/voa-uk-transcriptions
metrics:
  - wer
model-index:
  - name: Whisper Small Ukranian
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: mozilla-foundation/common_voice_10_0 uk
          type: mozilla-foundation/common_voice_10_0
          config: uk
          split: test
          args: uk
        metrics:
          - name: Wer
            type: wer
            value: 27

🇺🇦 Join Ukrainian Speech Recognition Community - https://t.me/speech_recognition_uk

⭐ See other Ukrainian models - https://github.com/egorsmkv/speech-recognition-uk

Code for fine-tuning: https://github.com/egorsmkv/whisper-ukrainian

Tests:

WER: 27% (model quality is 73%)

The test with the original Whisper model is WER 30.57% (model quality is 69.43%)

Tests were with https://github.com/egorsmkv/cv10-uk-testset-clean