Edit model card

Model card - Kinyarwanda coqui STT model

Model details

Intended use cases

  • Intended to be used for
    • simple keyword spotting
    • simple transcribing
    • transfer learning for better kinyarwanda and african language models
  • Intended to be used by:
    • App developpers
    • various organizations who want to transcribe kinyarwanda recordings
    • ML researchers
    • other researchers in Kinyarwanda and tech usage in kinyarwanda (e.g. Linguists, journalists)
  • Not intended to be used as:
    • a fully fledged voice assistant
    • voice recognition application
    • Multiple languages STT
    • language detection

Factors

  • Anti-bias: these are bias that can influence the accuracy of the model
    • Gender
    • accents and dialects
    • age
  • Voice quality: factors that can influence the accuracy of the model
    • Background noise
    • short sentences
  • Voice format: voices must be converted to the wav format
    • wav format

Metrics

  • word error rate on the Common Voice Kinyarwanda test set
Test Corpus WER
Common Voice 39.1%

Training data

Evaluation data

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Examples
Unable to determine this model's library. Check the docs .

Space using mbazaNLP/kinyarwanda-coqui-stt-model 1