Gradio

Whisper tiny Model Comparison [ whisper-tiny-pl-by-leliw vs original openai ] (v0.09)

Compare speech transcription between Polish model (leliw/whisper-tiny-pl) and original OpenAI model (openai/whisper-tiny).

Record or upload audio

Enable GPU acceleration (requires restart)

Use GPU

Force model to use a specific language token

Use Language Token

Language Token

Complete language token (e.g., '<|pl|>', '<|en|>')

Provide context to help model with transcription

Use Initial Prompt

Initial Prompt

Text that helps guide transcription, especially for specialized terms

Polish model result (leliw/whisper-tiny-pl)

Polish model logs

OpenAI model result (openai/whisper-tiny)

OpenAI model logs