Whisper

Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse audio and is also a multitasking model that can perform multilingual speech recognition, speech translation, and language identification.

• Whisper home page

Whisper can be installed to a SD Desktop virtual machine with SD Software installer.

The version provided for SD Desktop is based on Faster-Whisper-XXL.

After installation, Whisper is available as a command-line tool in SD Desktop.

Sample commands:

whisper audio.mp3 --model medium --threads 4

Sample command with diarization enabled:

whisper interview.mp4 --model large --language French --threads 4 --diarize pyannote_v3.0 --diarize_threads 4 --num_speakers 2 -o interview_results

Update on 5.5.2025

Before the latest update, the SD tool installer provided WhisperDO-based Whisper installation. Faster-Whisper-XXL provides now new features, like diarization, and better performance compared to the old WhisperDO.

If you want to upgrade your old WhisperDO installation to the new Faster-Whisper-XXL, you need to first remove the old version with commands:

rm /shared-directory/sd-tools/apptainer/whisper.sif
rm /shared-directory/sd-tools/bin/whisper

After that you can use the SD tool installer to install the new version of Whisper.

Do you find this page useful?

Yes

Thank you for your feedback!

No

Thank you! Your feedback is appreciated.

Please let us know how we could improve this page by contacting us or by creating an issue in GitHub. You can also edit this page yourself and contribute your improvements by creating a pull request.

May 8, 2025