Audacity AI Plugins adds effects such as noise reduction and music separation to Audacity editing.
Vaibhavs10/insanely-fast-whisper
insanely-fast-whisper accelerates Whisper transcription using Transformers, Optimum, and flash attention.
About Vaibhavs10/insanely-fast-whisper
insanely-fast-whisper is an open-source command-line tool that transcribes audio files on-device using Whisper with Transformers, Optimum and flash attention. It accepts a local file path or URL and saves transcription to output.json by default, with options for model selection, transcribe or translate tasks, language detection, batch size and device.
The project reports transcribing 150 minutes of audio in less than 98 seconds with Whisper Large v3, with benchmarks listed for Nvidia A100 80GB and Google Colab T4. Use is limited to NVIDIA GPUs and Mac devices with CUDA or MPS, and transcription is run with flags such as --device-id, --flash and --model-name.
Key features
- Transcribe audio files via on-device command-line interface
- Accept local file path or URL as input
- Support Whisper Large v3 and Distil-Whisper models
- Optional Flash Attention 2 acceleration flag
- Transcribe or translate tasks with language auto-detection
- Adjustable batch size and device selection for CUDA and MPS
Best for: People with an NVIDIA GPU or Mac who want to run Whisper transcription locally from the terminal.
Good for
- Reported to transcribe 150 minutes of audio in less than 98 seconds with Whisper Large v3
- Runs transcription on-device via the command line
- Supports Flash Attention 2 and Distil-Whisper models via flags
Watch out for
- Only works on NVIDIA GPUs and Mac, with CUDA or MPS devices
- pipx can install outdated version 0.0.8 on Python 3.11 without --ignore-requires-python flag
- Batch size may need to be reduced if out-of-memory errors occur
Vaibhavs10/insanely-fast-whisper at a glance
- Category
- AI & Machine Learning › AI Voice & Audio
- Type
- Open-source project
- Pricing
- Not confirmed. Check the site
- Open source
- Yes · Source code
- Language
- English
- Last checked
- Oct 6, 2026 · Online when checked
- Listed
- Oct 5, 2026
Vaibhavs10/insanely-fast-whisper alternatives
Tools and sites you could use instead, ranked by how closely they match.
Use Clips AI, an open-source Python library, to convert long-form videos into shorter clips in developer editing workflows.
EchoFox transcribes and summarizes WhatsApp audio messages so users can read them without playing the recording.
UniScribe transcribes audio and video in more than 60 languages and exports text from uploaded recordings.
AudioShake separates recordings into audio stems, making individual sound elements editable and available for other uses.
Interview VIP provides real-time AI answers during technical interviews on major meeting and coding platforms, with an undetectable mode described on its page.
Free trial
24 alternativesPopular in AI Voice & Audio
Well-known listings from the same topic.
Google Illuminateilluminate.google.comAlteredaltered.aiELSA Speakelsaspeak.comKreadoAIkreadoai.comPodwisepodwise.aiAddsubtitleaddsubtitle.ai
Vaibhavs10/insanely-fast-whisper FAQ
How do I transcribe a file on macOS?
Add the --device-id mps flag to the command.
How do I use Flash Attention 2?
Add --flash True to the transcribe command.
Can I use it without installing?
Yes, run pipx run insanely-fast-whisper --file-name .
Is something wrong or out of date?
Tell us and an editor will check it.
Own this website?
Claim this listing
Claim the listing and suggest updates yourself. An editor checks every claim first.