faster-whisper

A reimplementation of Whisper that transcribes several times faster on the same hardware. Local, MIT, and the sensible default.

Self-hosted
Active
Allowed adult content
Open source

Website Source

At a glance

Where it runs: Self-hostedYou run it on your own machine. Nothing leaves it unless you send it.
Adult content: AllowedAdult content is permitted within the published rules.
Trains on your content: NoYour content is not used to train models.
Status: ActiveMaintained and working today.
Account needed: No
Licence: MIT
Cost: Free and open source.
What it keeps: Local only.

What it takes to run

Graphics memory: No GPU needed
System memory: 4 GB, 8 GB recommended
Disk: about 5 GB
Runs on: Linux, macOS, Windows

Find what fits your machine on the tools page — pick your hardware from the first menu.

Same model, same output, far less time and memory. It is what most transcription tooling uses under the hood, and the reason running the large Whisper model on a modest card is practical.

Transcription is the one corner of this field where the free, local, private option is also the best one.

Worth knowing before you start

A library, not an app It expects to be called from Python or wrapped by something else.

What it does

Several times faster than the reference implementation
Word-level timestamps
Runs on CPU
Many languages

Models it runs

Whisper large-v3

Signs of life

Checked automatically. These are the only figures on this page a machine wrote, and they say when they were taken.

Last answered: Yes, 3 hours ago Its website responded when we asked.
Stars on GitHub: 25,416 2,076 forks.
Last commit: 300 days ago2025-11-19 14:40 UTC
Latest release: v1.2.1
Archived copy: Wayback Machine A snapshot, in case this one stops answering.