Skip to content
Home›AI & agents›AI tools›faster-whisper
Transcribe & subtitle

faster-whisper: a quicker Whisper engine for Python users

An MIT-licensed Python library that runs OpenAI's Whisper up to 4x faster at the same accuracy with less memory, by the project's own account [2][1].

Does
Local transcripts with word timings
Best for
Interview and podcast creators with an NVIDIA GPU or fast CPU
Cost
Free (MIT licence) [1]
Runs
Locally, on CPU or NVIDIA GPU
Open the repositoryOfficial siteQuick start
Updated 2 Nov 2026

Whisper speech-to-text, run through the CTranslate2 inference engine [2]. You call it from a Python script; there is no app or built-in command. WhisperX [5] and Buzz [6] use it as an engine.

What it does

Licence
MIT [1]
Interface
Python library only; community command-line clients such as whisper-ctranslate2 [2]
Requirements
Python 3.9+; no FFmpeg install needed, as PyAV decodes the audio [2]
GPU
NVIDIA only: cuBLAS and cuDNN 9 for CUDA 12 [2]. No Apple GPU path documented
Output
Timed segments in Python; word timings with word_timestamps=True; optional Silero voice activity filter [2]
Project benchmark
13 min of audio, large-v2, RTX 3070 Ti: 1m03s at fp16 against 2m23s for openai/whisper (project's own test) [2]

Integrations

InterfaceStatusDetail
MCP serverMCP · CommunityCommunity wrappers
Command lineCLI · CommunityNo built-in command; whisper-ctranslate2 is a community client built on it [2].
API / libraryAPI · Built inPython API

Requirements and setup

  • Python 3.9+ [2].
  • GPU: NVIDIA with cuBLAS and cuDNN 9 for CUDA 12; older CUDA needs an older ctranslate2 [2].
  • CPU: device="cpu", compute_type="int8" [2].
  • Models download from the Hugging Face Hub on first load [2].

Licence, cost and privacy

  • Runs locally once the model is downloaded [2].

Limitations

  • No subtitle writer: build SRT from segment times or use a community client [2].
  • segments is a generator; transcription starts only when you loop over it [2].
  • Default beam size is 5 (openai/whisper: 1); match settings before comparing [2].
  • The voice activity filter only removes silence over 2 seconds by default [2].
  • Speed figures are the project's own [2]; VloggingPro has not tested them.
Quick start

Quick start

Install, then run the script. It prints each word with its start and end time, for word-by-word captions [2].

Copied from the documentation · source [2]
pip install faster-whisper
Assembled from documented options, not printed in the docs · source [2]
from faster_whisper import WhisperModel

model = WhisperModel("large-v3", device="cpu", compute_type="int8")
segments, _ = model.transcribe("audio.mp3", word_timestamps=True)

for segment in segments:
    for word in segment.words:
        print("[%.2fs -> %.2fs] %s" % (word.start, word.end, word.word))

Save as transcribe.py and run python transcribe.py. On an NVIDIA GPU, use device="cuda", compute_type="float16" [2].

Uses for vloggers

  • Batch-transcribing a back catalogue of interviews or podcasts on an NVIDIA machine.
  • Word timings for animated word-by-word captions [2].

Workflows that use it

Some setupLocal AI subtitles

Turn vlog speech into SRT and VTT subtitle files on your own computer, with an app or the command line, then check every name.

Buzz transcriptionAuto Subs (tmoroney)whisper.cppFFmpeg / ffprobe

Alternatives

Transcribe & subtitleWhisperX

Generate aligned word-level subtitles and diarisation workflows

Free codeOn your computerDeveloperBSD-2-Clause licence
MCP · CommunityCLI · Built inAPI · Built in
Transcribe & subtitlewhisper.cpp

Run efficient Whisper inference on local hardware

Free codeOn your computerDeveloperMIT licence
MCP · CommunityCLI · Built inAPI · Built in
Transcribe & subtitleWhisper (OpenAI)

Speech recognition and subtitles for local post-production

Free codeOn your computerDeveloperMIT licence
MCP · CommunityCLI · Built inAPI · Built in
Transcribe & subtitleBuzz transcription

Desktop transcription / speech-to-text workflow

Free codeDesktop app · On your computerNo codingMIT licence
MCP · None foundCLI · Built inAPI · None found

Sources

  1. raw.githubusercontent.com/SYSTRAN/faster-whisper/master/LICENSE fetched 2 Nov 2026
  2. raw.githubusercontent.com/SYSTRAN/faster-whisper/master/README.md fetched 2 Nov 2026
  3. raw.githubusercontent.com/jianfch/stable-ts/main/README.md fetched 2 Nov 2026
  4. GitHub - jianfch/stable-ts: Transcription, forced alignment, and audio indexing with OpenAI's Whisper · GitHub fetched 2 Nov 2026
  5. raw.githubusercontent.com/m-bain/whisperX/main/README.md fetched 2 Nov 2026
  6. Introduction | Buzz fetched 2 Nov 2026