FA

faster-whisper

Reimplementation of Whisper for fast local transcription

Self-Hosted AI ★ 24.8k stars Medium setup MIT

faster-whisper is a reimplementation of OpenAI's Whisper model using the CTranslate2 inference engine. It transcribes audio several times faster while using less memory, making local speech recognition practical.

Key features

  • CTranslate2 acceleration
  • Lower memory usage
  • Word-level timestamps
  • CPU and GPU support

Strengths

  • Released under the MIT license
  • Mature project with 24.8k GitHub stars
  • Written in Python

faster-whisper replaces

Compare faster-whisper

13 head-to-head comparisons.

Similar self-hosted ai apps