whisper.cpp
This program aims to facilitate running automatic speech recognition (ASR) locally both on CPU with near real-time speed and on many consumer GPU/Accelerator backends including CUDA, Apple's ANE and Vulkan Compute depending on compile-time flags.
Cost / License
- Free
- Open Source (MIT)
EUPlatforms
- Mac
- Windows
- Linux
- Android
- iPhone
- WebAssembly
- FreeBSD
- Raspberry Pi
- Docker
- Java
Features
- Command line interface
- Speech to text
- Support for NVIDIA CUDA acceleration
- Speech Recognition
- Local AI
whisper.cpp News & Activities
Recent activities
- niksavc liked whisper.cpp
- Maoholguin updated whisper.cpp
- Samenaga added whisper.cpp
Samenaga added whisper.cpp as alternative to Voxtral, Spokenly, TypeWhisper and Murmure
whisper.cpp information
Developed by
ggml-org Licensing
Open Source (MIT) and Free product.Written in
Alternatives
8 alternatives listedSupported Languages
- English
- Japanese
- Spanish
AlternativeTo Categories
AI Tools & Services, OS & UtilitiesGitHub repository
- 51,723 Stars
- 5,900 Forks
- 1216 Open Issues
- Updated
What is whisper.cpp?
This program aims to facilitate running automatic speech recognition (ASR) locally both on CPU with near real-time speed and on many consumer GPU/Accelerator backends including CUDA, Apple's ANE and Vulkan Compute depending on compile-time flags.
It can open flac, mp3, ogg and wav files by default, and many more if compiled with FFMPEG support. It can format its output as srt (default), vtt, plain text, json and more; either in their original languages or translated to English.
The backing project provides OpenAI models adapted to their own binary format for download through Hugging Face.