Whisper
OpenAI's multilingual speech recognition and translation models, from tiny to large-v3 and turbo, with a Python CLI.
- License
- MIT
- Platforms
- Windows, macOS, Linux, CLI
- Own API keys
- No
- Checked
- 2026-09-24
Whisper alternatives
Offline push-to-talk dictation: hold a hotkey, speak, and the text is pasted into any app; optional LLM cleanup via an OpenAI-compatible API.
CTranslate2 reimplementation of Whisper, up to 4x faster with less memory and 8-bit quantization.
Dependency-free C/C++ port of Whisper with CUDA, Vulkan, Metal, quantization, VAD and an HTTP server.
Desktop app to transcribe and translate audio and video offline with Whisper or any OpenAI-compatible API.
Lightweight open-weight 82M-parameter text-to-speech model with 54 voices in 8 languages, fast even on CPU.
Open-source TTS with zero-shot voice cloning: Turbo (English), Multilingual (23+ languages incl. Russian) and a CPU-friendly Nano.
FAQ
Is Whisper free?
Code and model weights are MIT-licensed and free to run locally; only OpenAI's separate cloud transcription API is paid.
Does Whisper work with free models?
It uses the vendor's own models; connecting third-party free APIs is not supported or not documented.
What license does Whisper use?
MIT. The source code is open.