Whisper

Whisper is OpenAI's open-source, general-purpose speech recognition model. It transcribes audio, can translate speech in other languages into English, and detects which language is being spoken. Once installed, you run it from the command line on audio files you provide, and it can write subtitle files. The audio is processed on your own computer, with no OpenAI API key needed. What sets it apart is that both the code and the model weights are public, and many captioning and transcription tools are built on it.

Who it's for:people who caption videos or transcribe interviews and don't mind setting up a little developer tooling. Readers who don't code or avoid the command line can use ready-made apps on this site that are built on it.

Before you start: it needs Python, PyTorch and ffmpeg, installs with pip, and runs on Windows, macOS and Linux. The first run downloads a model, and bigger models need a more powerful computer. Audio is processed locally by default; whether third-party tools built on Whisper upload anything depends on those tools. The code and model weights are MIT-licensed; check the official documentation.

Based on the project's public documentation. Not tested by AI Inspiration World.