FFmpeg to add transcription functionality using OpenAI's Whisper

It has been revealed that the multimedia tool ' FFmpeg ' will be adding a transcription feature using OpenAI's transcription AI ' Whisper .' At the time of writing, the release of FFmpeg 8.0 is in preparation, and the transcription feature will also be included in FFmpeg 8.0.

[FFmpeg-devel] [PATCH] Whisper audio filter
https://ffmpeg.org/pipermail/ffmpeg-devel/2025-July/346391.html

FFmpeg 8.0 Merges OpenAI Whisper Filter For Automatic Speech Recognition - Phoronix
https://www.phoronix.com/news/FFmpeg-Lands-Whisper

FFmpeg is an open-source multimedia tool with a wide range of functions, including encoding and decoding video and audio, extracting specified sections from video, and creating videos from a sequence of images. FFmpeg can be used as a standalone command line tool, and also runs inside various apps and web services. Many people benefit directly or indirectly from FFmpeg.

It has been revealed that FFmpeg will be adding a transcription function using OpenAI's Whisper. Whisper is a transcription AI that supports Japanese and can produce highly accurate transcriptions simply by loading a recording file. The Whisper model data is available for free and can be downloaded to your PC and run locally.

Summary of how to transcribe recorded files using voice recognition for free using OpenAI's 'Whisper' - GIGAZINE

The transcription feature added to FFmpeg can be used by building with the option '--enable-whisper'. To do this, you will need to have the library ' whisper.cpp ' on your PC, which allows you to use Whisper in various apps.

The transcription function supports output to JSON and SRT files, and also supports GPU acceleration. You can check the code for FFmpeg's transcription function at the link below.

git.ffmpeg.org Git - ffmpeg.git/commit
https://git.ffmpeg.org/gitweb/ffmpeg.git/commit/13ce36fef98a3f4e6d8360c24d6b8434cbb8869b

In addition, an article explaining how to use the transcription function by Vittorio Palmisano, the patch creator, is available at the following link.

Run Whisper audio transcriptions with one FFmpeg command | by Vittorio Palmisano | Jun, 2025 | Medium
https://medium.com/@vpalmisano/run-whisper-audio-transcriptions-with-one-ffmpeg-command-c6ecda51901f

At the time of writing, preparations for the release of FFmpeg 8.0 are underway, and if preparations proceed as planned , it will be released in August 2025.