news.volyx.in

FFmpeg 8.0 adds Whisper support (code.ffmpeg.org)

1033 points by rilawa · 349 days ago · 325 comments on HN

Article summary

FFmpeg 8.0 has added support for Whisper, a speech-to-text system. This addition allows for the generation of subtitles and transcription of audio and video files. The Whisper filter can output text, SRT subtitles, or JSON data. The implementation of Whisper in FFmpeg is expected to improve automatic subtitle generation and speech-to-text capabilities.

Main themes

  • FFmpeg update
  • Whisper support
  • Speech-to-text
  • Subtitle generation
  • Bot filtering
  • Browser compatibility

What commenters say

  • The addition of Whisper support to FFmpeg enables dynamic generation of subtitles for movies with AI, which could be a significant improvement for video players.
  • Some commenters believe that VLC and FFmpeg are unrelated projects, while others argue that there is significant overlap between the two, including shared libraries.
  • The Anubis bot filter used on the FFmpeg website can cause issues for some users, particularly those with older browsers or certain browser configurations, and may be overly restrictive.
  • The implementation of Whisper in FFmpeg may have performance implications, particularly for users without GPU support, but some commenters have reported acceptable performance on relatively modern hardware.
  • Some users have experienced issues with the Anubis bot filter, including being blocked or receiving error messages, which may be due to IP reputation or browser configuration.
  • The use of Anubis as a bot filter is seen as a necessary measure to prevent scraping and protect the website, but some commenters argue that it can be configured to be less restrictive while still effective.
  • The Whisper filter in FFmpeg has potential applications beyond just subtitle generation, including speech-to-text capabilities for various use cases.
  • There are differing opinions on the ease of use and configuration of Anubis, with some commenters suggesting that it can be configured to work with older browsers, while others have experienced difficulties.