[Audio-video menu] 

GNU Home / Audio & video /

Audio/Video Documentation

Containers and Codecs

Audio

We distribute our audio files using Opus, a codec that offers low latency and high audio quality across a wide range of bitrates (see Opus FAQ). Opus is free/libre software, with source code released under a three-clause BSD license and protective, royalty-free licenses for the required patents (see license — page archived). We use either the WebM or the Ogg container, both unencumbered by known software patents.

Visitors to our media pages are likely to find some legacy audio files enconded in Vorbis and packaged in an Ogg container. Both Ogg and WebM are suitable containers for Vorbis. Unlike MP3, Vorbis has always been completely free and unencumbered by patents, and it's still widely supported by most browsers and platforms. The Vorbis wiki provides lists of software and hardware that support it.

Video

We distribute our video files using the AV1 codec in a WebM container. AV1 is a free/libre software video codec developed by the Alliance for Open Media, backed by royalty-free patent pledges from AOMedia's member companies and supported by most current browsers. It delivers substantially better compression than the codecs it replaces, meaning smaller files and lower bandwidth for the same visual quality.

We also offer some of our video files in WebM with the VP9 codec, for those who may prefer it (some legacy WebM files may still be found with the VP8 codec until we complete migration).

The WebM container only accepts free codecs by design, ensuring long-term compatibility without licensing encumbrances.

Deprecated Formats and Ongoing Remux

Several of the codecs we have used over the years are now either completly deprecated or supported only by a few browsers. Theora, Vorbis, Speex, and VP8 each represented important milestones for free, unencumbered media. Theora gave the free software world its first viable patent-free video codec, Vorbis did the same for audio against MP3 [1]. Speex pioneered free speech compression before Opus subsumed and surpassed it, and VP8 carried that lineage into the WebM container format—but all are outdated today, both in compression efficiency and in platform support. VP9, while still widely deployed, has likewise seen development stall and has been superseded by AV1 for our purposes.

We are actively working to remux and re-encode our entire back catalog to AV1 with Opus in WebM (or just Ogg for audio), so that everything we host benefits from a single well - supported pipeline. The codecs being gradually phased out are:

You may still encounter files in these formats while our migration is in progress.

Accessing Recordings

Most of the Ogg files have been encoded in a stream of 20 kb/s. Most Internet users, including those with slow connections, should be able to listen to the audio files streaming from our server.

Our WebM video files can be streamed by major browsers (e.g., Firefox), and streaming will start automatically by default. You can also stream them with VLC. For example:

vlc https://audio-video.gnu.org/audio/file.extension

Another possibility is to download the media file, and play it later:

wget -q https://audio-video.gnu.org/audio/file.extension

Wget with the -c switch will continue a transfer if a smaller file of the same name exists in the download location. For example,

wget -c https://audio-video.gnu.org/video/file.extension

Request for Recordings

If you are planning to attend a GNU event, please make a recording. It is polite to ask permission from the event organizer. If you wish to make the recording publicly available in a digital format, please choose one that is accessible to free software. AV1 video and Opus audio (in a WebM container) are the best choice today, as using these adds support to free, open standards and unencumbered formats.

Licensing your recording

For Richard Stallman's speech recordings, please choose the Creative Commons Attribution-NoDerivatives 4.0 International (CC BY-ND 4.0) license for the whole recording (why this license?).

Remember to embed the license into the digital file, which you can do by adding it as metadata. If you don't have the possibility of adding it as metadata in the file itself, we will do it for you; please state the license in your email or send it to us as a separate text file. By clearly declaring the license, people who receive a copy of the file will be aware of their freedoms.

Submitting your recording

If you have a good recording from a GNU event, and you would like to share it with us, please send an email to <audio-video@gnu.org>. We are interested in both audio and video recordings, and can accept the source in a wide range of formats. We will transcode it to a standardised codec for the site, and optimize the encoder settings.

When submitting recordings, please provide as much information as possible. For example (entries marked with an asterisk are important),

  • * title of the speech;
  • * date and place the speech and recording took place, including venue, address, city, and country;
  • * if the speech was part of a larger event, name of the event;
  • * confirmation that our usual license is OK to use: Creative Commons Attribution-NoDerivatives 4.0 International (CC BY-ND 4.0);
  • * who made and who owns the copyright for the recording;
  • names of organizers;
  • names of sponsors, if any;
  • names of organizations or people to be thanked, if any;
  • names of people who appear speaking in the recording, like presenters or other speakers;
  • URL of web pages related to the speech, such as announcements, reviews, etc., or server where the original recording is hosted.

We prefer to handle speech recordings as a single file rather than divided into parts, but if the original is in several parts, we can join them as required. If the original is compressed with a lossy compression format such as mpeg, divx or a proprietary format, please don't transcode the recording before sending it to us. Given that we may transcode it to a different bit rate and frame size, and insert copying information, adding yet another transcoding stage will simply degrade quality.

We don't have a public uploading facility for large media files. The best way to send us the recording is to place it on a website and send us the link. If the file is not accepted by the service due to its large size, one solution is to split it into chunks. For example, to split an 8 GB file into four 2 GB chunks:

split -n 4 file

If you experience any difficulties in this respect, we can probably find a solution.

Submitting Audio

We prefer original recordings in the original recorded sample rate up to 44100Hz. Monophonic is generally adequate for speech recordings and saves a lot of space over stereo. We will accept the recording in the original file type. If the original file is large, you may wish to transcode to 64 kb/s mono Opus in a Ogg container.

Submitting Videos

The same is true for video as for audio. Send the recording in the original frame size. If already compressed with a lossy codec, please send the original. If your original is uncompressed or has a very low compression/large file size, please compress using AV1 in a WebM container at a quality setting that preserves the original detail (for example, SVT-AV1 with --crf 30 or lower).

Submitting Subtitles and Transcriptions

If you have subtitles or a transcription of a speech by a GNU speaker, or would like to make one, please contact <audio-video@gnu.org>.

Footnote

  1. ↑ In 2007, the FSF launched the PlayOgg campaign. An article published the same year explains the issues with MP3, as well as Why Audio Format Matters.