Talk:Community Wishlist/Wishes/Multiple audio tracks for the same video

This page is for discussions related to the Community Wishlist/Wishes/Multiple audio tracks for the same video page.   Please remember to: Sign posts using the four tildes (~~~~) Remain civil and polite during discussions. Place new text under old text (start a new post).

'convert to mp4'

Not sure why you would need to convert to mp4 and back again. WebM can handle multiple audio tracks just as well as mp4. This sounds more like a limitation of the video editing software you are using. —TheDJ (talk • contribs) 13:18, 29 July 2024 (UTC)

Can we handle multiple audio tracks in Commons? I think not. Theklan (talk) 15:02, 29 July 2024 (UTC)

Currently not. One needs to download the video from WMC to see any other audio channels. Could you please mention that this was also requested here? Prototyperspective (talk) 16:07, 29 July 2024 (UTC)

You can add the information to the wish, for sure! Theklan (talk) 22:09, 29 July 2024 (UTC)

Nope, but there is just no reason to go to mp4. You can simply remux the file with another audio track and keep the same video track and write back a new webm file. MP4 does not have to be involved. —TheDJ (talk • contribs) 19:37, 29 July 2024 (UTC)

I have checked and the three leading softwares for video editing (Adobe Premiere, Final Cut Pro and DaVinci Resolve) can't handle webm. Theklan (talk) 22:08, 29 July 2024 (UTC)

OK, i did not count on commercial software packages being so out of sync with modern times :)

But still, in terms of quality preservation, I would export the audio track from the mp4, and then remux webm file with ffmpeg to inject the audio track and preserve the original quality of the video track. —TheDJ (talk • contribs) 20:20, 30 July 2024 (UTC)

It can be done? Yes. Are we going to attract video content creators asking for those steps? I don't think so. Theklan (talk) 10:37, 31 July 2024 (UTC)

Very much agree. People won't do this. Instead it should be done automatically or via a button (incl API for redubbing software) like "add audio channel to this video".

However, the part The current approach for dubbing a video is downloading the original video, converting it from .webm to .mp4, dubbing it seems wrong. Why do you convert to mp4? Audio channels can also be added to webm files and I've done that a few times with ffmpeg. By the way a problem there was that one needs to first remove the audio then add the audio channel then readd the audio from the unmuted video if one wants the audio channel to be channel 1 or the one that is played on WMC (e.g. when redubbing a German video to English and wanting English to be in channel 1) – let me know if there is a way around that. Prototyperspective (talk) 11:44, 31 July 2024 (UTC)

Multiple subtitle tracks would also complement this very well

While we're at it, it would be useful to have multiple subtitle tracks for different languages. Subtitle tracks also have the advatage that they can be(a) used to create spoken audio tracks using TTS software, and also (b) be created automatically from spoken audio tracks using transcription software, both of which are available as open source. The Anome (talk) 18:34, 3 August 2024 (UTC)

Aren't these already implemented with TimedTexts? One could write the subtitles to the video files but that's probably not what you mean. So it's about reading subtitles of videos, right? That would indeed be useful. However, on that end, often not even the only subtitles of videos get read – see phab:T368298. By the way, one thing I noticed using open source software to redub videos on WMC is that there's often issues when using the subtitles people made so often one gets a better result letting the software transcribe anew and then using that instead of the subtitles. Best would be a way to merge these (machine transcription can get e.g. names and occasionally a word wrong) or adjust the timings of the manually-made transcript, that's also what my other proposal is about. Prototyperspective (talk) 23:15, 3 August 2024 (UTC)

Multiple subtitle tracks can be done now. Take a look to this:

Theklan (talk) 12:32, 4 August 2024 (UTC)

Useful

This could be very useful and mitigate problems like bloated file-sizes (due to a video having an audio track for 300 languages) if this was implemented similar to TimedTexts where the audio tracks are stored separately from the video file (which just has the default audio track). I don't know if that's possible with current media formats.

There are now YouTube videos with multiple audio tracks, for example see the settings of this Arte documentary video (further info may be in the link posted in here).

I think the proposal is much more relevant and useful now that there are well-working AI redubbing tools / methods by which one can relatively quickly redub a video as described in c:Help:AI video dubbing. Rather than uploading many videos each with a different audio channel it may be best to just have audio channels, e.g. TimedAudio. One could also make it so that the other version video is loaded when selecting a different audio channel in the player config and the video adapting the language depending on where it's used / user-preferences. This would for example make it possible to add an English-language video or any kind of video with subtitles to Wikipedia articles across many languages where the same file gets an audio channel added after some time. It also prevent WMC getting flooded with 100 redubs per one video where people instead just add the redub audio to the existing one video. (A caveat is that file titles should also be translatable which is currently an advantage of uploading videos as separate files.) Prototyperspective (talk) 22:19, 24 November 2024 (UTC)