[Tools-discuss] Re: Session audio recording

[Lorenzo Miniero <[e-mail]>]

On Thu, 6 Nov 2025 08:36:15 -0800
Rob Sayre <[e-mail]> wrote:

> Well it makes sense, sort of. Even though hosting the files would not
> be expensive, someone has to take care of it, and it is a different
> task than writing Python and HTML while dealing with IETF details. I
> think it would be easy to put them on S3 or equivalent (Cloudflare,
> etc have competitors and the API is open, so the AWS tools work with
> most of them). That would create an ever-increasing bill as we add
> recordings.
> 
> In the meantime, it occurred to me that Ekr can use the "youtube-dl"
> tool to make this work right now.
> 
> https://github.com/ytdl-org/youtube-dl/blob/master/README.md#readme
> 
> -x, --extract-audio          Convert video files to audio-only files
>                                      (requires ffmpeg/avconv and
>                                      ffprobe/avprobe)
> 


Yes, I thought about tools like youtube-dl (or yt-dlp) too, but -x
would not be what I'd go to if you want to save the download bandwidth.
Those tools have a --list-formats option that allows you to list which
options are available for download, and some of them will be audio- or
video-only. For instance, from a check I made with a TLS recording from
IETF 124:

	yt-dlp -F "https://yewtu.be/watch?v=zTAuEx9Otys"

formats 233 and 234 look like audio-only available options, meaning
that they can be picked selectively:

	yt-dlp -f 233 "https://yewtu.be/watch?v=zTAuEx9Otys"

In this case, this only downloads 25MB (mp4 file with AAC codec).

Lorenzo


> thanks,
> Rob
> 
> On Thu, Nov 6, 2025 at 7:49 AM Tom Pusateri <[e-mail]>
> wrote:
> 
> > I have discussed this in the past for use with the mobile apps.
> >
> > I had asked for all the individual raw files be archived for audio
> > and video but this request was rejected. The problem seemed to be
> > that the IETF didn’t want to provide the storage for this. So they
> > generate a file to upload to YouTube and delete the originals.
> >
> > I wanted to play around with different ways to present the audio
> > and video through the mobile apps such that you could have just
> > audio and have it play in the background while driving or on a
> > walk. Or switch between video and high quality synchronized slides.
> >
> > I also experimented with speech to text systems and speaker
> > identification algorithms. But without a reliable source of the
> > audio files, I lost interest in the project.
> >
> > Tom
> >
> >  
> > > On Nov 5, 2025, at 11:05 PM, Rob Sayre <[e-mail]> wrote:
> > >
> > > Oh, some of them do have .m3u but I am not sure how that works.
> > > Maybe  
> > just a few paragraphs of documentation, since the feature already
> > seems to be there  (but only sometimes? idk). How do we get the
> > headphones link there?  
> > >
> > > thanks,
> > > Rob
> > >
> > > (not unaware that I am sending "how do I get the headphones
> > > button?" to  
> > an IETF list)  
> > >
> > >
> > > <Screenshot 2025-11-05 at 8.02.10 PM.png>
> > >
> > >
> > >
> > > On Wed, Nov 5, 2025 at 2:50 PM Eric Rescorla <[e-mail]> wrote:
> > >
> > >
> > > On Wed, Nov 5, 2025 at 2:46 PM Watson Ladd
> > > <[e-mail]>  
> > wrote:  
> > >
> > >
> > > On Wed, Nov 5, 2025, 2:38 PM Eric Rescorla <[e-mail]> wrote:
> > >
> > >
> > > On Wed, Nov 5, 2025 at 2:32 PM Rob Sayre <[e-mail]> wrote:
> > >
> > >
> > > On 11/5/25 1:02 PM, Eric Rescorla wrote:  
> > > > Is there a straightforward way to just get the audio recording
> > > > for a given session?
> > > >  
> > >
> > > Hi,
> > >
> > > It's a fair question--sometimes they have had links for that from
> > > MeetEcho, but this time around it seems like there are no mp3
> > > files, only YouTube and Cloudflare video (which confusingly looks
> > > like SoundCloud?).
> > >
> > >
> > > These should be on BitTorrent, but in practice it doesn't matter.
> > > YouTube works fine and the video overhead isn't too bad because
> > > it's just slides that don't really change most of the time over
> > > each frame, and you need to show the "Note Well" without wasting
> > > everyone's time by reading it out.
> > >
> > > What I'm looking to do here is not to avoid the bandwidth
> > > associated with the video but to be able to download the audio so
> > > I can feed it into a speech to text model.
> > >
> > > The model can also give you the baroque (is there any other kind)
> > > of  
> > ffmpeg invocation to strip out just the audio.  
> > >
> > > Indeed it could, if the video were available for download at some
> > >  
> > official URL.  
> > >
> > > -Ekr
> > >
> > >
> > > -Ekr
> > > -----------------------------------------------
> > > Tools-discuss mailing list -- [e-mail]
> > > To unsubscribe send an email to [e-mail]
> > > https://mailarchive.ietf.org/arch/browse/tools-discuss/
> > > -----------------------------------------------
> > > Tools-discuss mailing list -- [e-mail]
> > > To unsubscribe send an email to [e-mail]
> > > https://mailarchive.ietf.org/arch/browse/tools-discuss/  
> >
> >
> >  



-- 
I'm getting older but, unlike whisky, I'm not getting any better
https://lminiero.it