Re: [codec] Discussion around ITU LS

My two cents:

Time scaling is important for using a codec on the Internet, but so is e.g. a jitter buffer.  People who want a full software stack for interactive Internet audio applications should look elsewhere (WebRTC).  By exposing internal pitch information through the API we enable an efficient integration of the codec, and that should be where our responsibility ends.

Furthermore, leaving out superfluous functionality helps to minimize risk of IPR infringement.  

koen.

----- Original Message -----
From: "Anisse Taleb" <Anisse.Taleb@huawei.com>
To: "Jean-Marc Valin" <jmvalin@mozilla.com>
Cc: "Jonathan Rosenberg" <jonathan.rosenberg@skype.net>, codec@ietf.org
Sent: Thursday, September 15, 2011 1:25:18 AM
Subject: Re: [codec] Discussion around ITU LS

Dear Jean-Marc,

Thanks for your consideration. I don't think acoustic front-end algorithms such as echo cancellation and noise reduction are necessary to include. This does not mean that this functionality would not be necessary for operation on the internet (or any other access technology for that matter), However, I do not see them as functionality especially included for optimal operation on the internet. 

Kind regards,
/Anisse
________________________________________
From: Jean-Marc Valin [jmvalin@mozilla.com]
Sent: 14 September 2011 19:08
To: Anisse Taleb
Cc: Cullen Jennings; Jonathan Rosenberg; codec@ietf.org
Subject: Re: [codec] Discussion around ITU LS

> If that is not agreeable, I would suggest that the codec
> specification, either clearly and explicitly states that time scaling
> is not supported by the Opus codec and a link to an external code,
> NetEQ for example, is provided. Please consider this as a way to
> avoid confusion (which I strongly believe is the case) on the nature
> of the functionality offered by the "Internet Codec" developed by
> IETF.

I think this is a perfectly reasonable suggestion. Opus is not claiming
to solve all the world's problems and I'm fine with explicitly stating
so. So the draft will include a link to the NetEQ code as a suggestion
for both jitter buffer and time stretching. The same code base also does
noise reduction and echo cancellation, so we might want to point to that
as well.

Cheers,

        Jean-Marc
_______________________________________________
codec mailing list
codec@ietf.org
https://www.ietf.org/mailman/listinfo/codec