Re: [rtcweb] Query Regarding Mandatory audio codecs draft

Paul Coverdale <> Mon, 28 April 2014 13:46 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id C32341A0A18 for <>; Mon, 28 Apr 2014 06:46:25 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, MSGID_FROM_MTA_HEADER=0.001, RCVD_IN_DNSWL_NONE=-0.0001, SPF_PASS=-0.001] autolearn=ham
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id OziNz_O0X33q for <>; Mon, 28 Apr 2014 06:46:23 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 1E0811A0795 for <>; Mon, 28 Apr 2014 06:46:22 -0700 (PDT)
Received: from BLU0-SMTP21 ([]) by with Microsoft SMTPSVC(6.0.3790.4675); Mon, 28 Apr 2014 06:46:22 -0700
X-Originating-Email: []
Message-ID: <BLU0-SMTP218AE95A6E03BA856CC468D0470@phx.gbl>
Received: from PaulNewPC ([]) by BLU0-SMTP21.phx.gbl over TLS secured channel with Microsoft SMTPSVC(6.0.3790.4675); Mon, 28 Apr 2014 06:46:21 -0700
From: Paul Coverdale <>
To: 'Steev James' <>, 'Jean-Marc Valin' <>
References: <>
In-Reply-To: <>
Date: Mon, 28 Apr 2014 09:46:18 -0400
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="----=_NextPart_000_0044_01CF62C6.B4F95700"
X-Mailer: Microsoft Office Outlook 12.0
Thread-Index: Ac9iw9uhCCquOqr/QPOOwKLn2TaMvAAIef9A
Content-Language: en-us
X-OriginalArrivalTime: 28 Apr 2014 13:46:21.0698 (UTC) FILETIME=[3D463620:01CF62E8]
Subject: Re: [rtcweb] Query Regarding Mandatory audio codecs draft
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Real-Time Communication in WEB-browsers working group list <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Mon, 28 Apr 2014 13:46:25 -0000

I think we’re mixing up sampling frequency and bit-rate here.


By definition, G.711 uses a sampling frequency of 8 KHz and encodes each sample with an 8-bit logarithmically companded, encoding law (A or Mu), resulting in an output bit-rate of 64 kbit/s. 


Opus can operate with a number of different sampling frequencies (depending on the desired audio bandwidth) and output bit-rates, including 64 kbit/s.


The point I was trying to make is that, even if both codecs provide the same numerical bit-rate, they are not compatible between end-points. If a legacy end-point can only support G.711, it’s no good sending it anything else but G.711.




From: Steev James [] 
Sent: Monday, April 28, 2014 5:26 AM
To: Jean-Marc Valin
Subject: Re:Query Regarding Mandatory audio codecs draft


Means you say that G.711 uses less bandwidth than Opus at the expense of complexity.

Am I right ?


On Mon, Apr 28, 2014 at 11:23 AM, Jean-Marc Valin <> wrote:

Hash: SHA1

So, if the other endpoint only supports G.711 (legacy), then it
doesn't matter which codec you offer first. The the other endpoint
supports both Opus and G.711, but only at 8 kHz, then it comes down to
which of complexity or bandwidth is more important. On a corporate LAN
you'd likely want to use G.711 while for chatting from your home,
you'd usually want Opus. This is why the draft makes has no SHOULD
when it comes to narrowband.


On 28/04/14 12:40 AM, Steev James wrote:
> Paul, What do you say, if another end point is using both G.711 and
> Opus at 8kHz and why ?
> Steev.
>>> Is that means spec is recommending G.711 (PCMA/PCMU) for
>>> sampling
> rate
>>> less than or equal to 8 kHz?
>> No. Opus is still a good idea for 8 kHz, but it's not 2119
>> RECOMMENDED since there are valid (mostly complexity-related)
>> reasons to use G.711 despite the higher bit-rate.
>> Jean-Marc
> Opus isn't a good idea for 8 KHz if you are trying to inter-operate
> with another end-point using G.711.
> ...Paul

Version: GnuPG v1
Comment: Using GnuPG with Thunderbird -