Re: [codec] #8: Sample rates?

Jean-Marc Valin <jean-marc.valin@usherbrooke.ca> Wed, 14 April 2010 05:32 UTC

Return-Path: <jean-marc.valin@usherbrooke.ca>
X-Original-To: codec@core3.amsl.com
Delivered-To: codec@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 7C0FA28C246 for <codec@core3.amsl.com>; Tue, 13 Apr 2010 22:32:20 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.001
X-Spam-Level:
X-Spam-Status: No, score=0.001 tagged_above=-999 required=5 tests=[BAYES_50=0.001]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Q-nsOWrZREx0 for <codec@core3.amsl.com>; Tue, 13 Apr 2010 22:32:19 -0700 (PDT)
Received: from relais.videotron.ca (relais.videotron.ca [24.201.245.36]) by core3.amsl.com (Postfix) with ESMTP id D3F6628C260 for <codec@ietf.org>; Tue, 13 Apr 2010 22:30:24 -0700 (PDT)
MIME-version: 1.0
Content-transfer-encoding: 7bit
Content-type: text/plain; charset="ISO-8859-1"; format="flowed"
Received: from [192.168.1.10] ([70.81.109.112]) by VL-MH-MR003.ip.videotron.ca (Sun Java(tm) System Messaging Server 6.3-8.01 (built Dec 16 2008; 32bit)) with ESMTP id <0L0U00LKQPYH34X0@VL-MH-MR003.ip.videotron.ca> for codec@ietf.org; Wed, 14 Apr 2010 01:30:17 -0400 (EDT)
Message-id: <4BC552E8.9090701@usherbrooke.ca>
Date: Wed, 14 Apr 2010 01:30:16 -0400
From: Jean-Marc Valin <jean-marc.valin@usherbrooke.ca>
User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.1.9) Gecko/20100317 Thunderbird/3.0.4
To: Roman Shpount <roman@telurix.com>
References: <062.89d7aa91c79b145b798b83610e45ce71@tools.ietf.org> <003101cadb1c$828b3990$87a1acb0$@de> <j2l6e9223711004130926nfaa975e3y129cc8cc21c52a84@mail.gmail.com> <m2v28bf2c661004130941g2e2bf956ld512b5d162df9080@mail.gmail.com> <g2h6e9223711004131029m3bfeb1ddq1a0e2bbd8418102f@mail.gmail.com> <m2s28bf2c661004131111pd7880c03m5f225ad464819414@mail.gmail.com> <s2i6e9223711004131143v3f3d2123pc94fe430a59b5776@mail.gmail.com> <CB68DF4CFBEF4942881AD37AE1A7E8C74AB3D92271@IRVEXCHCCR01.corp.ad.broadcom.com> <y2q6e9223711004131303l15fb87ffoe1039c56d21c565f@mail.gmail.com> <20100413164818.546929eae97cjjr6@mail.skype.net> <u2u28bf2c661004132137x5f673ff7j7a43a61572e5483e@mail.gmail.com>
In-reply-to: <u2u28bf2c661004132137x5f673ff7j7a43a61572e5483e@mail.gmail.com>
Cc: codec@ietf.org
Subject: Re: [codec] #8: Sample rates?
X-BeenThere: codec@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: Codec WG <codec.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/codec>, <mailto:codec-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/codec>
List-Post: <mailto:codec@ietf.org>
List-Help: <mailto:codec-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/codec>, <mailto:codec-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 14 Apr 2010 05:32:20 -0000

On 2010-04-14 00:37, Roman Shpount wrote:
> To be honest, your slides show that SILK works better at superwideband,
> not that superwideband is better then wideband.

Then I suggest you have a look at the left graph in slide 7 of: 
http://www.ietf.org/proceedings/10mar/slides/codec-6.pdf
The blue bars show speech quality and the bar that's marked with "7 kHz" 
means 7 kHz bandwidth, which means *uncompressed* wideband audio.

> Normally, G.711 at 8 KHz
> has a higher MOS (4.4) score then SILK at superwideband (3.9).

Note that MOS is not an absolute measurement and MOS scored can only be 
compared when they come from the same test.

> To
> compare wideband vs superwideband (or full rate) you need to compare 16
> Khz linear PCM vs superwideband (or full rate) linear PCM. As far as
> full rate is concerned, most men cannot hear above 12-14 khz.

I disagree here. Don't know where you got that data, but it doesn't 
sound right. FTR, like many others, I'm over 30 and still easily hear 
the NSTC hozontal sync (15.3 kHz) coming out of CRT TVs.

> In side by
> side listening tests we did, using high end headsets (AKG K-701,
> Sennheiser HD 650) and professional DAC/headphone AMP (Benchmark DAC-1)
> using prerecorded voice samples (English, adult voices, no kids), nobody
> could distinguish wideband from superwideband, but this can be us, our
> samples, and untrained listeners.

Well, my listening tests that I linked above clearly disagree with your 
results. Maybe you were using samples that were already low-passed at 7 kHz!

Not saying that wideband (and even narrowband) doesn't have a use, just 
that full-band is definitely useful.

Cheers,

	Jean-Marc