Re: [codec] #5: Mention DTMF in requirements, Testing DTMF?

Steve Underwood <> Fri, 02 April 2010 06:23 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 5EA8E3A6899 for <>; Thu, 1 Apr 2010 23:23:59 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: 1.131
X-Spam-Level: *
X-Spam-Status: No, score=1.131 tagged_above=-999 required=5 tests=[BAYES_50=0.001, DNS_FROM_OPENWHOIS=1.13]
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id gY03LWpMW0Fu for <>; Thu, 1 Apr 2010 23:23:57 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 13F323A6A61 for <>; Thu, 1 Apr 2010 23:23:30 -0700 (PDT)
Received: from ( []) by with ESMTP id o326O27d009330 for <>; Fri, 2 Apr 2010 14:24:03 +0800
Message-ID: <>
Date: Fri, 02 Apr 2010 14:24:02 +0800
From: Steve Underwood <>
User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv: Gecko/20100301 Fedora/3.0.3-1.fc11 Thunderbird/3.0.3
MIME-Version: 1.0
References: <> <001d01cace82$1c8a0c80$559e2580$@de> <>
In-Reply-To: <>
Content-Type: text/plain; charset="UTF-8"; format="flowed"
Content-Transfer-Encoding: 8bit
Subject: Re: [codec] #5: Mention DTMF in requirements, Testing DTMF?
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: Codec WG <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Fri, 02 Apr 2010 06:23:59 -0000


I'm the author of the SpanDSP library people have referenced.

People seem to have only mentioned DTMF talk-off in these discussions, 
but the issue of corrupting the DTMF so it is less detectable is 
probably more of an issue.

Over the years there have been two reference data sets for testing DTMF 
talk off - the one from Mitel, and the one from Bellcore. Both started 
out as audio cassette tape products. The files I use for DTMF testing 
with SpanDSP are:

- A wave file produced by resampling to 8k samples/second the MP3 
version of the Mitel tape, which Mitel's heirs provide freely. If this 
is fed to TDM equipment it seems to give pretty much the same hits and 
near misses as the original Mitel cassette tape version. It seems, 
therefore, that aging of the original tape and transcription to MP3 have 
no unduly affected he signal.

- A set of wave files which were produced years ago from a pristine set 
of the Bellcore test tapes, using a cassette player with speed 
adjustment, so the pitch could be tweaked to be bang on target. Later, 
Telcordia/Bellcore stopped selling the tapes, and they gave me 
permission to hand out these files to a couple of people. Since they are 
transcribed from copyright tapes, I only handed them out where 
permission was sought. Now they seem to have introduced an audio CD 
version of the tapes.

You could use almost any long passages of speech for this kind of 
testing, but the above two data sets are the industry accepted ones, as 
you can compare different DTMF detectors reasonably well by comparing 
their false hits on this data. The Bellcore tapes are about 3 hours of 
audio, taken from the PSTN, where the call changes several times per 
seond. This is quite a tough test. The Mitel tape is about 30 minutes of 
audio passages recorded from the PSTN. Here they change the call every 
sentance or two, so the character os the test is somewhat different. It 
is much easier to get a good result with the Mitel tape, for various 
reasons, although I have found the Mitel tape a pretty good one for 
exercising talk-off issues in things like 2280Hz and 2600Hz signalling 
tone detectors.

The detection performance of DTMF detectors is usually quoted by their 
performance against the other data Mitel provided - a set of DTMF digits 
with various distortions of level, frequency and noise component. I took 
the spec for that data, and built a test suite which synthesises the 
relevant signals more precisely than the recordings on the Mitel tape. I 
encourage anyone interested in DTMF to make use of that work.


On 03/29/2010 06:17 AM, Wyss, Felix wrote:
> Christian,
> Telcordia sells the "Bellcore tape" DTMF talk-off immunity test 
> recordings as a set of 3 CDs:
> <>
> --Felix
> *From:* [] *On 
> Behalf Of *Christian Hoene
> *Sent:* Sunday, March 28, 2010 10:22
> *To:*
> *Subject:* Re: [codec] #5: Mention DTMF in requirements, Testing DTMF?
> Hi,
> I found some open source DTMF code at
> Does anybody has access to nice DTMF sample files such as those 
> describe in
> ?
> Christian
> ---------------------------------------------------------------
> Dr.-Ing. Christian Hoene
> Interactive Communication Systems (ICS), University of Tübingen
> Sand 13, 72076 Tübingen, Germany, Phone +49 7071 2970532
> *From:* Michael Knappe []
> *Sent:* Friday, March 26, 2010 11:38 PM
> *To:*;
> *Cc:*
> *Subject:* Re: [codec] #5: Mention DTMF in requirements
> I would agree with 'should'. Lot's of dtmf out there still, but also 
> out of band options (rfc 2833 / 4733) that make in-band carriage less 
> a concern for SIP telephony applications.
> Mik
> ------------------------------------------------------------------------
> *From*: <>
> *To*: Christian Hoene <>
> *Cc*: <>
> *Sent*: Fri Mar 26 18:10:31 2010
> *Subject*: Re: [codec] #5: Mention DTMF in requirements
> Perhaps analog DTMF carriage is a SHOULD?
> I agree that out-of-band is preferable, however, there still are 
> scenarios where coded DTMF tones will be sent.  It would be nice 
> (though IMHO not essential) for this carriage to work.
> Stephen Botzko
> On Fri, Mar 26, 2010 at 3:04 PM, Christian Hoene 
> < <>> wrote:
> Hi,
> > If the point of this issue to allow this codec to be used to 
> successfully
> > transit an IP network between two TDM networks,
> which is not the primary use-case.
> > I'd suggest that out-of-
> > band tone transport is far preferable to trying to ensure that the codec
> > will carry the tones adequately to allow them to be detected after
> > regeneration.
> I totally agree. Out-of-band is much better than inband.
> Sorry for my radical view but I believe that analog DTMF is an old and 
> obsolete technology not worth considering anymore.
> Christian