Re: DTMF v4

On 2012-12-14 22:10, Roman Shpount wrote:
> WebRTC-SIP Interworking Requirements Draft states that:
>
> A3-4 WebRTC MUST provide a means for the Javascript application to 
> invoke [RFC4733] DTMF events to be generated, and their duration, with 
> a default duration of 100ms.
>
> A3-6 WebRTC MUST NOT generate [RFC4733] events closer than 50ms 
> back-to-back.  In other words, even if                      the 
> Javascript calls the API repeatedly or provides a string of digits to 
> send, the browser must enforce a minimum of 50ms inter-event gap.
>
> There are several fairly old recommendations, such as ITU-T V.18, on 
> which these values are based, but the bottom line that these values 
> are usually considered safe.
>
ITU-T Recommendation Q.24 is the origin of the timing of DTMF.
http://www.itu.int/rec/T-REC-Q.24-198811-I/en
A table is given in Annex A for some countries.


It seems that a minimum 40 ms signal and minimum 30-70 ms gap are common 
requirements.
( the 70 ms gap is Australia. I suggest that you check if they still 
require that value. )

So 50  ms signal and 50 ms gap sound good for real produced values 
except for what is said for Australia in Q.24.
But some countries seemed to require a sum of signal + gap to be more 
than 120 or 125 ms, while more common seems to be 100 ms.

I assume that the duration mentioned in A3-4 is meant to be signal time 
plus silence time. Right?

/Gunnar
> _____________
>
> Roman Shpount
>
> On Fri, Dec 14, 2012 at 3:53 PM, Martin Thomson 
> <martin.thomson@gmail.com <mailto:martin.thomson@gmail.com>> wrote:
>
>     On 14 December 2012 12:40, Harald Alvestrand <harald@alvestrand.no
>     <mailto:harald@alvestrand.no>> wrote:
>     > Bernard mentioned a minimum gap (I think he said 50 ms), but I'm
>     not sure
>     > where that came from.
>
>     Minimum values come from experience with IVR interoperation for cases
>     where the tone is subsequently mixed into the audio.  Too short a tone
>     (or gap) and some IVRs fail to detect the tone.  I've seen minimum
>     tone duration of 100ms and gap of 50ms in some cases, though more
>     conservative values might be appropriate.  I have no good data to back
>     any specific choice.  I expect fluffy has more data than I.
>
>