[asterisk-dev] DTMF detection and generation code

[Iain Barker]

Vahan Yerkanian wrote:
>Are there any plans or even chances for DTMF detection and generation 
>(inband, rfc2833 and info) code being looked at and rewritten/fixed?

Steve Underwood wrote:
>Adding duration to IAX2 DTMF messges is a bad idea. RFC2833 is brain 
>dead for having it (actually RFC2833 is just rather brain dead)

Just adding a few general comments to this discussion.

Tone duration may not be particularly relevant for Asterisk when it is doing simple digit collection, but it is important for other applications. One example: A "Press 0 to buzz in" door entry system needs accurate start and end indications, or the tone duration.

Also, please remember that RFC2833 is not just used for DTMF tones. RFC2833 NSE's are a generic transport mechanism which is also used for things like fax and modem tone reporting, and other call-progress events not related to the signalling path.

For example, some VoIP gateways will disable echo cancellers on receiving a /ANS, or will switch from using RTP to using T.38 UDPTL if a Fax CNG and CED event s received.

One other thing that would help Asterisk interop for basic DTMF tones, is the RFC2198 redundancy standard. Some hardware vendors chipsets widely used in media gateways from "big name" companies support sending *only* RFC2833 inside RFC2198 redundant frames.

- Iain