Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!accuvax.nwu.edu!nucsrl!telecom-request From: Tad Cook Newsgroups: comp.dcom.telecom Subject: Re: Touchtone Detection Question Message-ID: <10605@accuvax.nwu.edu> Date: 8 Aug 90 19:38:34 GMT Sender: news@accuvax.nwu.edu Organization: very little Lines: 43 Approved: Telecom@eecs.nwu.edu X-Submissions-To: telecom@eecs.nwu.edu X-Administrivia-To: telecom-request@eecs.nwu.edu X-Telecom-Digest: Volume 10, Issue 551, Message 9 of 13 In article <10356@accuvax.nwu.edu>, gmc@wisvr.att.com (Glenn M Cooley) writes: > Some/most systems I've come across which have you enter data through > TT are able to correctly decode my input, long pulses, short pulses, > quick pulses, Bell phones, non-Bell phones. Other systems, such as > various answering machines are very fickle. I have to master a > certain pressing technique and can only use certain phones (non-PBX > Bell phones are the best) and still need to use several tries. > Why/comments/etc? One of the problems with designing a good quality DTMF receiver is insuring talk-off (falsing) immunity. Talk-off is when the receiver falsely detects a digit due to non-DTMF audio ... such as when you are dialing the payphone in the bar next to the jukebox. If music is playing (or you are talking) and it produces a momentary condition where two frequencies exist within the bandpass for the tones, a tone receiver could detect the false digit. One way to make this less likely is the make the tone acceptance bandwidth for each tone tighter. Another way is to make the "twist" acceptance (the difference in level between the high and low tone) more restrictive. If you do this, you can have another problem ... some phones may not be able to signal your receiver. This makes DTMF receiver design tricky. If you are building an answering machine with a really cheap DTMF receiver (maybe just some filters to detect a couple of digits) one way to prevent false digit detection would be to lengthen the time required for detection. Good DTMF receivers detect tones down to 40 ms, with 40 ms interdigit time. If you lengthen this to 500 ms, you will "never" get falsing, as the chances of the two tones being present in speech or on the jukebox for a half second is remote. Tad Cook Seattle, WA Packet: KT7H @ N7HFZ.WA.USA.NA Phone: 206/527-4089 MCI Mail: 3288544 Telex: 6503288544 MCI UW USENET:...uw-beaver!sumax!amc-gw!ssc!tad or, tad@ssc.UUCP