There was a post:
"Re: [OT]: DTMF decoding software"
*****************************************
Edson Brusque wrote:
>     I have a bunck of wav/mp3 files wich I need to
> extract the DTMF informations. I've been looking
> for some weeks for a simple software that can do
> DTMF decoding from recorded wav files.
>     Do you know some software (preferably for
> Windows) that can do this?

   Ask folks who attended Microchip Master's last
summer. (Are there any out there?). They might
comment "DTMF decoding on PIC18"

http://techtrain.microchip.com/masters2002/Classes/608/608 DTF.pdf

Interesting document on "Design of DTMF Detector
on PICR Microcontroller with 1-bit DFT using modified
Goertzel algorithm"

You may develop your own software using the idea.

   Mike.
*******************************************


Scott Dattalo wrote:
+++++++++++++++++++++++++++++++++++++++++++
On Tue, 29 Jul 2003, Dale Shelor wrote:

> Hello,
>
> What I'm looking to do is decode (or detect the present of) a sub
> audible audio tone using a PIC.  The tones are received on a radio
> receiver and range from 67.0Hz to 254.1Hz.  Accuracy needs to be
> around -+1%.  Faster the better!  False triggers are acceptable but
> need to be minimal.  Anyone ever done this or have tips on how to
> accomplish this?  I'm using CCS C but can always snip in ASM code.
> Code samples would be great!
>
> GE did this with an Intel 8049 at 6MHz, surely it can be done with a
> PIC 16F877 at 20MHz!  GE used a low pass filter to filter out all
> audio above 300Hz, then sent the low frequency audio into the CPU and
> decoded the tones.  GE did this very fast and accurate.


If you can use an analog filter to remove the high frequencies then it's
fairly straight forward to perform tone decoding with a PIC. For a
theoretical background, checkout:

http://www.dattalo.com/technical/theory/dtmf.html

I've used the techniques described there to decode DTMF and CAS tones.
Your frequencies are an order of magnitude smaller and consequently much
easier to decode. OTOH, DTMF consists of only 16 tones. You want much
finer resolution. It's possible to construct filter banks to improve the
resolution. For example, you can create binary banks that filter into
high and low frequencies. The filter that produces the stronger signal
is further subdivided and searched. After 7 levels, you'll resolve to 1
part out of 128. This technique only works if there really is just one
tone present.

Another approach is to break your frequency range into more than 2
banks. For example, you could pick 8 banks, determine which has the
strongest signal and then further subdivide that bank; etc. However,
from a computational point of view it amounts to about the same work.
When your dealing with two frequencies at once, you can resolve them
much more quickly/accurately than when you attempt to resolve 8
frequencies at once.

Scott
++++++++++++++++++++++++++++++++++++++++++++++

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email listserv@mitvma.mit.edu with SET PICList DIGEST in the body