Decoder for DTMF (and other MF signals) in Python
Small Python script for decoding DTMF tones from .WAV files.
pip install numpy
pip install scipy
pip install argparse
pip install csv
Running the program with the -h argument will produce the following:
C:\ToneDecoder\>python.exe main.py -h
usage: main.py [-h] [-f FRAME_MS] [-pf PEAK_FACTOR] [-st SILENCE_THRESHOLD] [-nts] filename csv_files [csv_files ...]
DTMF decoder
positional arguments:
filename Path to the input .WAV file
csv_files One or more CSV file(s)
options:
-h, --help show this help message and exit
-f FRAME_MS, --frame_ms FRAME_MS
Frame rate in ms (integer, optional. Default=50)
-pf PEAK_FACTOR, --peak_factor PEAK_FACTOR
Peak factor (float, optional. Default=0.1)
-st SILENCE_THRESHOLD, --silence_threshold SILENCE_THRESHOLD
Silence threshold (float, optional. Default=0.02)
-nts, --no_time_stamp
Flag to disable timestamps (default: False)
We can then check the file perfect-example.wav against the dtmf.csv file:
C:\ToneDecoder\>python.exe main.py perfect-example.wav dtmf.csv
00:00:00.000000000 -> 0
00:00:00.200000000 -> 1
00:00:00.400000000 -> 2
00:00:00.600000000 -> 3
00:00:00.800000000 -> 4
00:00:01.000000000 -> 5
00:00:01.200000000 -> 6
00:00:01.400000000 -> 7
00:00:01.600000000 -> 8
00:00:01.800000000 -> 9
Note however, that results may not be so perfect without clear audio. For example, attempting to decode not-perfect-example.wav we get:
C:\ToneDecoder\>python.exe main.py not-perfect-example.wav dtmf.csv
00:00:00.400000000 -> *
00:00:00.700000000 -> 7
00:00:00.750000000 -> 3
00:00:00.850000000 -> 7
00:00:00.900000000 -> D
00:00:00.950000000 -> 0
00:00:01.250000000 -> 6
00:00:01.500000000 -> B
00:00:01.550000000 -> 2
00:00:01.600000000 -> 1
00:00:01.650000000 -> 2
00:00:01.700000000 -> 7
00:00:01.750000000 -> #
00:00:01.800000000 -> 7
00:00:01.850000000 -> 4
00:00:02.050000000 -> 9
00:00:02.250000000 -> 2
00:00:02.400000000 -> 5
00:00:02.550000000 -> 8
00:00:02.600000000 -> *
00:00:02.650000000 -> C
00:00:02.700000000 -> D
00:00:02.750000000 -> 9
00:00:02.800000000 -> 7
00:00:02.850000000 -> 8
00:00:02.900000000 -> 1
00:00:02.950000000 -> A
00:00:03.000000000 -> 3
00:00:03.050000000 -> 6
00:00:03.100000000 -> #
00:00:03.150000000 -> 3
00:00:03.300000000 -> D
00:00:03.350000000 -> 8
00:00:03.400000000 -> 2
00:00:03.450000000 -> *
00:00:03.900000000 -> 8
00:00:03.950000000 -> 4
00:00:04.250000000 -> 7
00:00:04.300000000 -> 8
00:00:04.350000000 -> 5
00:00:04.450000000 -> 8
00:00:04.500000000 -> 5
00:00:04.700000000 -> #
00:00:04.800000000 -> 5
00:00:04.850000000 -> 8
00:00:04.900000000 -> C
00:00:05.000000000 -> 2
00:00:05.050000000 -> 6
00:00:05.250000000 -> 9
00:00:05.300000000 -> 6
00:00:05.350000000 -> 9
00:00:05.400000000 -> #
00:00:05.550000000 -> D
00:00:05.800000000 -> *
00:00:05.850000000 -> 7
00:00:05.900000000 -> 8
00:00:05.950000000 -> 7
00:00:06.050000000 -> C
00:00:06.100000000 -> *
00:00:06.150000000 -> 7
00:00:06.600000000 -> #
00:00:06.800000000 -> 8
00:00:06.950000000 -> C
00:00:07.050000000 -> D
00:00:07.100000000 -> 9
00:00:07.350000000 -> 8
00:00:07.450000000 -> *
00:00:07.500000000 -> 9
00:00:07.700000000 -> #
00:00:07.750000000 -> 9
00:00:07.800000000 -> D
00:00:07.850000000 -> 7
00:00:07.950000000 -> 0
00:00:08.000000000 -> 8
00:00:08.050000000 -> 7
00:00:08.200000000 -> 0
00:00:08.250000000 -> 9
00:00:08.300000000 -> 0
00:00:08.350000000 -> #
00:00:08.450000000 -> *
00:00:08.500000000 -> #
00:00:08.600000000 -> 3
00:00:08.650000000 -> 9
00:00:08.750000000 -> 0
If we try adjusting the frame-rate and peak-factor we get:
C:\ToneDecoder\>python.exe main.py not-perfect-example.wav dtmf.csv -f 100 -pf 0.4
00:00:00.900000000 -> 0
00:00:01.500000000 -> 1
00:00:02.200000000 -> 2
00:00:02.900000000 -> 3
00:00:03.900000000 -> 4
00:00:04.300000000 -> 5
00:00:05.000000000 -> 6
00:00:05.900000000 -> 7
00:00:06.800000000 -> 8
00:00:07.600000000 -> 9
If we do not wish to see the timestamp, we can use the -nts flag. So for example we decoding audiocheck.net_dtmf_123A456B789C_0#D.wav we get:
C:\ToneDecoder\>python.exe main.py audiocheck.net_dtmf_123A456B789C_0#D.wav dtmf.csv
123A456B789C*0#D
Finally, in addition to the dtmf.csv file there a ccitt5.csv file for CCITT-5 tones, and us_red_box.csv for US Red Box tones.