Simple Noise Activated Recording in Python. Previously known as vacr.
Python Shell
Latest commit c3d43ea Sep 13, 2013 @MerlijnWajer Merge pull request #2 from casebeer/dbfs-delta-silence-algo
dbFS delta silence algorithm



SNARP (Simple Noise Activated Recording in Python) is, as the title suggests, a program to record audio. It is primarily developed on the N900.

Eventually I'll split it up into a recording part and a Graphical frontend. The frontend would probably be specific to the N900 Maemo 5 QT Widget set.

Command Line Interface

To use SNARP from the UNIX command line, specify an output wav file to write to as a positional argument. By default, reads wav data from stdin:

$ cat test.wav | python output.wav

The audio in the output will have frames from silence sections of the input removed, but otherwise be a frame-for-frame copy of the input.

Reading data from the standard input makes it particularly easy to read from different audio devices using command line audio programs and pipes. Sox (cross platform) and arecord (Linux) are suggested. Using Sox' rec program to record system audio input, trimming out silences with SNARP:

$ rec -t wav - | python output.wav

To specify an input file rather than reading from the standard input, use the -i flag:

$ python -i test.wav output.wav

You may need to override the min and max sample levels used to define silent periods:

$ python -i test.wav --silence-min -100 --silence-max 100 output.wav

Other options and usage information can be found with python -h.

Original SNARP behavior

Use arecord via pipes to emulate SNARP's original default behavior:

$ arecord -D hw:0,0 -r 8000 |\
      python --silence-min 120 --silence-max 135 output.wav

And SNARP's original "podcaster" behavior:

$ arecord -D front:CARD=Podcaster,DEV=0 -r 48000 -f S24_3LE |\
      python --silence-min -1500000 --silence-max 1500000 output.wav