SVAR - Simple Voice Activated Recorder

It is a simple console application (low memory footprint and CPU usage) designed for recording audio when a specified signal level is exceeded. It is commonly known solution called Voice Operated Recording (VOR). When the signal level is low for longer than the fadeout time, audio recording is paused.

On Linux systems, capturing the audio signal is based on the ALSA technology. For all other systems, PortAudio library will be used. Alternatively, it is possible to force PortAudio back-end on Linux systems by adding -DENABLE_PORTAUDIO=ON to the CMake configuration step.

Currently this application supports four output formats:

• RAW (PCM 16bit interleaved)
• WAV (libsndfile)
• MP3 (mp3lame)
• OGG (libvorbis)

For low CPU consumption WAV is recommended - it is the default selection.

There is also possible to split output file into chunks containing continuous recording. New output file is generated every time a new signal appears (after the split time period). In such a case, the time of signal appearance can be determined by the output file name, which by default is in the format of "rec-DD-HH:MM:SS". It is possible to customize it with a strftime(3) format string.

For the fine adjustment of the activation condition (the signal level), one can run svar with the --sig-meter parameter. This activates the signal meter mode, in which the maximal peak value and the RMS is displayed. Activation threshold is based on the maximal peak value in the signal packed (time of tenth of the second).

Installation

mkdir build && cd build
cmake .. -DENABLE_SNDFILE=ON -DENABLE_MP3LAME=ON -DENABLE_VORBIS=ON
make && make install