wav2json

Generate waveformjs.org compatible json data out of wav files.

Author: Benjamin Schulz

email: beschulz[the a with the circle]betabugs.de

License: GPL

Note: the file examples/waveform.js is from http://waveformjs.org/

This application is based on wav2png but does generate json data instead of pngs.

Examples

wav2json song.wav --channels left right -o song.json

might produce output like

{
    "left":[0,0.499969,0.865997,0.999969,0.865997,0.499969,0,0.5,0.866028,0,0.866028,0.5],
    "right":[0,0.499969,0.865997,0.999969,0.865997,0.499969,0,0.5,0.866028,0,0.866028,0.5],
}

Performance

Similar to wav2png (e.g. very good). But it might be a little slower, when generating output for multiple channels. Since the input file is opened multiple times, I'd recommend to write it to a temporary file first, so you can take advantage of the operating systems file cache:

sox ../song.mp3 -c 2 -t wav tmp.wav
wav2json tmp.wav --channels left right mid side min max -o song.json
rm tmp.wav

Also note, that you can dramatically reduce the file size, by reducing the precision of the output. In my tests, even a precision of 1 looked good.

Installation

On Linux (Ubuntu, Debian)

install dependencies

apt-get install make g++ libsndfile1-dev libboost-program-options-dev

Build

cd build
make all

On Max OS

install dependencies

• Get the Xcode command line tools
  • Starting with Xcode 4.3, Apple does not install command line tools by default anymore, so after Xcode installation, go to Preferences > Downloads > Components > Command Line Tools and click Install. You can also directly download Command Line Tools for Xcode without getting Xcode.
• Install homebrew
• install libsndfile: in the shell: brew install libsndfile

Build

• in the shell: cd build && make all

Usage

wav2json version 0.2
written by Benjamin Schulz (beschulz[the a with the circle]betabugs.de)

usage: wav2json [options] input_file_name
example: wav2json my_file.wav

Allowed options:

Generic options:
  -v [ --version ]      print version string
  --help                produce help message

Configuration:
  -s [ --samples ] arg (=800)           number of samples to generate
  --channels arg (=left right )         channels to compute: left, right, mid,
                                        side, min, max
  -p [ --precision ] arg (=6)           precision of the floats, that are
                                        generated. [1..6], reduce for smaller
                                        sized files. Usually 2 should be
                                        sufficient!
  -o [ --output ] arg                   name of output file, defaults to <name
                                        of inputfile>.json
  -c [ --config-file ] arg (=wav2json.cfg)
                                        config file to use
  -d [ --db-scale ]                     use logarithmic (e.g. decibel) scale
                                        instead of linear scale
  --db-min arg (=-48)                   minimum value of the signal in dB, that
                                        will be visible in the waveform
  --db-max arg (=0)                     maximum value of the signal in dB, that
                                        will be visible in the waveform.
                                        Usefull, if you now, that your signal
                                        peaks at a certain level.
  -n [ --no-header ]                    Do not include the version info banner
                                        in the output

One thing, thats noteworthy is, that you can generate output for multiple cahannels:

• left is the left channel of the audio file.
• right is the right channel of the audio file
• mid is the mid channel of the audio file computed per sample as: mid = (left + right) / 2
• side is the side channel of the audio file computed per sample as: side = (left + right) / 2
• max is the maximum of the channels of the audio file computed per sample as: max = max(left, right)
• min is the minimum of the channels of the audio file computed per sample as: min = min(left, right)

Obviously outputting the right, mid, side, min, max channels does only make sense on stereo input. When the input is mono, the applications falls back to outputing the left channel and prints a warning to stderr.

When there are more than two channels present, the output might be a little non-sensical, depending on what information is stored in the first and second channel.

Also note, that if there are less samples in the audio file, than defined by --samples, only as many samples as there are in the audio file will be written to the json file. The interpolation should be done by waveform.js

If you find any issues, feel free to contact me. and most important: enjoy and have fun :D