YAAA (Yet Another Audio Analyzer) is a simple program written in python. It is still under development.
The idea is having a very simple powerful program to look at an audio singnal's time and frequency domain data interactively.
Using pyo, yaaa is able to play back sound, but also provides a granular sampling engine, so one can scrub through the sound. Pyqtgraph provides a very powerful interface that supports rapid updates.
A peak finding algorithm (written by Marcos Duarte) is used to extract and display spectral peaks. These frequnecies and amplitudes can be examined in a table view and can be exported. (For example to create modal models in faust or Max/MSP or similar.)
This is still in a very early stage of development, and has just been tested under linux. Since it is based on python, it should work under other platforms too.
Use pipenv to install dependencies and run yaaa:
- install pipenv
- clone the repository
- cd into its directory
- use pipenv to install dependencies:
pipenv install --ignore-pipfile - run:
pipenv python ./yaaa
- python 3.6
- pyqtgraph
- pyo
- numpy
- scipy
- librosa
- numba
The peak finding algorithm used here was created by Marcos Duarte, https://github.com/demotu/BMC and is included.
Most of the dependencies can be installed via conda (https://conda.io/docs/user-guide/install/download.html)
-
conda install pyqtgraph -
conda install scipy -
conda install numba -
conda install -c conda-forge librosa -
For Pyo follow the instructions here: http://ajaxsoundstudio.com/pyodoc/download.html But if you are on linux, compile from sources with instructions here: http://ajaxsoundstudio.com/pyodoc/compiling.html
That should be it. After that, you should be able to start the yaaa executeable from the terminal.
On Linux and Mac it probably makes sense to create a symlink to yaaa:
ln -s /path/to/yaaa/yaaa /usr/bin/yaaa
If you have a specific conda environment with all the dependencies installed you will need to use that. A small function in the .bashrc (or similar) instead of the symlink can be used e.g.
yaaa(){
"path/to/my/envoronment/python" "path/to/yaaa/yaaa" "$@"
}
Yaaa has a simple command line interface as shown below. But it can also simply be started and a new file can be loaded via the GUI.
usage: yaaa [-h] [-f F] [-c C] [-o O]
Yet Another Audio Analyzer.
optional arguments:
-h, --help show this help message and exit
-f F file to open
-c C provide a number of seconds to extract from the file
-o O provide a number of seconds as an offset from the beginning.
- Adjustment of peak finding algorithm via GUI
- decay rate computation of spectral peaks
- Can get pretty slow for large files. This tool is thought for analysis of impulse responses or other small portions of audio. probably automatic cropping will be introduced.
- Installation should get easier
- There is probably a bug for out of bound indizes when cropping or offestting via the CLI. Simple sanity checks need to be introduced.