A minimalist implementation of a vocal autotuner in Python, with no external dependencies except for Numpy.
Import doremi from the command line using PyPI:
pip install doremi
Then tune away:
import doremi
doremi.tune("examples/original.wav", "examples/Dmajor.wav", key="D", scale="major")That's all there is to it. All twelve notes in the Western chromatic scale are supported, and can be referenced through any valid enharmonic notation excluding double sharps and double flats (e.g. "F" is the same as "E#"). A frequency in hertz can also be passed in place of the key letter for more precise tuning control:
doremi.tune("input.wav", "tuned.wav", 442, "major")Valid arguments for the scale are major, minor, chromatic, and pentatonic.
Additional arguments for the attack of the tuning effect (i.e. the number of seconds before the tuning takes full effect) and the strength of the tuning effect (i.e. how much to apply the tuning, on a scale from 0 to 1) can also be passed in. For instance, setting attack to 0.1 and strength to 0.9 avoids the robotic 'AutoTune' effect, making the tuning more subtle:
doremi.tune("examples/original.wav", "examples/Gpentatonic.wav", key="G", scale="pentatonic", attack=0.1, strength=0.9)To produce a diagram showing the original and output pitches, set the 'debug' argument to True:
doremi.tune("examples/original.wav", "examples/Dpentatonic.wav", "D", "pentatonic", debug=True)Examples of tuned samples can be found in /examples.