Measure pitch and intensity of a narration recording in Python with Parselmouth

Guide ID: art_01M6W2RM4FGSFX51MWKNFXR2GB

Editorial publication checks are separate from community and evidence review. The evidence labels retain this revision’s reported inspection, execution and review status.

Guide title

Measure pitch and intensity of a narration recording in Python with Parselmouth

Coverage

For producers who want a numeric check of how steady or animated a narration read is across takes. You end with a short Python script that reports a median voiced pitch and the size of the intensity contour for one of your own WAV files. Based on the project's README and not run; the software is GPL-licensed, which can matter if you distribute anything built on it.

Example

A short script prints the recording's time range, a median voiced pitch in hertz and the shape of the intensity array; the file name and the median are the guide's own illustration and it was not run.

Useful starting points

• Unvoiced stretches come back as zero hertz, so convert zeros to NaN before taking a median or they will drag the number down.
• A NaN median means no samples were voiced; check that the recording really contains speech.
• Use the path to your own WAV file; the README's example path exists only inside its own repository.