auditok

Auditok is an open-source Python library for audio activity detection and segmentation. It provides tools to identify speech and non-speech regions within long recordings and to split the audio into shorter segments suitable for transcription or analysis.

Core functionality includes a configurable voice activity detector and segmenter. Auditok can process audio files or in-memory arrays and returns a list of detected segments with start and end times. Users can adjust parameters such as window length, hop size, energy threshold, minimum and maximum segment duration, and padding between segments. The library relies on simple heuristics, typically energy-based, with post-processing to merge or filter segments.

Implementation and compatibility: Auditok is written in Python and uses common scientific libraries such as NumPy and SciPy. It can read standard audio formats and produce segments compatible with downstream tools. Segment results can be exported as separate audio files or accompanied by metadata for further processing.

Applications and limitations: It is designed to preprocess long recordings—such as meetings, lectures, or interviews—by extracting speech portions to speed transcription or analysis. It is not a full speaker diarization solution, and performance may vary with noise and recording quality. The project is open-source and welcomes community contributions.

Availability: Auditok is available on PyPI and GitHub under an open-source license. Installation typically proceeds via pip install auditok.