SPIRALSonic Production Intelligence Lab
← All tools & apps

BFSegmenter Extractor

Sound-feature tool · macOS

Scores a folder of sounds for valence, arousal and Bark-band features, ready for analysis.

BFSegmenter Extractor interface

Overview

Extracts high-level sound descriptors — valence, arousal, and Bark-frequency features — from a folder of audio files, using the Essentia MIR library. Built to feed the lab's generative sound-design and soundscape-affect research.

How it works

BFSegmenter Extractor runs every audio file in a folder through the Essentia music-information-retrieval library and writes the results to a CSV file, one row per sound.

The features feed the lab's work on high-level sound descriptors. A model trained on them places any sound in a continuous valence–arousal space, and 36 pairs of high-level descriptors — aggressive/peaceful, chaotic/ordered and so on — are projected into that same space, so a sound can be scored against descriptors no one manually tagged it with.

The descriptor code is open source on GitHub.