Algorithms Overview

A curated list of the most common audio analysis algorithms available in the Essentia API. Submit an analysis job with one or more of these model names to get started.

For a complete list including technical names and versioning, see the Complete Algorithm Reference.

Music Genre

genre_discogs_400

Music style (genre) classification by 400 styles from the Discogs taxonomy.

genre_mtgjamendoExperimental

Multi-label genre classification (87 classes).

Mood & Context

approachability

Predicts if music is accessible for the general public (mainstream vs. niche).

engagement

Predicts if music evokes active attention ("lean forward" vs. "lean back" listening).

danceability

Music danceability classifier (danceable/non-danceable).

arousal_valence

Music arousal and valence (the amount of energy and positiveness in the emotion).

mood_party

Mood classifier (party/non-party).

mood_relaxed

Mood classifier (relaxed/non-relaxed).

mood_sad

Mood classifier (sad/non-sad).

Tonality

key

Key and scale (major/minor) estimation.

tonal_atonalExperimental

Tonal/atonal music classification.

Voice Analysis

voice_instrumental

Classification of music by presence or absence of voice.

gender

Classification by singing voice gender (male/female). Only applies to music with vocals.

Instrumentation

acousticness

Acoustic/non-acoustic instrumentation.

electronicness

Electronic/non-electronic instrumentation.

Loudness

loudness_ebur128

Loudness measured by the EBU R128 standard (integrated, short-term, momentary).