Playing Fast and Loose with Music Recognition
summary
A prototype music-recognition system (Music Oats) lets musicians embed short note/rhythm sequences as triggers and evolves across two versions tested with 23 participants in four workshops. It uses MIDI or audio plus a feature-extraction plugin, supports textual codes, regular expressions for wildcards, and user-set inexact matching and logical preconditions. Audio users struggle with ambient artifacts, so version 2 adds velocity/pitch filters and record-to-code; overall it enables negotiated “looseness” but needs better live feedback and UI for stage use.
AI-written summary · about
chapters (6)
- 0:00Start
- 2:55Performance View
- 4:47Findings
- 7:36Logical Preconditions
- 8:03Setting Up the System 11
- 9:49Visual Feedback
related talks
43mMaking machines that make music
41mMusic and Machine Learning
25mFrom Beats to Bytes: What Music Production Can Teach About Programming
41mCreating Intelligent Music Instruments with Machine Learning
41mMaking Machines that Make Music
45mAI Lofi HipHop - How Not To Generate AI-Music
58mMusic Generation via Masked Acoustic Token Modeling
05mInteractive Exploration-Exploitation Balancing for Generative Melody Composition