Humanizing Music Sequences via 1/f Noise Timing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Machine-generated music, such as artificial drum sequences, lacks a human touch due to its perfect and predictable rhythm, which is perceived as unnatural by audiences, and even human-performed music can acquire an artificial touch through editing, leading to a desire for generating or modifying music with a more natural feel.

Innovation Solution

A method and device that introduce natural random fluctuations into music sequences by generating random offsets following a 1/fα power spectral density, similar to those found in human drumming, to create a more humanized rhythm, using Gaussian distributed 1/fα noise, which is empirically determined and applied to both computer-generated and recorded music.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If machine-generated music sequences are produced with precise clocking mechanisms, then rhythm precision and tempo stability are improved, but the music becomes perceived as artificial and lacks human touch

Engineering Contradiction:
Improverhythm precisionVSAvoidartificial perception
Core Design Contradiction:
Manufacturing precisionVSObject-affected harmful factors

Solution Approach 1:

The patent applies parameter changes by modifying the temporal parameters of musical events. Specifically, it introduces random fluctuations in the timing offsets of beats and sounds, transforming the rigid, precise timing of machine-generated music into a more natural, human-like rhythm pattern. This resolves the contradiction by maintaining acceptable rhythm precision while eliminating the artificial perception through controlled temporal variation.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent implements dynamics by making the rhythm pattern adaptive and variable rather than static. It dynamically adjusts the timing of musical events based on probabilistic models that simulate human performance characteristics, allowing the music to exhibit natural variations in tempo and rhythm. This dynamic approach preserves rhythm precision on average while introducing the variability that prevents artificial perception.

Inventive Principle:
Principle #15Dynamics

2Reliability

If human-performed music is edited through analogue or digital post-processing, then musical shortcomings are corrected, but the music acquires an undesired artificial touch

Engineering Contradiction:
Improveperformance qualityVSAvoidartificial touch
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent applies parameter changes by subtly modifying temporal parameters of edited music without dramatically altering the original performance characteristics. It introduces small, natural-sounding timing variations that correct performance inconsistencies while maintaining the authentic feel of human performance, thus improving reliability without adding artificial touch.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent uses copying by creating a modified version of the original performance that retains the essential characteristics of human playing. Rather than replacing the original performance with a perfectly generated one, it copies the performance and applies subtle temporal adjustments that preserve the human element while correcting shortcomings.

Inventive Principle:
Principle #26Copying

Data Source

PatentUS7777123B2Method and device for humanizing musical sequences
Publication Date: 2010.08.17 MAX PLANCK GESELLSCHAFT ZUR FOERDERUNG DER WISSENSCHAFTEN EV
  • US7777123B2 patent drawing
  • US7777123B2 patent drawing
  • US7777123B2 patent drawing

AI summary

A method for humanizing a music sequence (S), the music sequence (S) comprising a multitude of sounds (s1, . . . , sn) occurring on times (t1, . . . , tn) comprises the stepsgenerating, for each time (ti) a random offset (oi),adding the random offset (oi) to the time (ti) in order to obtain a modified time (ti+oi); andoutputting a humanized music sequence (S′) wherein each sound (si) occurs on the modified time (ti+oi).According to the invention, the power spectral density of the random offsets has the form1fα.wherein 0<α<2.