Timbre Creation System Using Deep Learning Patches

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current methods for recreating a sound from scratch in music production are time-consuming and costly, often requiring session musicians, sample packs, or preset packs, which are inefficient and expensive.

Innovation Solution

A computer-implemented timbre creation method that generates a digital fingerprint of a sound using timbre analysis filters and deep learning to create patches for a synthesizer, allowing for the reproduction of matching or complementary timbres.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If session musicians or voice actors are hired to record samples, then sound quality is improved, but cost and time consumption increase significantly

Engineering Contradiction:

Improvesound qualityVSAvoidtime consumption

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The system creates a digital fingerprint copy of the target sound's timbre characteristics and uses this fingerprint to generate synthesizer patches that replicate the sound, eliminating the need to record actual samples with musicians

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical process of recording with musicians and voice actors with an automated computational system that analyzes timbre characteristics and generates synthetic patches through algorithmic processing

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Loss of time

If sample packs or preset packs are purchased and modified, then time consumption is reduced, but cost increases and unique matching is limited

Engineering Contradiction:

Improvetime consumptionVSAvoidsound matching accuracy

Core Design Contradiction:

Loss of timeVSAdaptability or versatility

Solution Approach 1:

The system performs timbre analysis on the target sound to extract characteristic features, uses these features to generate initial patches, and iteratively refines the patches by comparing them against the original sound's fingerprint until optimal matching is achieved

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system analyzes and extracts specific timbre parameters from the target sound, then adjusts synthesizer patch parameters to match these extracted characteristics, enabling precise customization rather than relying on pre-fixed presets

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If deep learning is performed to create patches matching the digital fingerprint, then timbre matching accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvetimbre matching accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs timbre analysis and fingerprint extraction beforehand to pre-process the target sound into structured feature data, which simplifies the subsequent deep learning patch generation process by providing ready-to-use target characteristics

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11158297B2Timbre creation system

Publication Date: 2021.10.26 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A timbre creation method, system, and computer program product include performing a timbre analysis of a sound from an input source to generate a digital fingerprint of the sound, performing deep learning to create a patch that matches the digital fingerprint, and generating a second patch for a synthesizer which reproduces a timbre that complements the digital fingerprint based on the patch.