Audio Decomposition Using Precomputed Intermediate Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio processing methods require significant computational resources and time to decompose mixed audio signals into individual musical timbres, which can be unsatisfactory on devices with limited hardware resources, such as mobile phones or standalone DJ systems.
Innovation Solution
A method that involves preparing intermediate data in advance, either locally or remotely, which can be quickly loaded and combined with input audio data to separate musical timbres, reducing the need for resource-intensive calculations and allowing for efficient processing even on devices with limited resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If real-time audio decomposition is performed using artificial intelligence systems, then decomposed tracks can be obtained during playback, but processing time and resource consumption increase significantly
Solution Approach 1:
The patent applies preliminary action by pre-computing decomposition results for specific time segments of audio data before actual playback. The system divides the audio data into segments, performs AI-based decomposition on each segment in advance to obtain intermediate data, and stores these results. During playback, only the final combination step is needed, dramatically reducing real-time processing requirements and energy consumption while maintaining the ability to provide decomposed tracks during playback.
2Ease of operation
If audio decomposition is performed on devices with limited hardware resources, then portability and accessibility are improved, but processing capability and decomposition quality deteriorate
Solution Approach 1:
The patent applies segmentation by dividing the audio data into multiple time segments and processing each segment independently to create intermediate data. This segmentation allows the system to perform complex AI decomposition on manageable portions of audio data in advance, then efficiently combine these segments during playback on resource-constrained devices. This approach enables high-quality decomposition results to be achieved while maintaining compatibility with devices having limited hardware resources.
Data Source
AI summary
A method for processing audio data, comprising providing song identification data identifying a particular song from among a plurality of songs or identifying a particular position within a particular song, loading intermediate data associated with the song identification data from a storage medium or from a remote device. The method also comprises obtaining input audio data representing audio signals of the song as identified by the song identification data. The audio signals comprise a mixture of different musical timbres, including at least a first musical timbre and a second musical timbre different from said first musical timbre. The method comprises combining the input audio data and the intermediate data with one another to obtain output audio data. The audio data represent audio signals of the first musical timbre separated from the second musical timbre.