Voice Over Music Recording Audio Separation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing electronic devices lack the capability to efficiently record a user's voice over music without requiring specialized recording equipment, and they struggle to separate the user's voice from the music in real-time.

Innovation Solution

The system architecture enables electronic devices to record a user's voice over music by using one or more microphones to capture both the music output from a speaker and the user's singing, then separates the user's voice from the music using digital signal processing or machine learning models, allowing for recombination with digital audio content.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If electronic devices use standard microphones and speakers to record voice over music, then the ease of operation and accessibility are improved, but the ability to separate voice from music and achieve high-quality recordings deteriorates

Engineering Contradiction:

Improveease of operationVSAvoidrecording quality

Core Design Contradiction:

Ease of operationVSManufacturing precision

Solution Approach 1:

The patent replaces traditional mechanical audio recording systems with digital signal processing and machine learning models. The system uses software-based voice separation algorithms (such as neural networks) to distinguish and isolate the user's voice from the music, substituting complex mechanical separation equipment with intelligent digital processing that can accurately separate audio sources using standard device microphones.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Manufacturing precision

If electronic devices use specialized recording equipment to capture high-quality voice over music, then the recording quality is improved, but the device complexity and cost increase

Engineering Contradiction:

Improverecording qualityVSAvoiddevice complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent creates a virtual copy of the professional recording experience by using machine learning models trained on high-quality recording data. These models learn from extensive audio datasets to replicate the separation and processing capabilities of specialized equipment, allowing standard devices to achieve professional-quality results through software intelligence rather than hardware complexity.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system dynamically adjusts audio processing parameters in real-time based on the detected audio environment. The machine learning models analyze audio characteristics and automatically optimize processing parameters such as voice separation thresholds, noise suppression levels, and mixing ratios, eliminating the need for manual configuration and specialized equipment setup.

Inventive Principle:

Principle #35Parameter changes

3Productivity

If electronic devices process and separate voice from music in real-time, then the productivity and user experience are improved, but the energy consumption and computational load increase

Engineering Contradiction:

Improveprocessing speedVSAvoidenergy consumption

Core Design Contradiction:

ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent implements adaptive processing that applies full computational power only when voice separation is actively needed, rather than continuously processing all audio data. The system can switch between processing modes based on user interaction, applying intensive machine learning inference only during recording segments and using lighter processing for playback or idle states, thus reducing overall energy consumption while maintaining high productivity when required.

Inventive Principle:

Principle #16Partial or excessive action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This solution allows users to create high-quality recordings of their singing voice with prerecorded music without needing specialized equipment, enabling real-time playback and processing such as noise suppression and style matching.

Implementation Method 1

obtain, using one or more microphones of an electronic device, an audio input stream comprising: a voice of a person from a first source, and music from a second source that differs from the first source

Methodology Applied

Scientific EffectMicrophone transduction:

Data Source

PatentUS20250157443A1Voice over music recording for electronic devices

Publication Date: 2025.05.15 APPLE INC

AI summary

Systems, devices, and methods for voice over music recording are provided. A person may sing along with music being output into the physical environment of the person by a loudspeaker. One or more microphones may be used to record both the voice of the person singing, and the music output by the speaker. An electronic device may cancel the portion of the microphone signals containing the music output by the speaker, and remix the voice of the person with digital audio content corresponding to the music.