Voice Separation Using Ultrasonic Synchronization and Calculated Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Degenerate Unmixing Estimation Technique (DUET) algorithm for voice separation faces challenges in selecting an appropriate range for clustering relative delay and attenuation, leading to potential phase wrap effects due to ambiguous range selection and inadequate time synchronization, especially with high sampling frequencies and large synchronization errors.
Innovation Solution
A method and system that calculate and select a clustering range for relative delay parameters based on the distance between microphones and sampling frequency, using Fourier transforms and inverse Fourier transforms, and achieve time synchronization through ultrasonic signals to eliminate synchronization errors, ensuring accurate voice separation without phase wrap effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional time synchronization method is used, then system complexity is reduced, but time synchronization precision deteriorates due to millisecond-level system time vs. tens of kilohertz sampling frequency
Solution Approach 1:
The patent introduces an ultrasonic signal as an intermediary carrier to transmit timing information between devices. The ultrasonic signal serves as a mediator that bridges the gap between millisecond-level system time and microsecond-level sampling frequency, enabling precise time synchronization without complex hardware modifications.
Solution Approach 2:
The patent replaces the traditional mechanical/time-based synchronization system with an acoustic-based system using ultrasonic signals. This substitution leverages the physical propagation characteristics of sound waves to achieve precise timing, replacing reliance on system clock synchronization with acoustic signal-based timing extraction.
2Measurement precision
If clustering range is selected based on experience, then implementation is simple, but voice separation accuracy deteriorates due to phase wrap effects from invalid data points
Solution Approach 1:
The patent performs preliminary calculations of the clustering range based on known parameters (microphone distance, sampling frequency, maximum frequency) before executing the voice separation algorithm. This preliminary action ensures that the clustering range is optimally configured to avoid phase wrap effects, eliminating the need for trial-and-error adjustments during implementation.
Solution Approach 2:
The patent transforms the clustering range from an empirically selected parameter to a calculated parameter based on physical system characteristics. By changing how the parameter is determined (from experience-based to formula-based), the system achieves higher accuracy while maintaining simple implementation through deterministic calculation.
3Reliability
If DUET algorithm is applied without proper time synchronization, then algorithm simplicity is maintained, but separation reliability deteriorates due to synchronization errors at high sampling frequencies
Solution Approach 1:
The patent implements time synchronization as a preliminary step before applying the DUET algorithm. By synchronizing timestamps using ultrasonic signals in advance, the system ensures that subsequent voice separation operations rely on accurately time-aligned data, significantly improving reliability without complicating the core algorithm.
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 approach enables efficient and accurate voice separation by automatically selecting clustering ranges and achieving precise time synchronization, reducing computational complexity and avoiding phase wrap issues, thus improving the DUET algorithm's performance in various environments.
Implementation Method 1
performing a Fourier transform on the received signals
Implementation Method 2
performing an inverse Fourier transform on each subsets
Implementation Method 3
achieve time synchronization through ultrasonic signals to eliminate synchronization errors
Data Source
AI summary
The present disclosure provides method and system for voice separation based on DUET algorithm, and the method comprises receiving signals from microphones; performing a Fourier transform on the received signals; calculating a relative attenuation parameter and a relative delay parameter for each data point; selecting a clustering range for the relative delay parameters based on a distance between the microphones and a sampling frequency of the microphones, clustering the data points within the clustering range for the relative delay parameters into subsets, and performing an inverse Fourier transform on each subsets. According to the present disclosure, it is possible to provide an efficient and intelligent solution to deploy DUET on the software and/or hardware.