Sound Source Localization via Acoustic Wave Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Electronic devices face performance degradation due to acoustically reflective surfaces, which hinder accurate sound source localization and speech recognition by confusing sound reflections, leading to poor user experience.
Innovation Solution
Implementing sound source localization using acoustic wave decomposition (AWD) with Multi-Channel Linear Prediction Coding (MCLPC) coefficients to decompose sound fields into disjoint acoustic plane waves, allowing devices to identify the direction of speech by calculating noise statistics and signal quality metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional sound localization methods are used in environments with acoustically reflective surfaces, then the device can operate with simple processing, but the localization accuracy deteriorates due to confusion from sound reflections
Solution Approach 1:
The patent segments the acoustic signal into distinct components using acoustic wave decomposition (AWD), separating the direct sound path from reflected sound paths. This allows the system to process and analyze each wave component independently, improving localization accuracy by identifying the direct path signal while filtering out reflections from acoustically reflective surfaces.
Solution Approach 2:
The patent introduces Multi-Channel Linear Prediction Coding (MCLPC) coefficients as an intermediary tool to model and predict the acoustic environment. These coefficients serve as a mediator between the raw microphone signals and the final localization decision, enabling the system to average out room impulse responses and distinguish direct sound from reflections through statistical analysis.
2Measurement precision
If the device processes audio data to improve speech recognition in noisy environments, then speech recognition accuracy improves, but computational resources and processing time increase
Solution Approach 1:
The patent performs preliminary acoustic wave decomposition and noise statistics calculation before the main speech recognition process. By pre-processing the audio data to separate direct sound from reflections and pre-calculating noise characteristics, the system reduces the computational burden on subsequent speech recognition stages, improving overall efficiency while maintaining accuracy.
Solution Approach 2:
The system uses the microphone array's own captured audio data to generate noise statistics and signal quality metrics without requiring external reference signals. This self-service approach allows the device to adapt to its specific acoustic environment autonomously, improving speech recognition in varying conditions without additional computational overhead from external calibration sources.
Data Source
AI summary
A system that performs sound source localization (SSL) using acoustic wave decomposition (AWD) or an approximation. When a device detects a wakeword represented in audio data, the device performs SSL processing in order to determine a position of the user relative to the device (e.g., estimate angle of the user). The device calculates noise statistics based on first audio data representing the wakeword and second audio data preceding the wakeword. Thus, upon detecting the wakeword, the device calculates the noise statistics and a signal quality metric corresponding to the wakeword. In addition, the device uses Multi-Channel Linear Prediction Coding (MCLPC) coefficients to average out the room impulse response. Using the noise statistics, the MCLPC coefficients, and the audio data, the device performs AWD processing to decompose the sound field to disjoint acoustic plane waves, enabling the device to identify the most likely direction for the line-of-sight component of speech.