Voice Processing Power Reduction via Multi-Core Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile devices face increased power consumption due to advanced functionalities and varied usage patterns, particularly in voice communications, where existing methods do not effectively optimize power usage for clear voice communications.
Innovation Solution
Implementing speech codecs on a CPU with SIMD extensions and post-processing modules like acoustic echo cancellation and noise reduction on a DSP core, distributing these functions across different processor cores to reduce current consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speech codecs and post-processing modules are implemented on a single DSP/VLIW processor, then clear voice communication is achieved, but current consumption increases
Solution Approach 1:
The patent divides the voice processing system into separate modules implemented on different processor cores: speech encoding/decoding on CPU with SIMD extensions, acoustic echo cancellation on DSP core, and noise reduction on DSP core. This segmentation allows each module to run on the most suitable processor, reducing overall power consumption while maintaining voice communication quality.
Solution Approach 2:
The patent applies local quality by matching specific processing tasks with appropriate processor architectures: CPU with SIMD for speech codecs, DSP for echo cancellation and noise reduction. This ensures each part of the system has the quality characteristics needed for its specific function, optimizing overall efficiency.
2Reliability
If multiple post-processing modules are added for clear voice communication, then voice quality improves, but device complexity increases
Solution Approach 1:
The patent segments voice processing into distinct functional modules (speech encoding, echo cancellation, noise reduction) that can be independently implemented and optimized on different processor cores, managing complexity through modular architecture.
Solution Approach 2:
The patent uses a multi-functional SoC platform that integrates CPU with SIMD extensions and DSP cores, allowing a single device to handle multiple voice processing functions efficiently without requiring separate dedicated hardware for each function.
Data Source
AI summary
The present invention provides a system and method for optimizing power consumption in voice communication in mobile devices. The system comprises pre-processing modules, a speech codec encoder module, a speech codec decoder module and post-processing modules. The pre-processing and post-processing modules are implemented on a DSP/VLIW processor, while the speech encoder and decoder modules are implemented on a CPU with SIMD extensions. This pipelined implementation of modules in multi-core reduces current consumption in the SoC by up to 50 percent compared to an implementation of the modules in a single DSP/VLIW core. The significant reduction in current consumption of the modules enables reduction of power consumption in the talk time.