Periodic Interference Detection in Speech Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Speech processing systems face challenges in identifying and isolating periodic interference, such as electromagnetic noise, which can be mistaken for speech, leading to reduced intelligibility and quality, especially in low-processing-power devices like vehicles or hand-held systems.
Innovation Solution
A system that encodes a limited frequency band by varying pulse amplitude, separates signals into frequency bins to identify harmonic signals by comparing average acoustic power over time, and sets flags or markers to distinguish noise from speech without analyzing pitch, using techniques like Fast Fourier Transform and time-smoothed measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speech processing systems use traditional noise suppression methods (reducing transmission power, changing protocols, or using shielding), then speech quality may be improved, but hardware complexity and cost increase
Solution Approach 1:
The patent replaces physical hardware solutions (shielding, transmission power control) with a software-based signal processing system. The system uses digital signal processing to analyze frequency spectra and identify periodic interference patterns, substituting mechanical/electrical hardware modifications with algorithmic noise suppression that requires no additional physical components.
Solution Approach 2:
The patent introduces an intermediary signal processing layer between the microphone input and the speech processing pipeline. This intermediary system analyzes the frequency domain representation of the signal, identifies periodic interference components, and selectively suppresses them before further processing, acting as a mediator that separates desired speech from unwanted periodic noise without requiring hardware changes.
2Object-affected harmful factors
If systems use shielding or hardware modifications to block interference, then noise suppression improves, but implementation difficulty and cost increase
Solution Approach 1:
The patent replaces physical shielding and hardware modifications with a software-based detection and suppression system. Instead of modifying the physical environment or hardware components to block interference, the system uses digital signal processing to identify and eliminate periodic noise components from the captured audio signal, dramatically simplifying implementation.
Solution Approach 2:
The system performs self-service by automatically detecting and suppressing periodic interference without requiring external hardware modifications or complex manufacturing processes. The software algorithm autonomously analyzes the input signal, identifies periodic patterns, and applies suppression, making the system easy to manufacture and deploy in various devices without specialized hardware production.
3Measurement precision
If systems analyze pitch to identify harmonic signals, then interference detection accuracy improves, but processing time and computational power increase
Solution Approach 1:
The patent extracts the essential characteristic of periodic interference (periodicity in the frequency spectrum) without requiring full pitch analysis. By taking out only the critical feature - the periodic pattern in spectral magnitude across frequency bins - the system achieves accurate interference detection while avoiding the computational overhead of complete pitch analysis algorithms.
Solution Approach 2:
The system performs partial action by analyzing only the magnitude spectrum for periodic patterns rather than conducting full pitch analysis. This partial approach focuses computational resources on detecting the presence and characteristics of periodic interference sufficient for suppression, without the excessive action of complete pitch determination, thereby reducing processing time while maintaining detection accuracy.
4Measurement precision
If systems process the full frequency spectrum to identify interference, then detection accuracy improves, but computational load increases
Solution Approach 1:
The patent applies partial action by processing only the magnitude spectrum rather than the full complex frequency spectrum. This approach extracts sufficient information for periodic interference detection (periodic patterns in magnitude) while discarding unnecessary computational processing of phase information, thereby reducing computational load while maintaining detection accuracy.
Solution Approach 2:
The system extracts only the essential component needed for interference detection - the periodic pattern in spectral magnitude - from the full frequency spectrum. By taking out this critical feature and ignoring other aspects of the spectrum that are not necessary for periodic noise identification, the system achieves accurate detection with reduced computational energy consumption.
Data Source
AI summary
A system improves speech detection or processing by identifying registration signals. The system encodes a limited frequency band by varying the amplitude of a pulse width modulated signal between predefined values. The signal is separated into frequency bins that identify amplitude and phase. The registration signal is measured by comparing a difference in average acoustic power in a plurality of adjacent bins over time.