Pitch Detection Using Signal and Residual Domain Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pitch detection methods in speech and audio encoding technologies face challenges due to varying levels of complexity and accuracy across different domains, such as time, frequency, signal, and residual domains, leading to inefficiencies and reduced performance.
Innovation Solution
A method and apparatus that perform pitch detection in the signal domain to obtain a candidate pitch, followed by linear prediction to generate an LP residual signal, and then refine the pitch search within a candidate pitch range in the residual domain using auto correlation or energy comparison, ensuring accurate detection across domains.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If pitch detection is performed in a single domain (time, frequency, signal, or residual domain), then the algorithm complexity is reduced, but the pitch detection accuracy deteriorates
Solution Approach 1:
The pitch detection process is segmented into two distinct phases operating in different domains: first, a coarse pitch search is performed in the signal domain to identify a candidate pitch range, and second, a refined pitch search is performed in the residual domain within that candidate range. This segmentation allows each phase to operate optimally in its suitable domain, reducing overall complexity while maintaining high accuracy.
Solution Approach 2:
The invention transitions between different domains (signal domain and residual domain) to perform pitch detection. By changing the domain dimension, the algorithm leverages the advantages of each domain: the signal domain provides a good starting point for coarse search, while the residual domain enables precise refined search, thereby solving the contradiction between complexity and accuracy.
2Device complexity
If pitch detection is performed in the time domain, then the algorithm complexity is low, but the pitch detection accuracy deteriorates
Solution Approach 1:
The algorithm segments the pitch detection task into a coarse search phase (performed in signal domain with lower complexity) and a refined search phase (performed in residual domain with higher precision). This segmentation allows the system to achieve high accuracy without requiring the entire process to be computationally intensive.
Solution Approach 2:
The signal domain pitch detection is performed as a preliminary action to identify a candidate pitch range before conducting the refined search in the residual domain. This preliminary action narrows down the search space, making the subsequent high-precision search more efficient and less complex.
3Measurement precision
If pitch detection is performed in the frequency domain, then the pitch detection accuracy is improved, but the algorithm complexity increases
Solution Approach 1:
The algorithm segments the pitch detection process so that only the refined search phase operates in the residual domain (analogous to frequency domain in terms of transformation), while the coarse search operates in the simpler signal domain. This segmentation reduces the overall computational complexity while maintaining the accuracy benefits of frequency-domain methods.
Solution Approach 2:
Instead of performing full frequency-domain analysis throughout the entire pitch detection process, the invention applies frequency-domain-like residual domain analysis only partially - specifically, only within the candidate pitch range identified by the signal domain search. This partial application maintains accuracy while reducing complexity.
Data Source
AI summary
A pitch detection method and apparatus are disclosed. The method includes: performing pitch detection on an input signal in a signal domain, and obtaining a candidate pitch; performing linear prediction (LP) on the input signal, and obtaining an LP residual signal; setting a candidate pitch range that includes the candidate pitch; searching the candidate pitch range for the LP residual signal, and obtaining a selected pitch.