Very Short Pitch Lag Detection for Speech and Music Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech coding technologies struggle with efficiently encoding very short pitch lags in music and singing voice signals, leading to perceptual distortion due to incorrect pitch lag transmission, as they are shorter than the conventional minimum pitch limitation.
Innovation Solution
Implement a method for very short pitch detection and coding using a combination of time domain and frequency domain techniques, including pitch correlation and low frequency energy detection, to accurately encode pitch lags within a predefined range smaller than the conventional minimum pitch limitation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional pitch detection methods are used, then the coding process is simple, but very short pitch lags cannot be detected accurately leading to perceptual distortion
Solution Approach 1:
The pitch detection process is segmented into two distinct phases: time domain analysis using autocorrelation for initial pitch lag estimation, and frequency domain analysis using spectral peaks detection for verification and refinement. This segmentation allows each method to handle specific aspects of pitch detection, improving overall accuracy for very short pitch lags while managing complexity through modular processing
Solution Approach 2:
The invention transitions from single-domain (time domain only) pitch detection to multi-domain analysis by incorporating both time domain autocorrelation and frequency domain spectral analysis. This dimensional expansion enables the system to detect very short pitch lags that are invisible or ambiguous in a single domain, resolving the contradiction between accuracy and complexity
2Reliability
If the pitch lag range is extended to include very short pitch lags, then music and singing voice coding quality improves, but the risk of incorrect pitch lag transmission increases
Solution Approach 1:
The system implements feedback mechanisms where frequency domain spectral peak information is used to verify and correct time domain pitch detection results. This cross-validation feedback loop ensures that very short pitch lags are accurately identified and transmitted, preventing information loss while maintaining coding quality for music and singing voice signals
Solution Approach 2:
The invention performs preliminary frequency domain analysis to identify spectral peaks before final pitch lag determination. This preliminary action in the frequency domain provides advance information about expected pitch frequencies, enabling more reliable pitch lag coding and transmission for very short pitch periods
Data Source
AI summary
A method includes detecting whether there is a very short pitch lag in a speech or audio signal that is shorter than a conventional minimum pitch limitation using a combination of time domain and frequency domain pitch detection techniques. The pitch detection techniques include using pitch correlations in a time domain and detecting a lack of low frequency energy in the speech or audio signal in a frequency domain. The detected very short pitch lag is coded using a pitch range from a predetermined minimum very short pitch limitation that is smaller than the conventional minimum pitch limitation.