Adaptive Post-Filter for Speech Signal Periodicity Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional adaptive post-filtering techniques for speech signals face challenges in accurately controlling long-term post-filtering, leading to undesired distortion due to fluctuations in local periodicity measurements and parameter changes at segment boundaries.
Innovation Solution
The method involves measuring smoothed periodicity of audio signal segments through low-pass filtering and interpolating filter parameters between adjacent segments to minimize distortion, using a block diagram system with a filter controller and adaptive post-filter to adjust long-term post-filtering based on smoothed periodicity thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If long-term post-filtering is applied to increase periodicity of speech signal, then perceptual quality of periodic speech is improved, but distortion is introduced when applied to non-periodic signal segments
Solution Approach 1:
The patent dynamically changes the filter parameter γ based on the measured periodicity of the speech signal. When periodicity exceeds a threshold, γ is set to zero to disable long-term post-filtering, preventing distortion in non-periodic segments. When periodicity is below the threshold, γ is adjusted to enhance periodicity improvement, thereby adapting the filter behavior to match signal characteristics and avoiding harmful effects.
2Adaptability or versatility
If filter parameters are adapted on a segment-by-segment basis to fit local characteristics, then adaptability to local signal features is improved, but distortion is introduced at segment boundaries due to parameter changes
Solution Approach 1:
The patent measures the periodicity of the speech signal in advance before applying long-term post-filtering. By evaluating the periodicity characteristic beforehand, the system can determine the appropriate γ value for each segment, ensuring that filtering is only applied when beneficial and parameter transitions are managed to minimize boundary distortion.
3Object-generated harmful factors
If γ parameter is set to zero when periodicity measure exceeds threshold to avoid increasing periodicity of non-periodic signals, then distortion in non-periodic segments is reduced, but fluctuations in periodicity measure still cause undesirable results
Solution Approach 1:
The patent implements a feedback mechanism where the periodicity of the speech signal is continuously measured and used to adjust the γ parameter. This closed-loop control ensures that long-term post-filtering is activated or deactivated based on real-time signal characteristics, maintaining consistent filtering performance despite fluctuations in periodicity measurements.
Data Source
AI summary
A method for adaptive long-term filtering of an audio signal, such as a decoded speech signal. The method includes measuring a smoothed periodicity of an audio signal segment, such as an audio frame, wherein the smoothed periodicity is measured by low-pass filtering an instantaneous periodicity of the audio signal segment. The periodicity of the audio signal segment is then increased in a manner that depends upon whether the smoothed periodicity is less than a predetermined threshold. By utilizing a smoothed periodicity measurement in this fashion, more accurate control of the post-filter is provided as compared to conventional solutions. Additionally, the method includes deriving filters by interpolating between filter responses of adjacent audio signal segments to minimize distortion at segment boundaries.