Time-Warped Audio Transform Coding With Common Frame Warp
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio coding techniques face inefficiencies in encoding signals with pitch variations, leading to audible discontinuities and increased bit rate demands due to uncontrollable duration and bit-rate overhead in time-warped signals.
Innovation Solution
A method that estimates a common time warp for consecutive audio frames, allowing for efficient block transform coding by integrating time-warp operations with window functions and resampling, reducing audible discontinuities and bit-rate overhead by transmitting warp parameters instead of pitch information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If simple time warping is applied to achieve constant pitch over voiced segments, then pitch uniformity is improved, but duration control deteriorates and bandwidth limitations are violated
Solution Approach 1:
The patent changes the parameter representation from absolute pitch values to relative warp factors. By encoding the ratio between actual pitch and reference pitch (warp factor w = log2(p/p0)), the system achieves pitch uniformity while maintaining duration control, as the warp factors are applied locally to individual frames rather than globally to the entire signal.
Solution Approach 2:
The patent segments the audio signal into individual frames and applies time warping independently to each frame using warp factors. This segmentation approach allows duration control to be maintained at the frame level while achieving pitch uniformity within each segment, avoiding the global duration violations of simple time warping.
2Measurement precision
If absolute tape speed information is transmitted for time-warped signal reconstruction, then pitch accuracy is improved, but bit-rate overhead increases substantially
Solution Approach 1:
The patent transforms the information representation from absolute pitch values (requiring high precision and generating large bit-rate overhead) to relative warp factors (requiring lower precision and generating minimal bit-rate overhead). The warp factor w = log2(p/p0) captures pitch information in a compressed form that maintains accuracy while reducing the number of bits required for transmission.
Solution Approach 2:
Instead of transmitting pitch information directly (the conventional approach), the patent transmits the inverse or relative information - the warp factor representing the ratio between actual and reference pitch. This inversion allows the decoder to reconstruct accurate pitch by applying the warp factor to a reference pitch, significantly reducing bit-rate overhead.
3Productivity
If large transform sizes are used for stationary signals, then coding gain is improved, but adaptability to transient events deteriorates
Solution Approach 1:
The patent introduces dynamic adaptability by allowing the transform size to vary across different frames based on the signal characteristics. While maintaining large transform sizes for stationary segments to achieve high coding gain, the system can switch to smaller transform sizes when transient events are detected, thus adapting to changing signal conditions without significantly increasing mean coding cost.
Solution Approach 2:
The patent segments the audio signal into frames and applies different transform sizes to different segments based on their stationarity. This allows large transforms to be applied to stationary segments for high coding gain while using smaller transforms for transient segments, achieving both high productivity for stationary signals and good adaptability for transient events.
Data Source
Figure 1
Figure 2~2B
Figure 3A~3B
AI summary
A spectral representation of an audio signal having consecutive audio frames can be derived more efficiently, when a common time warp is estimated for any two neighbouring frames, such that a following block transform can additionally use the warp information. Thus, window functions required for successful application of an overlap and add procedure during reconstruction can be derived and applied, the window functions already anticipating the re-sampling of the signal due to the time warping. Therefore, the increased efficiency of block-based transform coding of time-warped signals can be used without introducing audible discontinuities.