Transient Handling in Audio Time Stretching via Subband Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for time stretching audio signals, such as phase vocoders and pseudo-quadrature mirror filterbanks, often blur transient events and introduce spurious echoes, affecting the vertical coherence of signals and altering pitch inaudibly during transposition.
Innovation Solution
An apparatus and method that utilize an analysis filterbank to generate subband signals, a time manipulator with an overlap-add stage, a transient detector, and a transient adder to individually handle transients, allowing for selective reduction or exclusion of transients during time stretching, and subsequent precise addition back to the signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If time stretching is performed using phase vocoder or pQMF methods, then the replay speed can be changed while maintaining pitch, but transient events become blurred and spurious echoes are introduced
Solution Approach 1:
The audio signal is divided into multiple subband signals through an analysis filterbank, allowing different processing strategies to be applied to different frequency regions. Transient detection is performed on individual subbands, enabling selective handling of transient events in specific frequency ranges while maintaining overall time stretching functionality.
Solution Approach 2:
Different processing approaches are applied to different parts of the signal based on transient detection results. In subbands where transients are detected, the overlap-add advance value is adjusted or transient blocks are excluded from time stretching, while other subbands undergo normal time stretching processing. This local adaptation preserves transient clarity where needed while maintaining time stretching efficiency elsewhere.
2Manufacturing precision
If stronger time stretching is applied in the environment of transients, then transient blurring is reduced, but pitch changes occur in a spuriously audible manner
Solution Approach 1:
Transient blocks are extracted and excluded from the time stretching process in their respective subbands. By removing transient-containing blocks from the overlap-add processing, the original transient signal is preserved without the dispersion and pitch distortion that would otherwise occur during time stretching of transient regions.
Solution Approach 2:
The transient detector acts as an intermediary that identifies transient blocks and controls the overlap-add stage to exclude these blocks from time stretching. This intermediary mechanism coordinates between the time stretching process and transient preservation, allowing the system to automatically adjust processing based on detected transient events.
3Manufacturing precision
If transients are excluded from time stretching, then dispersion and echo effects are prevented, but the device complexity increases due to transient detection and selective processing
Solution Approach 1:
The processing system dynamically adjusts the overlap-add advance value and block selection based on real-time transient detection results. The transient detector continuously monitors subband signals and provides control signals to the overlap-add stage, enabling the system to adapt its behavior moment-by-moment according to the presence or absence of transient events.
Solution Approach 2:
Transient detection is performed in advance on individual subband signals before the overlap-add processing stage. By identifying transient blocks beforehand, the system can prepare appropriate processing strategies (such as adjusting overlap-add advance values or excluding specific blocks) without adding complexity during the actual time stretching operation.
Data Source
AI summary
An apparatus for processing an audio signal has an overlap-add stage for overlapping and adding blocks of a corresponding one of a plurality of subband signals using an overlap-add-advance value being different from a block extraction advance value. The apparatus further has a transient detector for detecting a transient in the audio signal or a subband signal of the plurality of subband signals. The overlap-add stage is configured for reducing an influence of a detected transient or for not using the detected transients when adding. The apparatus further has a transient adder for adding a detected transient to a subband signal generated by the overlap/add stage. A related method for processing an audio signal has, inter alia, either reducing an influence or discarding a detected transient when overlapping and adding.