Audio Transform Coding with Direct Long-to-Short Window Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio coding techniques require additional delay to detect transitory signals, such as strong attacks in speech or percussive signals, necessitating the use of transition windows to adapt window sizes, which is not acceptable in certain telecommunications applications.
Innovation Solution
A method for transform coding/decoding of digital audio signals that allows direct transition from a long window to a short window without introducing additional delay, by detecting the particular event within the current frame and applying a short window if the event is detected at the beginning of the frame, or a long window if not detected, thereby eliminating the need for transition windows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If transition windows are used to adapt window sizes for detecting transitory signals, then the detection accuracy of strong attacks is improved, but the coding delay is increased
Solution Approach 1:
The patent applies preliminary action by detecting the presence of a strong attack at the beginning of the current frame before finalizing the window size decision. This allows the encoder to proactively prepare for the potential need to use short windows, eliminating the delay that would otherwise require waiting until the next frame to detect the attack and switch window sizes.
Solution Approach 2:
The patent implements dynamics by making the window size adaptive and changeable within the current frame based on real-time detection of strong attacks. The system dynamically switches between long and short windows depending on the detected signal characteristics, allowing rapid response to transitory signals without the lag introduced by fixed or slowly-adapting window size schemes.
2Measurement precision
If short windows are used for coding current frame when strong attack is detected, then the temporal resolution is improved, but the reconstruction quality may deteriorate
Solution Approach 1:
The patent applies local quality by using different window sizes (short or long) in different temporal regions of the signal based on local characteristics. When a strong attack is detected in a specific region, short windows are applied locally to that region to capture transient details, while long windows can be used in other regions where they provide better frequency resolution and reconstruction quality. This localized adaptation optimizes both temporal resolution and reconstruction quality where needed.
3Reliability
If transition windows are used to pass from long window to short window, then the perfect reconstruction is maintained, but the device complexity is increased
Solution Approach 1:
The patent extracts or removes the transition window component from the coding structure, allowing direct switching between long and short windows without requiring intermediate transition windows. This simplification is achieved by detecting strong attacks at the beginning of frames and directly applying appropriate window sizes, thereby eliminating the additional complexity of implementing and managing transition window sequences while maintaining acceptable reconstruction quality.
Data Source
AI summary
The invention relates to transform coding/decoding of a digital audio signal represented by a succession of frames, using windows of different lengths. For the coding within the meaning of the invention, it is sought to detect (51) a particular event, such as an attack, in a current frame (Ti); and, at least if said particular event is detected at the start of the current frame (53), a short window (54) is directly applied in order to code (56) the current frame (Ti) without applying a transition window. Thus, the coding has a reduced delay in relation to the prior art. In addition, an ad hoc processing is applied during decoding in order to compensate for the direct passage from a long window to a short window during coding.