Audio Coding Temporal Noise Shaping Filter Energy Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Temporal Noise Shaping (TNS) in audio coders can introduce artefacts at low bitrates, degrading audio quality, particularly in speech and tonal music signals, due to its limitations in managing prediction gain and impulse response energy.
Innovation Solution
An encoder apparatus with a controller that dynamically adjusts the TNS filtering by selecting between higher and lower impulse response energy filters based on frame metrics, allowing for intermediate filtering statuses to minimize artefacts while maintaining the benefits of TNS, by modifying the first filter to obtain a second filter with reduced impulse response energy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If TNS filtering is applied at low bitrate, then quantization noise is shaped to be masked by the signal, but click-like or noise-like artefacts are introduced that degrade audio quality
Solution Approach 1:
The patent applies dynamics by making the TNS filter energy adaptive rather than fixed. The filter impulse response energy is dynamically adjusted based on the prediction gain of the current frame. When prediction gain is high, full TNS filtering is applied; when prediction gain is low, the filter energy is reduced to avoid artefacts. This dynamic adaptation resolves the contradiction between noise shaping effectiveness and artefact introduction.
Solution Approach 2:
The patent changes the parameter of filter impulse response energy based on prediction gain thresholds. By modifying the energy parameter of the TNS filter according to signal characteristics, the system achieves effective noise masking when appropriate while preventing artefact generation when prediction gain is insufficient. This parameter adaptation directly addresses the technical contradiction.
2Manufacturing precision
If TNS filtering with high impulse response energy is used, then quantization noise masking is improved, but artefacts appear on speech and tonal music signals at low bitrate
Solution Approach 1:
The system dynamically adjusts filter energy based on prediction gain measurements. High energy filtering is applied only when prediction gain exceeds a threshold, ensuring effective noise masking. When prediction gain is below the threshold, energy is reduced to prevent artefacts on speech and tonal signals. This dynamic control resolves the contradiction between masking effectiveness and artefact prevention.
Solution Approach 2:
The filter energy parameter is changed according to prediction gain levels. The system computes prediction gain for each frame and adjusts the TNS filter energy parameter accordingly, using high energy when prediction gain is high and low energy when prediction gain is low. This parameter adaptation resolves the contradiction between noise masking precision and artefact avoidance.
3Object-affected harmful factors
If TNS is disabled to avoid artefacts, then audio quality on problematic frames is preserved, but pre-echo artefacts on signals with sharp attacks are not reduced
Solution Approach 1:
Rather than completely disabling TNS, the system dynamically adjusts filter energy based on prediction gain. This allows TNS to remain active and reduce pre-echo artefacts on signals with sharp attacks (high prediction gain) while reducing filter energy on problematic frames (low prediction gain) to avoid click-like artefacts. This dynamic approach resolves the contradiction between pre-echo reduction and quality preservation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There are discussed methods and apparatus for performing temporal noise shaping. An apparatus may comprise a temporal noise shaping, TNS, tool (11) for performing linear prediction, LP, filtering (S33, S35, S36) on an information signal including a plurality of frames; and a controller (12) configured to control the TNS tool (11) so that the TNS tool (11) performs LP filtering with: a first filter (14a) whose impulse response has a higher energy (S36); and a second filter (15a) whose impulse response has a lower energy (S35) than the first filter, wherein the second filter is not an identity filter, wherein the controller (12) is configured to choose (S34) between filtering (S36) with the first filter (14a), and filtering (S35) with the second filter (15a) on the basis of a frame metrics.