Speech Encoder Spectral Tilt Change Detection for Comfort Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech encoding technologies face challenges in efficiently detecting perceptually significant changes in background noise during inactive frames, leading to audible artifacts and discontinuities in comfort noise generation, as they often rely solely on energy fluctuations and fail to distinguish between different types of noise.
Innovation Solution
A method that generates a sequence of spectral tilt values for inactive frames, calculates changes within this sequence, and decides whether to transmit a silence description based on the calculated change, allowing for more accurate detection of perceptually significant noise changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speech encoders transmit silence descriptors for all inactive frames, then comfort noise generation is continuous, but bandwidth consumption increases and transmission efficiency decreases
Solution Approach 1:
The patent implements discontinuous transmission by periodically transmitting silence descriptors only for inactive frames where spectral tilt changes exceed a threshold, rather than continuously transmitting for all inactive frames. This periodic action maintains comfort noise generation reliability while improving transmission efficiency by reducing the number of transmitted descriptors.
2Device complexity
If speech encoders rely solely on energy fluctuations to detect noise changes, then detection simplicity is maintained, but measurement precision of perceptually significant changes deteriorates
Solution Approach 1:
The patent changes the detection parameter from simple energy fluctuations to spectral tilt values, which provide more precise measurement of perceptually significant noise changes. The encoder calculates spectral tilt for inactive frames and compares changes against a threshold, enabling accurate detection of background noise variations while maintaining reasonable algorithm complexity.
3Productivity
If speech encoders use lower bit rate for inactive frames, then transmission efficiency improves, but noise characterization accuracy deteriorates
Solution Approach 1:
The patent extracts the critical spectral tilt information from inactive frames and transmits only when necessary (when spectral tilt changes exceed the threshold). This selective extraction approach maintains transmission efficiency by reducing the number of low-bit-rate descriptors transmitted, while ensuring adequate noise characterization accuracy when changes occur.
Data Source
Figure 1A~1D
Figure 2~3
Figure 4~6
AI summary
Disclosed configurations include systems, methods, and apparatus arranged to generate a sequence of spectral tilt values that is based on inactive frames of a speech signal. For each of a plurality of inactive frames of the speech signal, a transmit decision is made according to a change calculated among at least two corresponding values of the sequence. The outcome of the transmit decision determines whether a silence description is transmitted for the corresponding inactive frame.