Noise-Dependent Artificial Bandwidth Expansion for Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial bandwidth expansion (ABE) solutions for speech signals suffer from poor performance in noisy situations, leading to artifacts that degrade the naturalness and intelligibility of speech, particularly due to the limited frequency bandwidth and misclassification of phoneme groups.
Innovation Solution
A noise-dependent artificial bandwidth expansion algorithm that determines signal type and noise conditions to adjust the expansion of narrowband speech signals to wideband, using features like gradient index, SNR estimates, and noise modes to minimize artifacts and maximize intelligibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional narrowband bandwidth (less than 4 kHz) is used for speech transmission, then transmission efficiency and system capacity are improved, but speech quality and intelligibility deteriorate due to missing high frequency components
Solution Approach 1:
The patent applies dynamics by making the bandwidth expansion algorithm adaptive rather than static. The system dynamically adjusts the expansion parameters based on real-time noise conditions (SNR estimates, noise mode classification), allowing the bandwidth expansion to be optimized for each specific acoustic environment. This resolves the contradiction by enabling high-quality wideband speech when conditions permit while maintaining efficient narrowband transmission when noise is present.
Solution Approach 2:
The patent changes the parameter of bandwidth from a fixed narrow value to a variable wide value based on noise conditions. By estimating signal-to-noise ratio and classifying noise modes, the system changes the effective bandwidth parameter dynamically, expanding to wideband (up to 8 kHz) when noise is low and maintaining narrowband when noise is high, thus resolving the trade-off between transmission efficiency and speech quality.
2Device complexity
If fixed artificial bandwidth expansion algorithm is applied without noise dependency, then computational complexity is reduced, but performance in noisy situations deteriorates due to artifacts and misclassification
Solution Approach 1:
The patent transforms the fixed algorithm into a dynamic one that adapts to noise conditions. By continuously estimating SNR and classifying noise modes (stationary, transient, babble), the system dynamically adjusts expansion parameters, preventing misclassification of phoneme groups in noisy conditions and reducing artifacts while maintaining reasonable computational complexity through efficient feature extraction and classification.
Solution Approach 2:
The patent implements feedback by using noise condition estimates (SNR, noise mode) to continuously adjust the bandwidth expansion algorithm's behavior. This closed-loop approach allows the system to learn from the acoustic environment and adapt its parameters accordingly, improving performance in noisy situations without requiring excessive computational resources for complex processing.
3Speed
If bandwidth expansion is applied without considering noise conditions, then processing speed is maintained, but intelligibility deteriorates due to expanded noise and artifacts in high frequency band
Solution Approach 1:
The patent applies dynamics by making the expansion process adaptive to noise conditions. The system quickly estimates noise characteristics (SNR, noise mode) and dynamically adjusts whether to apply full bandwidth expansion or modified expansion, maintaining fast processing while improving intelligibility by preventing noise expansion in high frequency band when conditions require it.
Solution Approach 2:
The patent changes the processing parameters based on noise conditions, adjusting the degree of bandwidth expansion dynamically. When noise is detected, the system modifies parameters to reduce or eliminate high frequency expansion, thereby preventing intelligibility degradation from expanded noise while maintaining fast processing speeds through efficient parameter adjustment rather than complete reprocessing.
Data Source
AI summary
A method, device, system, and computer program product expand narrowband speech signals to wideband speech signals. The method includes determining signal type information from a signal, obtaining characteristics for forming an upper band signal using the determined signal type information, determining signal noise information, using the determined signal noise information to modify the obtained characteristics for forming the upper band signal, and forming the upper band signal using the modified characteristics.