Time Domain Bandwidth Extension for Narrowband Speech
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bandwidth extension methods fail to effectively extend speech signals in noisy environments, as they struggle to accurately model noise, leading to reduced accuracy and poor speech intelligibility.
Innovation Solution
A system that extends the high-frequency spectrum of narrowband audio signals by introducing non-linearity for vowels and using random noise for consonants, differentiating between vowel and consonant features, and employing adaptive filtering to suppress background noise, allowing for flexible shaping of the extended signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If bandwidth extension methods are used to extend speech signals, then speech quality and bandwidth are improved, but accuracy deteriorates in noisy environments due to inability to model noise effects
Solution Approach 1:
The patent segments the speech signal into vowel and consonant portions using voice activity detection, and applies different bandwidth extension methods to each segment. For vowels, it uses non-linear processing to generate harmonics, while for consonants, it uses random noise generation. This segmentation allows the system to maintain accuracy for periodic sounds while handling aperiodic noisy segments differently, thus resolving the contradiction between bandwidth extension accuracy and robustness in noisy environments.
2Measurement precision
If new communication networks are built to support wideband speech transmission, then speech quality is improved, but cost and implementation time increase
Solution Approach 1:
The patent extracts and extends the high-frequency spectrum components from narrowband speech signals using signal processing techniques. By taking out the bandwidth limitation problem from the network infrastructure and solving it at the signal processing level through harmonic generation and spectral extension, the system achieves wideband speech quality without requiring expensive network infrastructure changes.
3Measurement precision
If bandwidth extension is applied to reconstruct speech, then speech quality is improved under ideal conditions, but performance deteriorates when noise is present due to difficulty in modeling noise effects
Solution Approach 1:
The patent applies different processing qualities to different parts of the speech signal based on local characteristics. For vowel segments with periodic structure, it applies deterministic harmonic generation with high fidelity. For consonant segments and noisy portions, it uses stochastic noise generation with controlled spectral characteristics. This local differentiation allows the system to maintain high reconstruction accuracy where applicable while being robust to noise where deterministic modeling would fail.
Data Source
AI summary
A system extends the high-frequency spectrum of a narrowband audio signal in the time domain. The system extends the harmonics of vowels by introducing a non linearity in a narrow band signal. Extended consonants are generated by a random-noise generator. The system differentiates the vowels from the consonants by exploiting predetermined features of a speech signal.