Signal Processor Pitch Model Harmonic Preservation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech enhancement algorithms struggle to accurately estimate the a priori Signal-to-Noise-Ratio (SNR) in poor SNR environments, leading to destruction of harmonic structure and introduction of unwanted artefacts like musical tones, and are limited in modelling pitch harmonic jitter and non-integer multiples of fundamental frequencies.
Innovation Solution
A signal processor with a modelling block that generates a pitch-model-signal based on a periodic function spanning discrete frequency bins, allowing for flexible representation of pitch frequencies and jitter, and a manipulation block that combines this signal with the frequency-domain-input-signal to produce an enhanced output.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If existing speech enhancement algorithms are used in poor SNR environments, then noise suppression is achieved, but harmonic structure is destroyed and musical tone artefacts are introduced
Solution Approach 1:
The patent changes the parameterization approach by using a periodic function with fundamental frequency and jitter parameters instead of traditional spectral subtraction parameters. This allows adaptive adjustment of pitch-related parameters to preserve harmonics while suppressing noise, resolving the contradiction between noise suppression and harmonic preservation.
Solution Approach 2:
The patent introduces a pitch model signal as an intermediary representation that captures the periodic structure of speech. This intermediary model serves as a bridge between the noisy input signal and the enhanced output, allowing harmonic structure to be preserved through explicit pitch modelling while noise is suppressed.
2Measurement precision
If traditional pitch modelling is used, then integer multiples of fundamental frequency are captured, but pitch harmonic jitter and non-integer multiples are not accurately represented
Solution Approach 1:
The patent makes the pitch model dynamic by incorporating jitter parameters that allow the harmonic frequencies to deviate from exact integer multiples of the fundamental frequency. This dynamic parameterization enables accurate representation of both precise fundamental frequencies and the natural variations (jitter) observed in real speech, resolving the contradiction between measurement precision and adaptability.
3Device complexity
If fixed bandwidth modelling is used, then processing is simplified, but arbitrary bandwidth adaptation is limited
Solution Approach 1:
The patent creates a universal pitch model that can operate across arbitrary bandwidths by parameterizing the periodic function to match the available frequency range. The same core model structure adapts to different bandwidth scenarios (narrowband, wideband, super-wideband) without requiring separate processing paths, thus achieving versatility without proportionally increasing complexity.
Data Source
AI summary
A signal processor comprising: a modelling block, configured to receive a frequency-domain-input-signal, a fundamental-frequency-signal representative of a fundamental frequency of the frequency-domain-input-signal; and configured to provide a pitch-model-signal based on a periodic function, the pitch-model-signal spanning a plurality of discrete frequency bins, each discrete frequency bin having a respective discrete frequency bin index, wherein within each discrete frequency bin the pitch-model-signal is defined by: the periodic function; the fundamental frequency; the frequency-domain-input-signal; and the respective discrete frequency bin index. The signal processor further comprises a manipulation block, configured to provide an output-signal based on the frequency-domain-input-signal and the pitch-model-signal.