Speech Analyzer Pitch Detection via Autocorrelation Waveform

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing techniques fail to accurately detect the fundamental frequency of voice signals due to fluctuations and irregular harmonic tones, making efficient pitch frequency detection challenging, especially in hoarse or trembling voices.

Innovation Solution

A speech analyzer is developed with a voice acquisition unit, frequency conversion unit, autocorrelation unit, and pitch detection unit that calculates the pitch frequency by shifting the frequency spectrum, interpolating discrete data, and performing regression analysis on the autocorrelation waveform to exclude noise and formant effects, allowing for accurate pitch frequency estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional fundamental frequency detection methods are used, then the detection process is simple, but the detection accuracy deteriorates due to voice fluctuations and irregular harmonic tones

Engineering Contradiction:

Improvepitch frequency detection accuracyVSAvoiddetection process complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an autocorrelation waveform as an intermediary representation. Instead of directly analyzing the complex voice signal, the system converts the voice signal to an autocorrelation waveform where pitch information is enhanced and noise is suppressed. This intermediary transformation enables accurate pitch detection even from hoarse or trembling voices that would be difficult to analyze directly.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces direct frequency analysis with autocorrelation analysis. Rather than using conventional spectral analysis methods that directly examine frequency components, the system uses autocorrelation to transform the signal into a domain where periodicity is enhanced. This substitution of analysis method enables accurate pitch detection from signals that are difficult to analyze using traditional mechanical frequency detection approaches.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Productivity

If the frequency spectrum is shifted discretely to calculate autocorrelation, then the calculation speed improves, but the frequency resolution deteriorates

Engineering Contradiction:

Improvecalculation speedVSAvoidfrequency resolution

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent performs preliminary discrete autocorrelation calculation to obtain discrete autocorrelation values at specific frequency shifts. This preliminary action enables fast computation while maintaining the necessary frequency information. The discrete values obtained are then used in subsequent analysis to determine pitch frequency, combining computational efficiency with adequate resolution for voice analysis.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent changes the parameter of frequency shift from continuous to discrete values for the autocorrelation calculation. By calculating autocorrelation at discrete frequency intervals rather than continuously, the system achieves faster computation speed. The discrete frequency points are sufficiently dense to maintain adequate resolution for pitch detection in voice signals, balancing speed and precision requirements.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If formant components are included in the autocorrelation waveform, then the waveform captures comprehensive voice characteristics, but the pitch detection accuracy deteriorates due to formant interference

Engineering Contradiction:

Improvepitch frequency detection accuracyVSAvoidvoice characteristic information

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent extracts and removes formant components from the autocorrelation waveform. By identifying and eliminating these interfering components, the system isolates the pitch-related information in the waveform. This extraction process enables accurate pitch frequency detection by removing the distorting influence of formants while preserving the essential pitch characteristics of the voice signal.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent converts the presence of formant components, which initially interfere with pitch detection, into a beneficial process. By analyzing the autocorrelation waveform and identifying formant patterns, the system uses these formant components as reference information to enhance pitch detection. The formants, rather than being merely obstacles, become useful markers that help identify and isolate the pitch frequency more accurately.

Inventive Principle:

Principle #22Blessing in disguise (Convert harm into benefit)

Data Source

PatentUS8738370B2Speech analyzer detecting pitch frequency, speech analyzing method, and speech analyzing program

Publication Date: 2014.05.27 AGI INC

AI summary

A speech analyzer includes a speech acquiring section, a frequency converting section, an autocorrelation section, and a pitch detection section. The frequency converting section converts the speech signal acquired by the speech acquiring section into a frequency spectrum. The autocorrelation section determines an autocorrelation waveform by shifting the frequency spectrum along the frequency axis. The pitch detection section determines the pitch frequency from the distance between two local crests or troughs of the autocorrelation waveform.