Pitch Frequency Detection Using Learned Fundamental Sound Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing speech processing methods struggle to accurately estimate pitch frequency, particularly in cases with small harmonic sound components or noise sections, leading to erroneous detection and missed pitch frequencies.
Innovation Solution
A speech processing method that calculates a spectrum from input frames, determines speech-like frames, learns fundamental sound characteristics, and detects pitch frequency based on learned values, using techniques like Fourier transforms and Hanning windows to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional correlation-based pitch detection is used, then the method is simple to implement, but the detection accuracy deteriorates in cases with small harmonic sound components or noise sections
Solution Approach 1:
The patent segments the pitch detection process into multiple stages: initial pitch detection from the input spectrum, identification of harmonic sounds at integer multiples of the initial pitch, and iterative refinement by comparing detected harmonic patterns against expected harmonic structures. This segmentation allows the system to handle cases with small harmonic components by building detection confidence progressively through multiple validation steps.
Solution Approach 2:
The patent introduces an intermediary learning value that represents the magnitude of the fundamental sound, which mediates between the raw spectrum data and the final pitch detection decision. This learning value is updated iteratively based on the detected harmonic structure, serving as a confidence indicator that resolves ambiguities in noisy or low-harmonic conditions by comparing against learned expectations from previous frames.
2Duration of action of stationary object
If the system processes all frames including noise sections, then the processing is continuous, but erroneous pitch frequency detections increase
Solution Approach 1:
The patent performs preliminary pitch detection and harmonic identification before making final detection decisions. By pre-identifying the initial pitch from the spectrum and预先 checking for the presence of harmonic sounds at integer multiples, the system prepares validation criteria in advance that filter out erroneous detections from noise sections while maintaining continuous processing of all frames.
Solution Approach 2:
The patent implements feedback through iterative learning where the learning value representing fundamental sound magnitude is continuously updated based on detected harmonic structures. This feedback mechanism allows the system to adapt to the speech characteristics over time and distinguish true pitch frequencies from noise by comparing current detections against learned patterns from previous frames, thereby improving reliability while maintaining continuity.
3Measurement precision
If the system uses learned values from previous frames, then the detection accuracy improves, but the system complexity increases
Solution Approach 1:
The patent makes the learning value mechanism universal by applying the same learning and comparison process to every frame of speech input. The single learning value representing fundamental sound magnitude serves multiple functions: it acts as a reference for detecting the current pitch, a criterion for validating harmonic detection, and a continuous adaptation mechanism. This multi-functionality achieves improved accuracy without proportionally increasing system complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of pitch frequency estimation by distinguishing speech frames from noise and small harmonic components, reducing erroneous detections and improving overall precision.
Implementation Method 1
The frequency conversion unit 10 is a processing unit that calculates the frequency spectrum of the input speech by Fourier transformation of the input speech.
Implementation Method 2
The correlation calculation unit 11 is a processing unit that calculates a correlation value between cosine waves of various frequencies and an input spectrum for each frequency.
Data Source
AI summary
A speech processing method for estimating a pitch frequency includes: specifying, for each determination result of a speech-like-frame, a fundamental sound by using a plurality of local maximum values included in a spectrum of a respective frame determined as the speech-like-frame; obtaining a learned value by performing learning processing on a magnitude of the fundamental sound specified from each determination result of the speech-like-frame, the learned value including an average value and a variance of the magnitude of the fundamental sound specified from each determination result of the speech-like-frame; and executing a detection process by using the learned value, the detection process including detecting a pitch frequency of the respective frame determined as the speech-like-frame by using a threshold, the threshold being obtained by subtracting the variance included in the learned value from the average value included in the learned value.