Voice Encoding Pitch Pulse Search Range Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional speech codecs for VoIP face quality degradation due to frame erasure, especially in the speech onset portion, where inaccurate pitch period detection and lack of distinct glottal pulses hinder effective concealment processing.
Innovation Solution
A speech encoding apparatus that uses pitch pulse information for erasure concealment, employing a determining section to set a search range for pitch pulse position, a selecting section to choose candidates, and an error minimizing section to find the optimal pitch pulse position by generating an adaptive codebook component, thereby minimizing decoding errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the pitch period is not accurate (e.g., double pitch period or half pitch period), then the conventional technique cannot detect the correct glottal pulse position, but increasing search range may increase computational complexity
Solution Approach 1:
The patent performs preliminary actions by determining a search range based on the current frame's pitch period before actually searching for the pitch pulse position in the previous frame. This preliminary range determination guides the subsequent search process, ensuring that potential pulse positions are identified within the correct temporal window without requiring an exhaustive search of the entire previous frame, thus balancing accuracy with computational efficiency.
Solution Approach 2:
The patent employs feedback mechanisms where the pitch period detected in the current frame is fed back to determine the search range for the previous frame's pitch pulse position. This feedback loop ensures that the search range adapts to the actual pitch characteristics, allowing the system to maintain accurate pulse detection while adjusting the search complexity based on the detected pitch period length.
2Measurement precision
If there is no distinct glottal pulse in the excitation signal, then the pulse position of the highest amplitude may not be optimum, but expanding search range increases computational complexity
Solution Approach 1:
The patent applies local quality by focusing the search for pitch pulse positions on specific local regions within the excitation signal rather than treating the entire signal uniformly. By determining a search range based on the current frame's pitch period and concentrating candidates within this localized temporal window, the system identifies optimal pulse positions in regions most likely to contain glottal pulses, improving detection accuracy without requiring a comprehensive search of the entire signal.
Solution Approach 2:
The patent segments the excitation signal into relevant temporal segments based on the pitch period information from the current frame. This segmentation allows the system to divide the search space into manageable portions, evaluating candidate pulse positions within specific segments rather than the entire signal, thereby reducing computational complexity while maintaining detection precision in the most critical regions.
3Reliability
If redundant information is transmitted for erasure concealment, then quality degradation is reduced, but transmission bandwidth increases
Solution Approach 1:
The patent performs preliminary action by encoding and transmitting pitch pulse position information from the previous frame in advance, before any erasure concealment is needed. This preliminary transmission of redundant pitch pulse data enables the decoding side to perform effective concealment processing when frame erasures occur, improving reliability without requiring additional real-time data transmission during the erasure event itself.
Solution Approach 2:
The patent implements copying by replicating pitch pulse position information from the previous frame and transmitting it as redundant data alongside the current frame's encoded information. This copied information serves as a backup that can be used for erasure concealment, enhancing the system's robustness against frame losses while adding only a small amount of redundant data to the transmission stream.
Data Source
AI summary
Provided is an audio encoding device which can detect an optimal pitch pulse when using pitch pulse information as redundant information. The device includes: a search start decision unit (121) which decides the oldest point among a plurality of points where a pitch pulse may exist; a pitch pulse candidate selection unit (122) which defines a search range as a range between the search start point and the point preceding the point of the head of the current frame by one and selects a decoding sound source vector having a large amplitude in this search range as a pitch pulse position candidate; a selector switch (125) which successively switches a plurality of pitch pulse position candidates inputted from a pitch pulse candidate selection unit (122) for output to a pulse sequence generation unit (123) and an error minimization unit (124); a pulse sequence generation unit (123) which generates as a pulse sequence, a vector generated as an adaptive codebook component from the pitch pulse in the current frame when a pitch pulse is set to be a pitch pulse position candidate inputted from the selector switch (125).