Keyword Detection Using Pre-computed Audio Frame Probabilities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current keyword detection technologies face issues with repeated computation and high computational complexity when using a sliding window approach for audio analysis.
Innovation Solution
A method and apparatus for keyword detection that determine the probability of an audio frame corresponding to a character unit in a preset keyword, and then calculate the probability of consecutive audio frames forming the keyword, to efficiently identify voice clips of the keyword.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sliding window approach is used for keyword detection, then detection coverage is improved, but computational complexity increases
Solution Approach 1:
The patent pre-calculates and stores the probability that each audio frame corresponds to each character unit before keyword detection. This preliminary action allows the detection process to directly retrieve pre-computed probabilities rather than recalculating them for each keyword search, significantly reducing computational complexity while maintaining comprehensive detection coverage across all possible keywords
Solution Approach 2:
The patent creates a probability matrix that copies and stores the correspondence relationships between audio frames and character units in advance. This copied probability information is then reused for detecting multiple different keywords without requiring repeated computation, thereby reducing the computational burden while ensuring thorough detection
2Reliability
If sliding window approach is used for keyword detection, then detection coverage is improved, but calculation overhead increases
Solution Approach 1:
The system performs preliminary calculation of frame-to-character-unit probabilities before the actual keyword detection process. By completing this computationally intensive task in advance, the system avoids redundant calculations during keyword detection, thereby reducing calculation overhead and energy consumption while maintaining comprehensive detection coverage
Solution Approach 2:
The probability matrix serves as a self-service resource that the detection system can repeatedly utilize without requiring external computation for each keyword. The pre-computed probabilities automatically serve multiple detection tasks, reducing the need for repeated heavy computations and lowering overall calculation overhead
3Measurement precision
If probability calculation method is used, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent pre-computes the probability that each audio frame corresponds to each character unit and stores it in a matrix. This preliminary computation enables accurate probability-based detection for multiple keywords without repeating the complex calculations, thus improving detection accuracy while reducing the computational burden during actual keyword detection
Solution Approach 2:
The system creates a copied representation of frame-character correspondence probabilities that can be efficiently queried during detection. This copied probability information allows the system to perform accurate detections by simple lookups rather than complex recalculations, improving accuracy while managing computational complexity
Data Source
AI summary
A keyword detection method and apparatus, and an electronic device and a storage medium. The method comprises: for a target audio clip in a target audio, determining a first probability that a target audio frame in the target audio clip corresponds to a target character unit, the target character unit being a character unit comprised in a preset keyword, and the position of the target audio frame in the target audio clip corresponding to the position of the target character unit in the preset keyword (110); determining, according to the first probability, a second probability that the target audio clip corresponds to the preset keyword, the second probability indicating the probability that audio frames in the target audio clip are sequentially character units in the preset keyword (120); and determining, according to the second probability, whether the target audio clip is a speech segment of the preset keyword (130).