Music Content Summary Generation for Similar Audio Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional similar music search methods are inefficient due to low processing speed, classification errors, and incorrect retrieval of music based on mood and genre, as they rely on entire music files and complex feature extraction processes.
Innovation Solution
A method and apparatus that automatically extracts audio features from the compression zone of music files to generate summaries, classify mood and genre, and retrieve similar music by measuring similarity between query music and files with matching mood and genre, using MDCT-based timbre and tempo features, Bayesian Information Criterion clustering, and Euclidean distance calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire music file is used for mood classification and similar music search, then the search comprehensiveness is improved, but the processing speed deteriorates
Solution Approach 1:
The music file is divided into multiple short segments, and only representative segments are selected and processed for feature extraction. This segmentation approach maintains search comprehensiveness by covering different parts of the music while significantly reducing the total processing volume and improving speed.
Solution Approach 2:
The patent extracts and processes only the essential feature information from music segments rather than analyzing the entire music file. By taking out only the necessary audio features (MFCC, LPC, ZCR) from selected segments, the system achieves both comprehensive search coverage and efficient processing.
2Measurement precision
If segment clustering is performed by using short distance between segments or HMM to measure similarity, then the classification accuracy is improved, but the computation complexity and processing time deteriorate
Solution Approach 1:
The patent uses simple distance-based similarity measurement instead of complex HMM models. This approach achieves acceptable classification accuracy with significantly reduced computation complexity, treating the complex HMM approach as an unnecessary overhead that can be discarded for simpler alternatives.
3Measurement precision
If timbre, tempo, and intensity are extracted from the non-compression zone, then the feature extraction accuracy is improved, but the processing speed deteriorates due to required decoding
Solution Approach 1:
The patent extracts audio features directly from the compression zone of the music file without requiring full decoding to the non-compression zone. By taking out and processing features from the compressed representation, the system maintains feature extraction accuracy while dramatically improving processing speed by eliminating the time-consuming decoding step.
4Ease of operation
If mood classification is performed regardless of genre, then the classification simplicity is improved, but the retrieval accuracy deteriorates
Solution Approach 1:
The patent segments the classification process into two independent stages: first genre classification, then mood classification within each genre. This segmented approach maintains simplicity by keeping each stage focused while improving overall accuracy by considering both genre and mood characteristics.
Solution Approach 2:
The patent adds genre as an additional dimension to the classification system. Instead of single-dimensional mood classification, the system now operates in a two-dimensional space of genre and mood, which improves retrieval accuracy by providing more nuanced music categorization while maintaining operational simplicity through structured processing.
Data Source
AI summary
A similar music search method, including: extracting an audio feature value from a compression zone of a music file and generating a music content summary using the extracted audio feature; classifying a mood and a genre of the music file by using the generated music content summary; extracting a similarity feature for a similarity retrieval from the music content summary; storing an information on the classified mood and a genre and the extracted similarity feature information in a database; receiving a query music; detecting the mood and the genre of the query music; measuring a similarity between the query music and music files having a mood and a genre identical to the query music by referring to the database; and retrieving the similar music with respect to the query music based on the measured similarity.