A dictionary learning and source recovery based approach to classify diverse audio sources

2015/10/27 by K V Vijay Girish, Girish, K V Vijay, T. Ananthapadmanabha +3
Computer Science · #Blind Source Separation Techniques #FOS: Computer and information sciences #H.5.1 #Music and Audio Processing #Sound (cs.SD) #Speech and Audio Processing

paper · pdf · doi:10.48550/arxiv.1510.07774

openalex publication_date 2015/10/27 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

A dictionary learning based audio source classification algorithm is proposed to classify a sample audio signal as one amongst a finite set of different audio sources. Cosine similarity measure is used to select the atoms during dictionary learning. Based on three objective measures proposed, namely, signal to distortion ratio (SDR), the number of non-zero weights and the sum of weights, a frame-wise source classification accuracy of 98.2% is obtained for twelve different sources. Cent percent accuracy has been obtained using moving SDR accumulated over six successive frames for ten of the audio sources tested, while the two other sources require accumulation of 10 and 14 frames.

Citations

• Atomic Decomposition by Basis Pursuit
• Subjective comparison and evaluation of speech enhancement algorithms
• rm K-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse Representation

Related

• Learning Sparse Adversarial Dictionaries For Multi-Class Audio Classification
• An efficient supervised dictionary learning method for audio signal recognition
• Source separation with weakly labelled data: An approach to computational auditory scene analysis
• Data-driven audio recognition: a supervised dictionary approach
• An End-to-End Audio Classification System based on Raw Waveforms and Mix-Training Strategy