Noise-Robust environmental sound classification method based on combination of ICA and MP features
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This paper presents an environmental sound classification method that is noise-robust against sounds recorded by mobile devices, and presents evaluation of its performance. This method is specifically designed to recognize higher semantics of context from environmental sound. Conventionally, sound classifications have used acoustic features in the frequency domain extracted from sound data using signal processing techniques. Although the most popular feature is Mel-frequency Cepstral Coefficients (MFCC), MFCC is inappropriate for mixture sound with noise. Independent Component Analysis (ICA) can extract sound characteristics even when the source is corrupted by noise because components within the source are assumed to be independent. In recent years, Matching Pursuit (MP) has been addressed to extract time-domain features. It has been applied to various applications. The feature is effective for recognizing and classifying environmental sounds that include time-variant sound such as birdsongs, alarms, and vehicle sounds. In this way, some innovative techniques have been proposed to recognize and classify environmental sounds recorded on mobile devices. However, we have not yet obtained a decisive method to attain a higher recognition and classification rate against environmental sounds with various noises such as unintended sounds and white noise. To address this problem, we propose a noise-robust classification method using a combination of Independent Component Analysis (ICA) and MP. It is possible to reduce noise effects for feature extraction. From performance evaluations, we confirmed that the proposed method can provide about 8% better classification than that of MFCC feature extraction.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it