• Corpus ID: 53539824

Automatic Urban Sound Classification Using Feature Learning Techniques

@inproceedings{Jacoby2014AutomaticUS,
  title={Automatic Urban Sound Classification Using Feature Learning Techniques},
  author={Christopher Jacoby},
  year={2014}
}
Automatic sound classification is a burgeoning field in audio informatics, which has been growing in parallel to developments in machine learning and urban informatics. In this thesis, we define two primary bottlenecks to research in automatic urban sound classification: the lack of an established taxonomy, and the meager supply of annotated real-world data. We begin by assembling a new taxonomy of urban sounds to create a foundation for future work in urban sound classification, which is an… 
2 Citations
Real-Time Noise Classifier on Smartphones
TLDR
A real-time noise classifier based on a smartphone by utilizing only the mel-frequency cepstral coefficient (MFCC) as the feature vector is presented, which drastically reduced the computation complexity and achieved 92.06% accuracy.
Seeing Cities Through Big Data: Research, Methods and Applications in Urban Informatics
This book introduces the latest thinking on the use of Big Data in the context of urban systems, including research and insights on human behavior, urban dynamics, resource use, sustainability and

References

SHOWING 1-10 OF 47 REFERENCES
Support vector machines and self-organizing maps for the recognition of sound events in urban soundscapes
TLDR
A combined approach including unsupervised learning (specifically, Self-Organizing Maps) for clustering and collecting real world samples and supervised learning for labeling is introduced for sound event recognition in urban environments.
Environmental sound recognition using MP-based features
TLDR
A novel method based on matching pursuit to analyze environment sounds for their feature extraction that is flexible, yet intuitive and physically interpretable, and can be used to supplement another well-known audio feature, i.e. MFCC, to yield higher recognition accuracy for environmental sounds.
Content-based Retrieval of Environmental Sounds by Multiresolution Analysis
TLDR
Experimental results show that the approach always outperforms a method based on traditional MFCC features and Euclidean distance, improving retrieval rates from 51% to 62% and the similarity measure is defined in terms of the Kullback-Leibler divergence between GGDs.
Nanyang Technological University Sound Event Recognition and Classification in Unstructured Environments
TLDR
Two novel feature extraction methods are proposed for recognition of environmental sounds in severe noisy conditions, based on the visual signature of the sounds, and one of them has a good performance and is particularly suited to severe mismatched conditions.
A database and challenge for acoustic scene classification and event detection
TLDR
This paper introduces a newly-launched public evaluation challenge dealing with two closely related tasks of the field: acoustic scene classification and event detection.
THE SENSOR CITY INITIATIVE : COGNITIVE SENSORS FOR SOUNDSCAPE TRANSFORMATIONS
TLDR
The authors outline the challenges posed by a cognitive sensor system and how it can approach a more humanlike understanding of soundscape, such as through the detection of meaningful events, deciding what data to record, and in-field sensor placement.
Spectral vs. spectro-temporal features for acoustic event detection
  • Courtenay V. Cotton, D. Ellis
  • Computer Science, Physics
    2011 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA)
  • 2011
TLDR
This work proposes an approach to detecting and modeling acoustic events that directly describes temporal context, using convolutive non-negative matrix factorization (NMF), and discovers a set of spectro-temporal patch bases that best describe the data.
CONTENT-BASED RETRIEVAL FROM UNSTRUCTURED AUDIO DATABASES USING AN ECOLOGICAL ACOUSTICS TAXONOMY
TLDR
A method to search for environmental sounds in unstructured databases with user-submitted material using a Support Vector Machine as a learning algorithm to classify sounds according to a general sound events taxonomy based on ecological acoustics is described.
Learning Features from Music Audio with Deep Belief Networks
TLDR
This work presents a system that can automatically extract relevant features from audio for a given task by using a Deep Belief Network on Discrete Fourier Transforms of the audio to solve the task of genre recognition.
Soundtrack classification by transient events
TLDR
Retrieval results based on transient features alone are comparable to an MFCC-based system, and fusing the two representations achieves a relative improvement of 7.5% in mean average precision (MAP).
...
1
2
3
4
5
...