Audio Scene Classification with Deep Recurrent Neural Networks


We introduce in this work an efficient approach for audio scene classification using deep recurrent neural networks. An audio scene is firstly transformed into a sequence of high-level label tree embedding feature vectors. The vector sequence is then divided into multiple subsequences on which a deep GRU-based recurrent neural network is trained for sequence-to-label classification. The global predicted label for the entire sequence is finally obtained via aggregation of subsequence classification outputs. We will show that our approach obtains an F1-score of 97.7% on the LITIS Rouen dataset, which is the largest dataset publicly available for the task. Compared to the best previously reported result on the dataset, our approach is able to reduce the relative classification error by 35.3%.
Original languageEnglish
Title of host publicationProc. 18th Annual Conf. of the Intl. Speech Communication Association (INTERSPEECH)
Number of pages5
Volume 2017-August
Place of PublicationStockholm, Sweden
Publisher International Speech Communication Association (ISCA)
Publication date01.08.2017
Publication statusPublished - 01.08.2017
Event18th Annual Conference of the International Speech Communication Association - Stockholm, Sweden
Duration: 20.08.201724.08.2017
Conference number: 132696


Dive into the research topics of 'Audio Scene Classification with Deep Recurrent Neural Networks'. Together they form a unique fingerprint.

Cite this