Artificially Synthesising Data for Audio Classification and Segmentation to Improve Speech and Music Detection in Radio Broadcast

Venkatesh, Satvik and Moffat, David and Kirke, Alexis and Shakeri, Gozel and Brewster, Stephen and Fachner, Jörg C. and Odell-Miller, Helen and Street, Alexander J. and Farina, Nicolas and Banerjee, Sube and Miranda, Eduardo R. (2021) Artificially Synthesising Data for Audio Classification and Segmentation to Improve Speech and Music Detection in Radio Broadcast. In: 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Toronto, ON, Canada.

Full text not available from this repository.
Official URL: https://doi.org/10.1109/ICASSP39728.2021.9413597

Abstract

Segmenting audio into homogeneous sections such as music and speech helps us understand the content of audio. It is useful as a pre-processing step to index, store, and modify audio recordings, radio broadcasts and TV programmes. Deep learning models for segmentation are generally trained on copyrighted material, which cannot be shared. Annotating these datasets is time-consuming and expensive and therefore, it significantly slows down research progress. In this study, we present a novel procedure that artificially synthesises data that resembles radio signals. We replicate the workflow of a radio DJ in mixing audio and investigate parameters like fade curves and audio ducking. We trained a Convolutional Recurrent Neural Network (CRNN) on this synthesised data and outperformed state-of-the-art algorithms for music-speech detection. This paper demonstrates the data synthesis procedure as a highly effective technique to generate large datasets to train deep neural networks for audio segmentation.

Item Type: Conference or Workshop Item (Paper)
Keywords: Audio Segmentation, Audio Classification, Music-speech Detection, Training Set Synthesis, Deep Learning
Faculty: Faculty of Arts, Humanities & Social Sciences
Depositing User: Lisa Blanshard
Date Deposited: 28 Sep 2021 15:45
Last Modified: 24 Nov 2021 13:23
URI: https://arro.anglia.ac.uk/id/eprint/706984

Actions (login required)

Edit Item Edit Item