Anglia Ruskin Research Online (ARRO)
Browse

File(s) not publicly available

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

conference contribution
posted on 2023-07-26, 15:32 authored by Satvik Venkatesh, David Moffat, Alexis Kirke, Gozel Shakeri, Stephen Brewster, Jörg C. Fachner, Helen Odell-Miller, Alexander J. Street, Nicolas Farina, Sube Banerjee, Eduardo R. Miranda
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.

History

Page range

636-640

ISSN

2379-190X

Publisher

IEEE

ISBN

978-1-7281-7605-5

Conference proceeding

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Name of event

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

Location

Toronto, Canada

Event start date

2021-06-06

Event finish date

2021-06-11

Language

  • other

Legacy posted date

2021-09-28

Legacy Faculty/School/Department

Faculty of Arts, Humanities & Social Sciences

Usage metrics

    ARU Outputs

    Categories

    No categories selected

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC