Master 2018 2019
Stages de la spécialité SAR
Automatic drum loop transcription using deep learning


Site : Le Sound
Lieu : Toulouse
Encadrant : Chunghsin Yeh
Dates :du 01/03/2018 au 31/07/2018
Rémunération :554€
Mots-clés : Parcours ATIAM : Acoustique, Parcours ATIAM : Informatique musicale, Parcours ATIAM : Traitement du signal

Description

Le Sound is developing next-generation interactive audio tools for sound design and post-production in the creative industry including film, broadcast and video games. We propose innovative solutions based on cutting-edge procedural audio research developed in-house and in partnership with internationally recognized research institutes and laboratories.

The internship is about applying deep learning methods to improve automatic transcription of drum loops. The intern will be working within Le Sound team of Novelab in Toulouse and fulfil the following tasks :
- establish a semi-automatic database generation toolchain
- improve Le Sound’s drum transcription algorithm by deep learning
- evaluate the proposed method

The candidate is expected to
- have a general knowledge of deep learning algorithms
- have experience in using Tensorflow
- demonstrate good Python programming skills

Please send your application to jobs@audiogaming.net

Bibliographie

Richard Vogl,Matthias Dorfer,and Peter Knees,“Drum transcription from polyphonic music with recurrent neural networks,” Proceedings of the 17th International Society for Music Information Retrieval Conference (ISMIR), 2016.

Carl Southall, Ryan Stables, and Hockman Jason, “Automatic drum transcription for polyphonic recordings using soft attention mechanisms and convolutional neural networks,” Proceedings of the 18th International Society for Music Information Retrieval Conference (ISMIR), 2017.

C. Jacques, A. Roebel : “Automatic drum transcription with convolutional neural networks,” Proceedings of the Digital Audio Effects (DAFx-2018).