Singing voice detection with deep recurrent neural networks

Simon Leglaive (1, 2) , Romain Hennequin (1) , Roland Badeau (2)

1 Audionamix
2 TSI - Département Traitement du Signal et des Images

In this paper, we propose a new method for singing voice detection based on a Bidirectional Long Short-Term Memory (BLSTM) Recurrent Neural Network (RNN). This classifier is able to take into account a past and future temporal context to decide on the presence/absence of singing voice, thus using the inherent sequential aspect of a short-term feature extraction in a piece of music. The BLSTM-RNN contains several hidden layers, so it is able to extract from low-level features a simple representation fitted to our task. The results we obtain significantly outperform state-of-the-art methods on a common database.

Mots clés: Recurrent Neural Networks; Singing Voice Detection; Long Short-Term Memory; Deep Learning

Domaines: Traitement du signal et de l'image [eess.SP]

Simon Leglaive, Romain Hennequin, Roland Badeau. Singing voice detection with deep recurrent neural networks. 40th International Conference on Acoustics, Speech and Signal Processing (ICASSP), Apr 2015, Brisbane, Australia. pp.121-125. ⟨hal-01110035⟩