Singing Language Identification using a Deep Phonotactic Approach

TL;DR

This paper introduces a deep phonotactic approach for Singing Language Identification (SLID) that leverages multilingual phoneme recognition and recurrent classification, achieving unprecedented accuracy on polyphonic music datasets.

Contribution

It presents a modernized phonotactic system for SLID using CTC-based phoneme recognition and recurrent language classification, improving performance over previous methods.

Findings

Achieved unprecedented SLID performance on large datasets

Demonstrated effective out-of-set language identification

Validated system on polyphonic music with publicly available data

Abstract

Extensive works have tackled Language Identification (LID) in the speech domain, however their application to the singing voice trails and performances on Singing Language Identification (SLID) can be improved leveraging recent progresses made in other singing related tasks. This work presents a modernized phonotactic system for SLID on polyphonic music: phoneme recognition is performed with a Connectionist Temporal Classification (CTC)-based acoustic model trained with multilingual data, before language classification with a recurrent model based on the phonemes estimation. The full pipeline is trained and evaluated with a large and publicly available dataset, with unprecedented performances. First results of SLID with out-of-set languages are also presented.