Triplet loss based embeddings for forensic speaker identification in Spanish

TL;DR

This paper investigates the use of triplet loss-trained CNNs to generate speech embeddings for forensic speaker identification in Spanish, addressing a language with limited prior research and exploring embedding quality and likelihood ratio calculations.

Contribution

It introduces a novel approach using triplet loss for speech embeddings in Spanish forensic speaker identification and evaluates different spectrogram configurations.

Findings

Triplet loss-based embeddings outperform traditional methods.

Spectrogram configuration impacts embedding quality.

Proposed likelihood ratio methods improve speaker identification accuracy.

Abstract

With the advent of digital technology, it is more common that committed crimes or legal disputes involve some form of speech recording where the identity of a speaker is questioned [1]. In face of this situation, the field of forensic speaker identification has been looking to shed light on the problem by quantifying how much a speech recording belongs to a particular person in relation to a population. In this work, we explore the use of speech embeddings obtained by training a CNN using the triplet loss. In particular, we focus on the Spanish language which has not been extensively studies. We propose extracting the embeddings from speech spectrograms samples, then explore several configurations of such spectrograms, and finally, quantify the embeddings quality. We also show some limitations of our data setting which is predominantly composed by male speakers. At the end, we propose two approaches to calculate the Likelihood Radio given out speech embeddings and we show that triplet loss is a good alternative to create speech embeddings for forensic speaker identification.