Triplet loss based embeddings for forensic speaker identification in Spanish.
Emmanuel MaquedaJavier Alvarez-JimenezCarlos MenaIván MezaPublished in: CoRR (2021)
Keyphrases
- speaker identification
- language identification
- speech recognition
- gaussian mixture model
- speech signal
- feature extraction
- broadcast news
- noisy environments
- low dimensional
- dimensionality reduction
- em algorithm
- machine learning
- information extraction
- hidden markov models
- high dimensional
- feature space
- multiscale
- information retrieval