Deep Learning Techniques for Phoneme Recognition in Italian Children's Speech
Broca, a Conformer-based deep learning system pretrained on 8 days of adult speech and fine-tuned on a 165-minute dataset of Italian child speech collected through a range of standardized diagnostic tests, demonstrates that accurate, vocabulary-independent speech-to-IPA transcription can be achieved with minimal data.
N. Barbaro, Cristina Gena, F. Petriglia et al.
· 0 citations