Guardado en:
Detalles Bibliográficos
Autores principales: Zhu, Jian, Samir, Farhan, Chodroff, Eleanor, Mortensen, David R.
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2505.23170
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908384184238080
author Zhu, Jian
Samir, Farhan
Chodroff, Eleanor
Mortensen, David R.
author_facet Zhu, Jian
Samir, Farhan
Chodroff, Eleanor
Mortensen, David R.
contents We present ZIPA, a family of efficient speech models that advances the state-of-the-art performance of crosslinguistic phone recognition. We first curated IPAPack++, a large-scale multilingual speech corpus with 17,132 hours of normalized phone transcriptions and a novel evaluation set capturing unseen languages and sociophonetic variation. With the large-scale training data, ZIPA, including transducer (ZIPA-T) and CTC-based (ZIPA-CR) variants, leverage the efficient Zipformer backbones and outperform existing phone recognition systems with much fewer parameters. Further scaling via noisy student training on 11,000 hours of pseudo-labeled multilingual data yields further improvement. While ZIPA achieves strong performance on benchmarks, error analysis reveals persistent limitations in modeling sociophonetic diversity, underscoring challenges for future research.
format Preprint
id arxiv_https___arxiv_org_abs_2505_23170
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ZIPA: A family of efficient models for multilingual phone recognition
Zhu, Jian
Samir, Farhan
Chodroff, Eleanor
Mortensen, David R.
Computation and Language
Sound
Audio and Speech Processing
We present ZIPA, a family of efficient speech models that advances the state-of-the-art performance of crosslinguistic phone recognition. We first curated IPAPack++, a large-scale multilingual speech corpus with 17,132 hours of normalized phone transcriptions and a novel evaluation set capturing unseen languages and sociophonetic variation. With the large-scale training data, ZIPA, including transducer (ZIPA-T) and CTC-based (ZIPA-CR) variants, leverage the efficient Zipformer backbones and outperform existing phone recognition systems with much fewer parameters. Further scaling via noisy student training on 11,000 hours of pseudo-labeled multilingual data yields further improvement. While ZIPA achieves strong performance on benchmarks, error analysis reveals persistent limitations in modeling sociophonetic diversity, underscoring challenges for future research.
title ZIPA: A family of efficient models for multilingual phone recognition
topic Computation and Language
Sound
Audio and Speech Processing
url https://arxiv.org/abs/2505.23170