ILENIA_VOZ ASR System Fusion for Albayzin 2024 Speech to Text Challenge
This paper presents the ILENIA VOZ team’s Automatic Speech Recognition (ASR) system developed for the Albayzin 2024 Speech-to-Text (S2T) Challenge, integrating efforts from four language-specific initiatives: AINA (Catalan), Nos (Galician), GAITU (Basque), and VIVES (Valencian). Our primary system is a word-level ROVER fusion of multiple ASR models, achieving a 15.46% and 19.91% Word Error Rate on the RTVE 2022 and 2024 test sets, respectively. Additionally, three contrastive systems are presented. We use datasets from RTVE and various other corpora for model training and fine-tuning. The paper details our system architecture, the fusion techniques that we used, and presents results on the Albayzin-RTVE 2024 and 2022 test sets.
Authors
- Ibon Saratxaga (ORCID: https://orcid.org/0000-0001-7282-2765)
- Marlene Garcia (ORCID: https://orcid.org/0000-0001-9539-4548)
- Carlos Mena
- Christoforos Souganidis (ORCID: https://orcid.org/0000-0002-5013-1442)
- Inma Hernáez (ORCID: https://orcid.org/0000-0003-4447-7575)
- Carmen Magariños (ORCID: https://orcid.org/0000-0003-3525-1304)
- A. Sánchez (ORCID: https://orcid.org/0000-0001-6271-7339)
- Eva Navas (ORCID: https://orcid.org/0000-0003-3804-4984)
- Abir Messaoudi
- Javier González Corbelle (ORCID: https://orcid.org/0000-0002-9457-0279)
- Federico Costa (ORCID: https://orcid.org/0000-0002-1389-3595)
- Asier Herranz
- Carme Armentano-Oller
- Marc Casals-Salvador
- Sarah Solito
- Lucas Takanori Sanchez Shiromizu
Publication Details
- Published
- 2024-11-05
- DOI
- https://doi.org/10.21437/iberspeech.2024-70
- Primary Topic
- Speech Recognition and Synthesis
- Type
- article
- Field-Weighted Citation Impact
- 0.00