Open Model for Slovenian Acoustic Prosody Annotation
Opis
This service provides an open model for acoustic prosody annotation for the Slovenian language. The model supports the advancement of spoken language technologies and enables improved analysis of prosodic features in speech data and communication scenarios. It is based on available speech datasets and audio sources suitable for modelling Slovenian prosody, which are identified and reviewed in the initial phase of service development. Based on this analysis, appropriate development environments and workflows for modelling and dataset preparation are defined. The resulting model supports linguists, phoneticians, and communication experts with language tasks. Developers working in speech technologies and human—computer interaction can use it to improve the naturalness of user interaction in various scenarios. The development and training of the model leverages high-performance computing resources to process large speech datasets and efficiently extract predefined features. The HPC infrastructure supports a deep learning model training and evaluation pipeline. The generated model is available for inference on less demanding platforms, due to lower computational complexity.
Pričakovani rezultati
An open model for Slovenian acoustic prosody annotation, along with evaluation results and supporting documentation. The model is made publicly available through SLAIF and CLARIN.SI infrastructure, enabling its direct use by the speech technology community (e.g., language analysis, first and second language education, preservation of cultural heritage, archives). The acoustic prosody annotation model also enables integration into various applications, supporting enhanced Slovenian spoken interaction.
Kartica storitve
| Kategorija | AI products |
| Domena | Spoken Language Technologies, Human-Computer Interaction |
| Vertikala | Digital Society |
| Ciljni uporabniki | Researchers, Linguists, and Developers in the Field of Spoken Language Technologies and Human-Computer Interaction |
| Tip storitve | Open Model |
| Rezultat | Open model for Slovenian acoustic prosody annotation; evaluation results; supporting documentation. |
| Zahteve | |
| Omejitve | |
| Na voljo od | Q3/2027 |
| Licenca | |
| Slika | |
| Oznaka slike |