NLP seminar in Kääriku with TartuNLP and Tilde.
Program
| Time | Speaker | Title |
| 10:00 | Tanel Alumäe |
Using Pretrained Models for Speaker Identification |
| 10:30 | Joonas Kalda |
PixIT: Joint Training of Speaker Diarization and Speech Separation from Real-world Multi-speaker Recordings |
| 11:00 | Taido Purason |
LLMs in low-resource settings |
| 11:30 | Liisa Rätsep |
Text-to-speech in low-resource settings |
| 12:00 | Daniil Rõbnikov | Enhancing the Quality of Multilingual Zero-Shot Text-to-Speech Systems |
| 12:30 |
Lunch + swimming |
|
| 14:00 | Mark Fishel |
Interlingua revisited yet again |
| 14:30 | Andra Velve and Tiia Sildam |
Comparing Cascaded and End-to-End Approaches for Estonian Spoken Language Translation |
| 15:00 | Aleksei Dorkin |
A survey on adapting LLMs to new languages |
| 15:30 | Kairit Sirts |
Using GPT-4 for text annotation: experiences from annotating psychological phenomena |
| 16:00 | Pause | |
| 16:30 | Aivo Olev | Co-Designing a Speech Transcription User Interface: A Case Study |
| 17:00 | Tilde |
Latest language technology developments at Tilde |
| 18:00 | Free discussion | |
| 18:30 | Dinner |