Theme: Speech in Context: Advancing Technology for Diverse Human Communication
As a speech technology community in The Netherlands—including researchers, industry partners and users—we must adopt a critical perspective on mainstream speech technology, which often overlooks the diversity in speech, the rich layers of meaning conveyed through spoken communication, and the subtleties embedded within the speech signal. From pathological speech analysis to non-verbal vocalisations, we need to explore how speech technology can better serve diverse communication needs. This direction includes addressing challenges stemming from speaker characteristics, multilingual contexts, and varying situational demands in human-machine interaction.
09:30 - 10:00 Walk-in
10.00 - 10:05 Welcome by Roeland Ordelman (Stichting Open Spraaktechnologie)
10:05 - 11:00 Sociolectal and Idiolectal Finetuning in AI Models for Inclusive Language and Speech Processing (Prof. Dr. Antal van den Bosch, Universiteit Utrecht)
11:00 - 11:30 Coffee break
11:30 - 12:45
The Transformation of Language and Speech Models in Artificial Intelligence Research (Cristian Tejedor Garcia, Radboud University)
Challenges in Speech Technology for Minority Languages in the Netherlands (Wietse de Vries, Groningen University)
Challenges and Recommendations for Dutch Atypical Speech Data Collection, Annotation, Sharing, and Usage (Zhengjun Yue, Delft University of Technology)
Analyzing speech models with tools from psycholinguistics and communication science (Tom Lentz, Tilburg University)
13:00 - 13:45 Lunch
13:45 - 15:20
Listening to a Newspaper? Why and How (Heino Schaght)
Application of Speech Technology in Health Care (Hidde Schwietert)
The MediSpeech project by HealthTalk (Jan-Marc Verlinden)
The HOSAN project: Building Speech Models for All Dutch Voices (Roeland Ordelman)
Speech Recognition at SpraakLab (Marijn Huijbregts)
15:20 - 15:45 Short coffee break
Panel Discussion "Speech in Context: Advancing Technology for Diverse Human Communication"
16:15 - 17:00 Coffee & visit the posters and information market
Posters:
Dragos Balan: Evaluating the State-of-the-Art Automatic Speech Recognition systems for Dutch
Lingyun Gao: Building a Benchmark Evaluation Dataset for Automatic Dutch Child Reading Diagnostics
Thomas Kluiters: An open source model for Dutch ASR
Spyretta Leivaditi: Does Disease-Specific Data Affect Dutch Dysarthric ASR Performance?
Nina van Roij, Alessandra Polimeno, Bo Molenaar, Breixo Soliño Fernández, Emilia Barakova, Aoju Chen: Entraining Social Robots: The effect of pitch entrainment on primary school student engagement in robot-mediated tutoring sessions
Information Market:
NOTaS
Whisp
Juvoly
HOSAN
16:30 - 18:00 Drinks