Exploring the Potential and Limitations of Model Merging for Multi-Domain Adaptation in ASR
Studies model merging for multi-domain ASR and proposes a more stable merging strategy for European Portuguese speech systems.
Group-Aware Partial Model Merging for Children’s Automatic Speech Recognition
Introduces a parameter-efficient approach that combines clustering, partial fine-tuning, and model merging for children’s ASR.
CAMOES: A Comprehensive Automatic Speech Recognition Benchmark for European Portuguese
Create a new benchmark for European Portuguese ASR.
Children's Voice Privacy: First Steps And Emerging Challenges
First paper on children's voice privacy, exploring technical challenges.
Exploring Shared-Weight Mechanisms in Transformer and Conformer Architectures for Automatic Speech Recognition
Explores different configurations of weight-sharing strategies in transformer and conformer architectures for ASR.
Acoustic and Linguistic Biomarkers for Cognitive Impairment Detection from Speech
Explores combinations of acoustic and linguistic biomarkers for cognitive impairment detection.
Tackling Cognitive Impairment Detection from Speech: A Submission to the PROCESS Challenge
Using knowledge-based versus pre-trained embeddings for pathology detection.
Towards Improved Automatic Speech Recognition for Children
Thesis summary on improving ASR for children.
Improved Children's Automatic Speech Recognition Combining Adapters and Synthetic Data Augmentation
Combines adapters and synthetic data augmentation to improve children's ASR performance.
Introduction to Partial Fine-Tuning: A Comprehensive Evaluation of End-to-End Children's Automatic Speech Recognition Adaptation
Proposes and evaluates a new selective fine-tuning approach for children's ASR.
Multilingual Transfer Learning for Children Automatic Speech Recognition
Multi-task training for multilingual low-resource children's ASR.
Transfer Learning-Based Cough Representations for Automatic Detection of COVID-19
Using pre-trained embeddings for COVID-19 detection.
The INESC-ID Multi-Modal System for the ADReSS 2020 Challenge
Using speech and text pre-trained embeddings for pathological speech detection.
Pathological Speech Detection Using X-Vector Embeddings
Using x-vector embeddings for pathological speech detection across different diseases.
Label-Consistent Sparse Auto-Encoders
Explores auto-encoders with label consistency.