activity
20232026
most citedScaling strategies for on-device low-complexity source separation with Conv-Tasnet

2 citations · 5 across the 15 of their papers we have counts for

collaborators
Showing eess.ASShow all

6 papers · 1 filter

eess.AS2025

Scaling and Enhancing LLM-based AVSR: A Sparse Mixture of Projectors Approach

Umberto Cappellazzo, Minsu Kim, Stavros Petridis +2

Audio-Visual Speech Recognition (AVSR) enhances robustness in noisy environments by integrating visual cues. While recent advances integrate Large Language Models (LLMs) into AVSR,…

eess.AS2024

Efficient Fine-tuning of Audio Spectrogram Transformers via Soft Mixture of Adapters

Umberto Cappellazzo, Daniele Falavigna, Alessio Brutti

Mixture of Experts (MoE) architectures have recently started burgeoning due to their ability to scale model's capacity while maintaining the computational cost affordable. Furtherm…

eess.AS2023

Parameter-Efficient Transfer Learning of Audio Spectrogram Transformers

Umberto Cappellazzo, Daniele Falavigna, Alessio Brutti +1

Parameter-efficient transfer learning (PETL) methods have emerged as a solid alternative to the standard full fine-tuning approach. They only train a few extra parameters for each…

eess.AS2023★ 1 cited

Continual Contrastive Spoken Language Understanding

Umberto Cappellazzo, Enrico Fini, Muqiao Yang +3

Recently, neural networks have shown impressive progress across diverse fields, with speech processing being no exception. However, recent breakthroughs in this area require extens…

eess.AS2023★ 1 cited

Training dynamic models using early exits for automatic speech recognition on resource-constrained devices

George August Wright, Umberto Cappellazzo, Salah Zaiem +5

The ability to dynamically adjust the computational load of neural models during inference is crucial for on-device processing scenarios characterised by limited and time-varying c…

eess.AS2023

Sequence-Level Knowledge Distillation for Class-Incremental End-to-End Spoken Language Understanding

Umberto Cappellazzo, Muqiao Yang, Daniele Falavigna +1

The ability to learn new concepts sequentially is a major weakness for modern neural networks, which hinders their use in non-stationary environments. Their propensity to fit the c…