Picture for Paris Smaragdis

Paris Smaragdis

Unsupervised Improvement of Audio-Text Cross-Modal Representations

May 05, 2023
Zhepei Wang, Cem Subakan, Krishna Subramani, Junkai Wu, Tiago Tavares, Fabio Ayres, Paris Smaragdis

Figure 1 for Unsupervised Improvement of Audio-Text Cross-Modal Representations

Figure 2 for Unsupervised Improvement of Audio-Text Cross-Modal Representations

Figure 3 for Unsupervised Improvement of Audio-Text Cross-Modal Representations

Figure 4 for Unsupervised Improvement of Audio-Text Cross-Modal Representations

* Submitted to WASPAA 2023

Access Paper or Ask Questions

A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement

Feb 23, 2023
Zhepei Wang, Ritwik Giri, Devansh Shah, Jean-Marc Valin, Michael M. Goodwin, Paris Smaragdis

Figure 1 for A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement

Figure 2 for A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement

Figure 3 for A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement

Figure 4 for A Framework for Unified Real-time Personalized and Non-Personalized Speech Enhancement

* Accepted by ICASSP 2023

Access Paper or Ask Questions

Framewise WaveGAN: High Speed Adversarial Vocoder in Time Domain with Very Low Computational Complexity

Dec 08, 2022
Ahmed Mustafa, Jean-Marc Valin, Jan Büthe, Paris Smaragdis, Mike Goodwin

Figure 1 for Framewise WaveGAN: High Speed Adversarial Vocoder in Time Domain with Very Low Computational Complexity

Figure 2 for Framewise WaveGAN: High Speed Adversarial Vocoder in Time Domain with Very Low Computational Complexity

Figure 3 for Framewise WaveGAN: High Speed Adversarial Vocoder in Time Domain with Very Low Computational Complexity

* Submitted to ICASSP 2023, demo: https://ahmed-fau.github.io/fwgan_demo/

Access Paper or Ask Questions

Optimal Condition Training for Target Source Separation

Nov 11, 2022
Efthymios Tzinis, Gordon Wichern, Paris Smaragdis, Jonathan Le Roux

Figure 1 for Optimal Condition Training for Target Source Separation

Figure 2 for Optimal Condition Training for Target Source Separation

Figure 3 for Optimal Condition Training for Target Source Separation

* Submitted to ICASSP 2023

Access Paper or Ask Questions

Meta-Learning for Adaptive Filters with Higher-Order Frequency Dependencies

Sep 20, 2022
Junkai Wu, Jonah Casebeer, Nicholas J. Bryan, Paris Smaragdis

Figure 1 for Meta-Learning for Adaptive Filters with Higher-Order Frequency Dependencies

Figure 2 for Meta-Learning for Adaptive Filters with Higher-Order Frequency Dependencies

Figure 3 for Meta-Learning for Adaptive Filters with Higher-Order Frequency Dependencies

Figure 4 for Meta-Learning for Adaptive Filters with Higher-Order Frequency Dependencies

* Source code and audio examples: https://jmcasebeer.github.io/metaaf/higher-order

Access Paper or Ask Questions

Learning Representations for New Sound Classes With Continual Self-Supervised Learning

May 15, 2022
Zhepei Wang, Cem Subakan, Xilin Jiang, Junkai Wu, Efthymios Tzinis, Mirco Ravanelli, Paris Smaragdis

Figure 1 for Learning Representations for New Sound Classes With Continual Self-Supervised Learning

Figure 2 for Learning Representations for New Sound Classes With Continual Self-Supervised Learning

Figure 3 for Learning Representations for New Sound Classes With Continual Self-Supervised Learning

Figure 4 for Learning Representations for New Sound Classes With Continual Self-Supervised Learning

* Submitted to IEEE Signal Processing Letters

Access Paper or Ask Questions

Real-Time Packet Loss Concealment With Mixed Generative and Predictive Model

May 11, 2022
Jean-Marc Valin, Ahmed Mustafa, Christopher Montgomery, Timothy B. Terriberry, Michael Klingbeil, Paris Smaragdis, Arvindh Krishnaswamy

Figure 1 for Real-Time Packet Loss Concealment With Mixed Generative and Predictive Model

Figure 2 for Real-Time Packet Loss Concealment With Mixed Generative and Predictive Model

Figure 3 for Real-Time Packet Loss Concealment With Mixed Generative and Predictive Model

Figure 4 for Real-Time Packet Loss Concealment With Mixed Generative and Predictive Model

* Submitted to INTERSPEECH 2022

Access Paper or Ask Questions

Meta-AF: Meta-Learning for Adaptive Filters

Apr 25, 2022
Jonah Casebeer, Nicholas J. Bryan, Paris Smaragdis

Figure 1 for Meta-AF: Meta-Learning for Adaptive Filters

Figure 2 for Meta-AF: Meta-Learning for Adaptive Filters

Figure 3 for Meta-AF: Meta-Learning for Adaptive Filters

Figure 4 for Meta-AF: Meta-Learning for Adaptive Filters

* Source code and audio examples: https://jmcasebeer.github.io/projects/metaaf

Access Paper or Ask Questions