Hi,

I'm

Picture of the author
hmed

Adel

Attia

Attia

Deep Learning And Speech PhD Researcher

About

me

Picture of profile

ABOUT

Welcome to my website!
I am a Deep Learning engineer and PhD researcher with expertise in Natural Language Processing, Signal Processing, Statistics, and Unsupervised Learning.My primary focus is on developing advanced Deep Learning models for speech and audio processing, with a passion for exploring the limitless possibilities of this cutting-edge technology.

    Education

  • Ph.D. in Computer Engineering

    University of Maryland
    2020 ‐ Current

  • B.SC in electronics and communication engineering

    Alexandria University, Faculty of Engineering
    2015 ‐ 2020

Skills

Below are some of my skills, and I'm always looking to learn more.

Programming Languages

Python, Matlab, C, C++, C#, VHDL, Verilog

Deep Learning

Transformers, GANs, Autoencoders, Unsupervised learning, FairAI

Deep Learning Frameworks

Tensorflow, Pytorch, Scikit learn

Signal Processing

Audio Processing, Speech Processing, Computer Vision and Image Processing

Game Development

Unity Game Engine, VR Development

Others

Linux,Bash,Latex
Picture of the skill
Picture of the skill
Picture of the skill
Picture of the skill
Picture of the author
Picture of the author
Picture of the author
Picture of the author
Picture of the author
Picture of the author
Picture of the author
Picture of the author

Papers

Kid-Whisper: Towards Bridging the Performance Gap in Automatic Speech Recognition for Children VS. Adults
Ahmed Adel Attia, Jing Liu, Wei Ai, Dorottya Demszky, Carol Espy-Wilson, ICASSP 2024 (Under Review)
Improving Speech Inversion Through Self-Supervised Embeddings and Enhanced Tract Variables
Ahmed Adel Attia, Yashish M. Siriwardena, Carol Espy-Wilson. ICASSP 2024 (Under Review)
Masked Autoencoders Are Articulatory Learners
Ahmed Adel Attia, Carol Espy-Wilson. ICASSP 2023
Audio Data Augmentation for Acoustic to articulatory Speech Inversion using Bidirectional Gated RNNs
Yashish M. Siriwardena, Ahmed Adel Attia, Ganesh Sivaraman, Carol Espy-Wilson. EUSIPCO 2023
Download

Experience

Picture of the author


Jul 2019 ‐ Sep 2019
Deep Learning Research Intern
University Of Arizona
Arizona, USA
  • Conducted research on Generative Adversarial Networks (GANs) under the supervision of Prof. Ravi Tandon.
  • I worked on and helped with different projects on GANs.
  • I developed Mutual Information Neural Estimators that achieved over 98% accuracy, and unsupervised Outlier Detection systems.
Picture of the author


Jun 2021 ‐ Jan 2022
Deep Learning Consultant
Omnispeech, LLC
US - Remote
  • I worked on developing lightweight real‐time speech enhancement deep learning models.
  • I successfully scaled down large Speech Enhancement GAN models from 37 million parameters to less than 1 Million parameters, maintaining good clarity and noise cancellation.
  • I also developed an efficient data pipeline using TensorFlow Dataset API and TensorFlow profiler for more than a terabyte of audio data achieving optimal performance and ∼ 100% GPU utilization.
Picture of the author


2022 ‐ Present
Graduate Research Assistant
University Of Maryland
Maryland,USA
  • Conducting research to develop Deep Learning and Machine Learning algorithms for acoustic and articulatory speech data to better understand speech production.
  • Published a number of papers in top conferences in speech and signal processing and machine learning.
  • Picture of the author
    2022 ‐ Present
    Graduate Research Assistant
    University Of Maryland
    Maryland,USA
    • Conducting research to develop Deep Learning and Machine Learning algorithms for acoustic and articulatory speech data to better understand speech production.
    • Published a number of papers in top conferences in speech and signal processing and machine learning.
  • Picture of the author
    June 2021 ‐ January 2022
    Deep Learning Consultant
    Omnispeech, LLC
    Us ‐ Remote
    • I worked on developing lightweight real‐time speech enhancement deep learning models.
    • I successfully scaled down large Speech Enhancement GAN models from 37 million parameters to less than 1 Million parameters, maintaining good clarity and noise cancellation.
    • I also developed an efficient data pipeline using TensorFlow Dataset API and TensorFlow profiler for more than a terabyte of audio data achieving optimal performance and ∼ 100% GPU utilization.
  • Picture of the author
    July 2019 ‐ September 2019
    Deep Learning Research Intern
    University Of Arizona
    Arizona, USA
    • Conducted research on Generative Adversarial Networks (GANs) under the supervision of Prof. Ravi Tandon.
    • I worked on and helped with different projects on GANs.
    • I developed Mutual Information Neural Estimators that achieved over 98% accuracy, and unsupervised Outlier Detection systems.

Projects

Below are some selected projects that I have worked on during my industry experience, as well as class and capstone projects. These endeavors highlight my skills and dedication in various professional contexts, showcasing the breadth of my experience and capabilities.

WORK

Efficient Speech Enhancement GANs
Closed Source

Efficient Speech Enhancement GANs

Optimizing Tensorflow dataset pipeline for large audio dataset
Closed Source

Optimizing Tensorflow dataset pipeline for large audio dataset

VR Hostage Rescue Video Game
Open Source

VR Hostage Rescue Video Game

You Only Look Faster
Open Source

You Only Look Faster

Contact

me

ahmadadelattia@gmail.com
(+1) 469-596-4371