About
I completed my PhD at the University of Sheffield's Centre for Doctoral Training in Speech and Language Technology, where my thesis explored speech separation in noisy and reverberant acoustic environments. My work has been published in top-tier conferences including ICASSP, EUSIPCO, and ASRU, and I have contributed to multiple CHiME workshops.
Research Interests
Speech Enhancement
Deep learning models for noise reduction and speech quality improvement
Speech Separation
Single and multi-channel source separation in complex acoustic environments
Speech Recognition
Robust ASR systems for noisy and multi-speaker scenarios
Multi-Speaker Analysis
Processing and understanding of multi-speaker audio streams
Employment
Spearheading the research and development of modern noise-robust ASR systems.
Led research and development of advanced speech processing solutions for production ASR and TTS systems. Focused on developing novel approaches to speech enhancement, data filtering, and model optimization for real-world deployment scenarios.
Research and development of machine learning models for audio processing applications in consumer electronics, with focus on resource-constrained environments and edge deployment.
Applied research in speech processing and machine learning for healthcare applications, developing novel approaches to multi-speaker audio analysis and transcription systems.
Selected Publications
2025
-
AITW: The Annotated In-the-Wild Dataset for Filtering of In-the-Wild Speech DataInterspeech 2025, Rotterdam • DOI
-
Whilter: A Whisper-based Data Filter for "In-the-Wild" Speech Corpora Using Utterance-level Multi-Task ClassificationInterspeech 2025, Rotterdam • arXiv
2024
-
Speech Separation in Noisy Reverberant Acoustic EnvironmentsPhD Thesis, University of Sheffield • White Rose Repository
-
Transcription-Free Fine-Tuning of Speech Separation Models for Noisy and Reverberant Multi-Speaker Automatic Speech RecognitionInterspeech 2024 • DOI
-
Multi-CMGAN++: Leveraging Multi-Objective Speech Quality Metric Prediction for Speech EnhancementICASSP 2024, pp. 351-355 • DOI
-
Combining Conformer and Dual-Path-Transformer Networks for Single Channel Noisy Reverberant Speech SeparationICASSP 2024, pp. 11491-11495 • DOI
2023
-
On Time Domain Conformer Models for Monaural Speech Separation in Noisy Reverberant Acoustic EnvironmentsASRU 2023, pp. 1-7 • DOI
-
The University of Sheffield CHiME-7 UDASE Challenge Speech Enhancement SystemCHiME-2023 Workshop, pp. 33-38 • PDF
-
On Data Sampling Strategies for Training Neural Network Speech Separation ModelsEUSIPCO 2023, pp. 331-335 • DOI
-
Deformable Temporal Convolutional Networks for Monaural Noisy Reverberant Speech SeparationICASSP 2023, pp. 1-5 • DOI
-
Perceive and Predict: Self-supervised Speech Representation Based Loss Functions for Speech EnhancementICASSP 2023, pp. 1-5 • DOI
2022
-
Att-TasNet: Attending to Encodings in Time-Domain Audio Speech Separation of Noisy, Reverberant Speech MixturesFrontiers in Signal Processing, Vol. 2, Article 856968 • DOI
-
Utterance Weighted Multi-Dilation Temporal Convolutional Networks for Monaural Speech DereverberationIWAENC 2022, pp. 1-5 • DOI
-
Receptive Field Analysis of Temporal Convolutional Networks for Monaural Speech DereverberationEUSIPCO 2022, pp. 80-84 • DOI
Open Source Projects
A Python library for 1D Deformable Convolution in PyTorch, distributed via pip.
A repository of SpeechBrain / PyTorch recipes for published speech separation models.
Education
CDT in Speech and Language Technology
Thesis: Speech Separation in Noisy and Reverberant Acoustics Environments
IET Accredited