Md Shohidul Islam
Papers
1
Total Citations
3
H-Index
1
About
Md Shohidul Islam is a researcher specializing in speech signal processing and machine learning, with a particular focus on single-channel speech separation. His work addresses the challenging problem of isolating individual speech sources from a single microphone recording, a critical task for applications in hearing aids, telecommunications, and voice-controlled systems. Islam’s most-cited paper, "Dual transform based joint learning single channel speech separation using generative joint dictionary learning" (2022), introduces an innovative approach that combines dual-domain transformations with generative joint dictionary learning to improve separation accuracy. This method leverages the complementary strengths of time-frequency and other transform domains, enabling more robust and efficient extraction of target speech from mixed signals. While his citation count is still growing—with 3 citations for this key work—the novelty of his dual transform framework has laid a foundation for future advancements in the field. Islam’s contributions are particularly notable for their integration of generative models into dictionary learning, offering a fresh perspective on how to handle overlapping speech in noisy environments. As his research continues to evolve, it promises to enhance the performance of real-world speech processing systems.
Research Focus
Key Achievements
Top Papers
- 1