Blog(3)

• TFPSNet Time-Frequency Domain Path Scanning Network for Speech Separation

  Deep learning techniques have accomplished a big step forward on speech separation task. The current leading methods are based on the time-domain audio separation network (TasNet) [1]. TasNet uses a learnable encoder and decoder to replace the fixed T-F domain transformation. It takes waveform inputs and directly reconstructs sources, and computes time-domain loss with utterance-level permutation invariant training (uPIT). Several approaches are proposed based on TasNet framework, such as the Conv-TasNet [2] , the dual-path recurrent neural network (DPRNN) [3], the dual-path Transformer network (DPTNet) [4], RNN-free transformer-based neural network (SepFormer) [5] , a self-attentive network with a novel sandglass-shape, namely Sandglasset [6].

• IEEE 802.11bn (Ultra-High Reliability (UHR), Wi-Fi 8)

  Historically, IEEE 802.11 (commonly known as Wi-Fi) standards have prioritized boosting maximum theoretical performance with each iteration—evolving from 2 Mbps (Legacy IEEE 802.11) to 36 Gbps (IEEE 802.11be, Wi-Fi 7)—by implementing advancements such as extended bandwidth, higher-order modulation, and increased spatial streams.

• FedMargin - Federated Learning via Attentive Margin of Semantic Feature Representations

  Let’s play a simple game. Open the photo gallery on your phone and briefly scroll your images, do you see some patterns and recognize the objects you like on the images?

Research Areas(0)

Publications(359)

• Benchmarking Rotary Position Embeddings for Automatic Speech Recognition

  AuthorShucong Zhang, Titouan Parcollet, Rogier van Dalen,

  PublishedIEEE Workshop on Automatic Speech Recognition & Understanding (ASRU)

  Date2025-12-06

• Evaluation of LLMs in Speech is Often Flawed: Test Set Contamination in Large Language Models for Speech Recognition

  AuthorTitouan Parcollet, Rogier van Dalen, Shucong Zhang

  PublishedIEEE Workshop on Automatic Speech Recognition & Understanding (ASRU)

  Date2025-12-06

• Network GDT: GenAI Based Digital Twin for Automated Network Performance Evaluation

  AuthorSukhdeep Singh, Swaraj Kumar, Moonki Hong, Ashish Jain, Madhan Raj Kanagarahinam

  PublishedIEEE International Conference on Communications (ICC)

  Date2025-09-26

View More

News(44)

• [Blog] IEEE 802.11bn (Ultra-High Reliability (UHR), Wi-Fi 8)

  Historically, IEEE 802.11 (commonly known as Wi-Fi) standards have prioritized boosting maximum theoretical performance with each iteration—evolving from 2 Mbps (Legacy IEEE 802.11) to 36 Gbps (IEEE 802.11be, Wi-Fi 7)—by implementing advancements such as extended bandwidth, higher-order modulation, and increased spatial streams.

• Samsung Wins Best Paper Award at IEEE ICC 2025 for Breakthrough in GenAI-Based Digital Twin for Automated Network Evaluation

  Samsung Research (SR) is thrilled to announce that the research paper titled “Network GDT: GenAI based Digital Twin for Automated Network Performance Evaluation” has won the Best Paper Award at the Institute of Electrical and Electronics Engineers International Conference on Communications (IEEE ICC) 2025, one of the most prestigious flagship conferences in the field of telecommunications.

• Driving Technological Excellence: SRI-B Innovator Wins IEEE's Top Technologist Award

  Samsung Research & Development Institute, Bangalore (SRI-B), continues to set benchmarks in innovation and engineering as Dr. Madhan Raj, a distinguished technologist, has been named the ‘Technologist of the Year’ by the prestigious Institute of Electrical and Electronics Engineers (IEEE) India Council.

View More

Others(0)