Salima Mdhaffar

Research focus

Architecture Design (Transformers, SSMs, MoE) (2)Speech & Audio (2)Multimodal Models (2)Inference & Quantization (1)Eval Frameworks & Benchmarks (1)

Frequent co-authors

Tian Su (1)Tianhui Su (1)Tien-Ping Tan (1)T. Tan (1)

Papers (3)

Apr 14, 2026

Tian Su +6Apr 14, 2026·also Le Mans Université, Universiti Sains Malaysia

An Ultra-Low Latency, End-to-End Streaming Speech Synthesis Architecture via Block-Wise Generation and Depth-Wise Codec Decoding

Achieve real-time, high-fidelity speech synthesis with a 48ms latency by directly generating compressed audio codec tokens, bypassing the neural vocoder bottleneck.

Tian Su, Tianhui Su, Tien-Ping Tan +4

Architecture Design (Transformers, SSMs, MoE)Inference & Quantization Speech & Audio

Mar 9, 2026

Avignon UniversitéMar 9, 2026

Using Multimodal and Language-Agnostic Sentence Embeddings for Abstractive Summarization

Hallucination in abstractive summarization? Injecting named entities into the decoder input, along with multimodal embeddings, can keep your French BART model grounded.

Chaimae Chellaf, Salima Mdhaffar, Yannick Estève +1

Eval Frameworks & Benchmarks Multimodal Models Natural Language Processing

Maryem Bouziane +2Mar 9, 2026

Learning Multiple Utterance-Level Attribute Representations with a Unified Speech Encoder

Now a single speech foundation model can generate diverse utterance-level representations, like semantics and speaker identity, opening new possibilities for multimodal and multilingual applications.

Maryem Bouziane, Salima Mdhaffar, Yannick Esteve

Architecture Design (Transformers, SSMs, MoE)Multimodal Models Speech & Audio

Search

Salima Mdhaffar

Research focus

Frequent co-authors

Papers (3)