We present AURA-ST, a three-stage modular pipeline for low-resource speech-to-text translation submitted to the IWSLT 2026 African-Celtic Track 1. The architecture bypasses traditional cross-attention between audio and text modalities by treating projected acoustic representations as a native token prefix to a frozen large language model. A dual-stream encoder captures linguistic and paralinguistic features via a jointly trained semantic and a paralinguistic encoder. A convolutional subsampler then bridges the modality gap through a 4x temporal compression and a linear projection into the LLM embedding space. Finally, a MLP-targeted Low-Rank Adaptation adapter fine-tunes the frozen Gemma-4-E2B backbone for translation without catastrophic forgetting of base language model knowledge. We further identify and resolve the incompatibility between standard PEFT attention-level adapter injection and the Gemma-4 Per-Layer Embedding architecture that tends to cause gradient isolation. Trained on the IWSLT 2026 Track 1 data covering Hausa, Igbo, and Yoruba, the final system achieves a best proxy teacher-forced SacreBLEU of 91.29 on the validation set at Phase 3, with Phase 1 speech encoder validation loss converging to 0.651.

This paper describes the techniques designed for detecting, extracting and normalizing adverse events from social data as part of the submission for the Shared task, Task 1-SMM4H’22. We present an adaptive learner mechanism for the foundation model to identify Adverse Drug Event (ADE) tweets. For the detected ADE tweets, a pipeline consisting of a pre-trained question-answering model followed by a fuzzy matching algorithm was leveraged for the span extraction and normalization tasks. The proposed method performed well at detecting ADE tweets, scoring an above-average F1 of 0.567 and 0.172 overlapping F1 for ADE normalization. The model’s performance for the ADE extraction task was lower, with an overlapping F1 of 0.435.