other

SpeechLLM

otheractiveprovisionalspeechllm-34e94de9·1 events·first seen 16d ago

Aliases: SpeechLLM

Co-occurring entities

Phi4-Multimodal Qwen3.5 Omni simultaneous speech-to-text translation DOA (Decoder-Only Attention)

More like this (12)

Audio-LLM Dep-LLM LLM vLLM frontier LLMs whichllm Arabic LLMs LLM (CLI tool)LLM-as-a-Judge LLM Debate Competition LLM inference LLM CLI

Recent events (1)

6arXiv · cs.CL·16d ago·source ↗

DOA: Training-Free Decoder-Only Attention Policy for Long-Form Simultaneous Speech Translation with SpeechLLMs

The paper proposes Decoder-Only Attention (DOA), a training-free streaming policy for simultaneous speech-to-text translation (SimulST) that works with off-the-shelf decoder-only Speech LLMs. DOA derives proxy alignment signals from self-attention rather than cross-attention, enabling long-form simultaneous translation without retraining. Experiments on Phi4-Multimodal and Qwen3-Omni demonstrate low-latency performance approaching offline decoding quality, validating that decoder self-attention contains sufficient alignment information for streaming decisions.

Long Context Evolution Inference Economics Phi4-Multimodal SpeechLLM Qwen3.5 Omni +3 more