fix qwen3-omni audio length < 30s (#12674)

Co-authored-by: canghua <canghua.wjp@alibaba-inc.com>

python/sglang/srt/multimodal/processors/base_processor.py

@@ -245,6 +245,8 @@ class BaseMultimodalProcessor(ABC):
             }:
                 # Note(Xinyuan): for gemma3n, ref: https://github.com/huggingface/transformers/blob/ccf2ca162e33f381e454cdb74bf4b41a51ab976d/src/transformers/models/gemma3n/processing_gemma3n.py#L107
                 kwargs["audio"] = audios
+                kwargs["audio_kwargs"] = {}
+                kwargs["audio_kwargs"].setdefault("truncation", False)
             else:
                 kwargs["audios"] = audios