Qwen-Audio-Jev: A Unified Interface for Audio Decisions
Large audio language models (LALMs) have made substantial progress in audio understanding. However, autoregressive text generation introduces computational overhead and latency in high-throughput applications that require frequent selection, judgment, and scoring. To address this limitation, we introduce Qwen-Audio-Jev...
Pei-Xin Chen, Zhi-Gao Chen, Zhi-Yun Fan et al.
· 0 citations