SteerablePlex：可引导全双工语音模型与SimIF-Bench

1. 12:00
   arXiv · Audio and Speech

   SteerablePlex：全双工语音模型能被引导吗？

   研究提出 SimIF-Bench，用于评测对话模型是否能在规定场景内按顺序完成多个目标，结果显示当前开源全双工模型难以遵循此类约束。团队进一步提出基于 GDPO 的训练方法，让全双工模型在保持轮次切换能力的同时遵循文本指令，生成 SteerablePlex。连接异步后端语言模型后，其多阶段约束遵循能力优于现有开源模型和 GPT-Realtime。

Current heat 9·Comparable peak 10（Oct 9）·Comparable change over 24 hours –

The trend compares only the same participants observed continuously; its range may be smaller than the current heat count. Move or click on the chart to inspect hourly heat; use the left and right arrow keys to switch.