TEN-framework/TEN_Turn_Detection

TEN-framework/TEN_Turn_Detection is a 7.6 billion parameter model based on the Qwen2.5-7B transformer architecture, developed by TEN Team. It is specifically designed for intelligent turn detection in full-duplex human-AI dialogue, classifying user utterances into 'finished', 'wait', or 'unfinished' states. This model excels at context-aware turn management and multilingual support for both English and Chinese, enabling more natural conversation flow by reducing awkward interruptions.