LLianJJun

@maozhiqiang Have you tried increasing "batch_max_steps"?

@maozhiqiang I'm not sure. however, I have a breakpoint in the continuous section of the voiced sound component. so, the cause of the problem is suspected to be the receptive...