I’ve had good results with a three-pass routine similar to what you describe: first pass audio only (accept partial understanding), second pass with transcript for repair, third pass audio only again. The third pass is where I usually notice real gains in segmentation and confidence.

Another useful tweak is to cap transcript time so it doesn’t dominate the session. I keep listening as the main task and treat text as targeted support. For learners who want more consistent long-form input, I also rotate in story-based sources between drills; sites like tuotuoshu.com can fit well in that workflow.

Thanks for presenting this so clearly and practically.