This paper reports the IACAS-Thinkit’s system for the 7th CHiME challenge’s task 1: distant automatic speech transcription and segmentation with multiple recording devices. Our system includes training data augmentation, target speaker voice activity detection (TS-VAD) based speaker diarization (SD), time-domain speakerbeam based single channel target speaker extraction (TSE), guided source separation (GSS) based multi-channel speech separation and WavLM based speech recognition. Evaluated on the CHiME-7 evaluation set, our system for the main track achieves 25.0% macro-average Diarization-attributed Word Error Rate (DA-WER), with an absolute reduction of 30.27% over the baseline system; our system for the far-field acoustic robustness sub-track achieves 20.5% macro-average DA-WER, with an absolute reduction of 13.75% over the baseline system.