The USTC-Ximalaya system for the ICASSP 2022 multi-channel multi-party meeting transcription (M2MeT) challenge
This work addresses speaker diarization for real-world meeting transcription, but it is incremental as it builds on existing TS-VAD methods with data and post-processing enhancements.
The authors tackled the problem of speaker diarization in multi-speaker meetings by improving target-speaker voice activity detection, resulting in a relative reduction of diarization error rates by up to 66.55% on the Eval set and 60.59% on the Test set compared to classical clustering methods.
We propose two improvements to target-speaker voice activity detection (TS-VAD), the core component in our proposed speaker diarization system that was submitted to the 2022 Multi-Channel Multi-Party Meeting Transcription (M2MeT) challenge. These techniques are designed to handle multi-speaker conversations in real-world meeting scenarios with high speaker-overlap ratios and under heavy reverberant and noisy condition. First, for data preparation and augmentation in training TS-VAD models, speech data containing both real meetings and simulated indoor conversations are used. Second, in refining results obtained after TS-VAD based decoding, we perform a series of post-processing steps to improve the VAD results needed to reduce diarization error rates (DERs). Tested on the ALIMEETING corpus, the newly released Mandarin meeting dataset used in M2MeT, we demonstrate that our proposed system can decrease the DER by up to 66.55/60.59% relatively when compared with classical clustering based diarization on the Eval/Test set.