It’s helpful for tasks like transcribing calls at a call center, where you need to know who (customer or salesperson) said what. Diarization also improves the accuracy of speech recognition for each speaker. This saves on processing power and storage space since it can use mono recordings instead of stereo.