Language / Accent models - how does DMO learn

Dragon has inbuilt accent detection and we have had great success with all dialects on the AU data packs, albeit some heavily accented users do need to do more correcting in the early days in-order for DMO to learn and improve. The AU data pack may have some words ie: drug names etc specific to ANZ that may not be in other data packs so I would be a bit cautious. As you know, Dragon does need to learn especially for heavily accented users. There are on-line and offline adaptions (learning) and first offline adaptation takes place after 10mins of dictated audio. It then goes into a cycle of I think it’s every 15-20 days. Online learning happens as you continue to correct mistakes. The key is the users need to select and correct when they first start using DMO so that it can learn.