This week Nuance announced the imminent release of Dragon for Mac, Version 6. I spoke with Nuance and this new version takes advantage of several improved dictation technologies.

Deep Learning

Nuance has always been able to server-based algorithms to improve dictation accuracy but this new version will be the first time they can embed learning on a user's computer, allowing them to improve their own language and acoustic model. This, and other improvements, adds up to less required training and a reported 24% improvement in accuracy. (I'm looking forward to putting that to the test.)

Improved Transcription

The demonstration I saw showed significant improvement in transcription of existing audio. Not only is the transcription better, it's also much easier to train and operate. I particularly like the new batch mode, that lets you transcribe batches of audio files in one go.

Improved Text Control

Mixing typing and transcription has always been rough going on the Mac. With Dragon 6 for Mac, you'll be able to dictate in supporting apps and type at the same time without the wheels falling off. They are still working on the list of supporting apps for launch but Scrivener is already one of them.

The new version ships (digitally) on September 1 and there will be physical product shipments by mid-September.