This is a long shot but, do you think it is at all possible that they've had some kind of a breakthrough using their quantum processing work, and are using that to accelerate machine learning models and produce much better quality generative speech? After hearing about Bristlecone I can't rule it out.