Original article is here (paywalled 😑). Not sure how far to take these results with such a tiny N but interesting if it can be replicated with a larger sample size. It looks like they initially hypothesized that HRT had caused the changes but ended up rejecting that idea in favor of it being the voice training.