Update: Thanks @pitermach showing a great demo that it's actually Mist World Upsampling to 48 in this demo, not NVDA downsampling to 16!
I stitched together an audio file showing you how bad it is at ignoring the setting of -1 as the output. Instead #NVDASR tries to be too smart, enumerate the list and gather which you have set as your sound mapper output, and explicitly call that sound device when passing to the TTS outputs.
I updated this to add a little more at the end and show how Mist World treats audio output switching properly, that I now know is not proper.
Good night, Mastodon. This really ruined my weekend at first, until that amazing demo in my mentions by @pitermach clarified things. :)
Update: People are asking, "how can I tell?" Listen for the sharpness of S's and other consonants. If you have the ear you'll notice.

For anyone curious what TrueVoice ran through an AI model might sound like when speaking, I had it read a passage from an article. It's... Rather strange, that's for sure. I'm a little unsettled just by how hybrid it became.

Here's an AI Cover of the Jason Mraz Living in the Moment song by the Centigram TrueVoice engine - just as to my expectations, feeding lots of text with high inflection did improve the pitch range of this one significantly, so it's usable for a wider range of songs. Try it at:
https://www.jammable.com/centigram-truevoice-tts-gqE74

Another Formant TTS that didn't turn out bad as far as AI singing is Orpheus. However, its pitch range is very narrow, and breaks at higher pitches. It still didn't do too horribly with this Teenager in Love oldies tune, though. Sharing it for the amusement factor mostly. The voice can be used at https://www.jammable.com/orpheus-tts-english-IBHpp

I was bad, and made a Jammable voice for the Amstrad Speech Synthesizer Voice. Barely had enough audio, but it didn't come out too bad. Attached is a cover of it doing Harry Chapin - Cat's In The Cradle.
If you wish to try or use the voice, here's the link:
https://www.jammable.com/amstrad-speech-synthesizer-lYqeF

Definitely not as impressed with it as I was with Google LM. I think the voices don't sound as vibrant since they're using ElevenLabs "quick clone" and even if you do one of the magic voice options, not all of them come out great. (Mac OS Alex though somehow came out exactly on-point.) The hosts sound bored, horrible to me. This is why I threw it together in my lunch break, 30 minutes, fed it the Mist World Wiki and a few sources, edited the chapter outlines a bit to make sure it's not adding bad info, and threw it together. I would not be offended if people turn it off after the first minute, truly. Horrible. What would this thing do with API docs of speech synthesizers or info about old retro tech? Not great.

Was very curious how Good Suno V4 is, so I had it write me a song about Monolog95 with lyrics I highly custom-crafted. I'm attaching it here for people's amusement, maybe it'll give you a laugh too.

Fighting on Mist World. The usual. What else is new, the same old dramas get re-hashed, the same conversations happen again and again as new players join it. And then we have this masterpiece of an AI song about Admin 6, created by one of the players (Ginger) and for anyone not playing the game, hearing it should help them to know why you might want to wait with such horrible staff support.

Something fun with NVDA remote! Listen to this nice beat it can make.

There you have it. A recording of my personal voice driving VoiceOver speech in iOS 18, and a sneak-peak at some of the tutorial in the process. enjoy.