User avatar
🇨🇦Samuel Proulx🇨🇦 @fastfinge@interfree.ca
1y
I can't wait for how excited everyone's going to get when NVDA 2025.1 breaks eloquence and a bunch of other third party TTS addons by retiring WinMM and making Wasapi the only output method. There's going to be a lot of gnashing of teeth. The fix for threshhold is a one-line change, but I have no idea about ibmtts.
4
1
0
0
User avatar
James Scholes @jscholes@dragonscave.space
1y
@fastfinge And then shortly after, the switch to 64-bit! Woo-hoo!
1
0
0
0
User avatar
🇨🇦Samuel Proulx🇨🇦 @fastfinge@interfree.ca
1y
@jscholes I’m sure something horrible with TCP/IP and an eloquence server in another process will get done.
1
0
0
0
User avatar
James Scholes @jscholes@dragonscave.space
1y
@fastfinge Potentially negating the WASAPI work in the process, unless the 32-bit process sends the audio data back to be output by NVDA.
1
0
0
0
User avatar
🇨🇦Samuel Proulx🇨🇦 @fastfinge@interfree.ca
1y
@jscholes I wonder how the Sonata neural voices handle it? They’re doing something with RPC and hosting the AI in an external process.
0
0
0
0