It's kind of sad that Eloquence now sounds boring to me.

It's kind of sad that Eloquence now sounds boring to me. It's not that I'd say Speechplayer sounds "just as good" because the Espeak prosody is going to bug me until the day I die, but more that Eloquence just feels like "the same old" whereas using the other one makes me think "potentially new." And Jess told me "classic pitch" might remind people more of the intonation the Braille & Speak had, which I can totally see now. Still, I tested it. You can read the same word with Espeak, like "for" and "four" and it'll still inflect the same way. Or like the word "and" and "the". You type them as a single word, word-by-word reading, Espeak and SpeechPlayer both sound shit on it, even with the word final tuning thing, because all that'll do is lengthen it a little, but the stress of how the word is said will forever smell like Espeak.

@Tamasg I have a question. I knoww hy you don't want to change to another ready made phonimizer because it would make a 700 MB forment synth. I get this.

I argue you should do it anyway, as an experiment, so you can triangulate on the best way to move forward. Thoughts?

@nick oh, it's been done as an experiment earlier by @fastfinge, he actually built me an NVDA addon that used Gruut with an early SpeechPlayer copy. It was quite amusing. The driver (NVDA) has gone through a lot of changes since then, so not sure how easy would be to get a version of it working with the latest versions, but yeah, it did sound different on where it stressed vowels and consonants. However, I am thinking I can build an "eloquence prosody" mode into the frontend, maybe just rework Classic pitch to be a "pitch tone type" combobox which has classic Pitch, Eloquence-style, and Espeak. It's clear that when you use classic pitch the single-word reading sounding awkward problem also goes away, so I think the frontend itself can do a lot more to reshape what Espeak gives. Which is an interesting path to look into.

@Tamasg @fastfinge Did anyone record a sample of the Gruut branch?

@nick @fastfinge haha don't think so! I know I've lost it in all this work by now. Hmm. Might turn up sometimee though. Wouldn't be surprised. I did like it since it had a combobox for phonemizer type. We just had to bundle the entire Gruut dependency set within NVDA add-on which made it so uggly.

@Tamasg I have a question. I knoww hy you don't want to change to another ready made phonimizer because it would make a 700 MB forment synth. I get this.

I argue you should do it anyway, as an experiment, so you can triangulate on the best way to move forward. Thoughts?

@Tamasg @nick @fastfinge Wow, sounds very cool, maybe can be made as an optional install from the language pack section for the feature.

@fastfinge @Tamasg @nick thanks for this, pure goal.
Just to get some clarification, I unzipped the file and paste the folder inside the userconfig\addons folder, but can't see the synth name on the list after restart, what I forgot to do?

@fastfinge @Tamasg @nick Ah that's even weirder, when I examine the log for 25.3.2, the SQLight couldn't be loaded and I can understand why now. But when I load it using the portable most recent alfa version of NVDA, the ad on is showing up fine and it's enabled. But the synthesiser is not showing up on the list with control plus CapsLock plus S.

@fastfinge @Tamasg @nick That's interesting, may I know where the synth fetch the driver as in that folder there is no dlls
Later, I copy pasted the x86 and x64 folder from a newer copy of nvspeech. Still it didn't, show on the list, maybe because it's a newer driver. the weird thing is, it's not showing up in log as error even after turning on debug mode.