#stream updates

So I started doing a few streams earlier this year reading books aloud. I had a camera on the pages while I read, I'd comment on stuff, used ReactBot a little, at one point a whole 4 people were there.

I've been going through some bad depression moods, and I still want to stream and share books but I dont think I can talk for that long when my mood makes me go nonverbal. I know part of the mood slump is related to not having basically any audience, and I know I won't gain an audience by only streaming once a month or so. So I need to keep streaming even when I'm feeling nonverbal.

The solution I thought of was using TTS (text-to-speech) on books I have saved to my PC, using something like Calibre to read them in the background. There's free to use open source voices, but I wanted to make my own so it still kinda sounded like me and didn't steal someone's likeness. I don't like the idea of making content using someone else's voice, so I looked into making one with my own.

First I looked at the paid programs with good UIs, but not only are they really fucking expensive, they also have hardcoded length limits even on the paid version. I can't get those to read a full book unless I only go at around 30 minutes of reading a month while paying up to $100 monthly to do it, which isn't really an option. I'm on disability, I can't afford that shit.

So I looked at the open source options bc open source means its free. My problem: I can't code to save my life and my experience in the past with programming forums like GitHub and Reddit has been people becoming condescending or hostile when I explain that I have been actively trying to learn to code for 10+ years and it never works. It's interpreted as me not trying or taking "the easy route" because I can't do even the simple stuff. I wish I could. I gave it an honest shot this time, I spent so much fucking time in my command prompt, typing in installs and running python commands, searching errors online to figure out what was going wrong, editing py files and json files and yaml files, learning how to use Audacity to read the metadata of my recordings and make sure they were the right MHz to use in training--and none of it worked. Even when everything I found on horribly formatted GitHub directories SAID it should have been working, it was just a big non functional pile of errors.

So I've given up on making a TTS clone with my voice bc I can't get these "easy codes" to work and the paid ones don't work for what I want or are ridiculously expensive (and mostly still have limits on usage length that mean they wouldnt work).

If anyone reads this and knows how to get these things working properly for cloning--Coqui, Tortoise, Piper, XTTS, any other open source one--reach out and let me know how to do it, because I really tried. Better yet, if you can prove you're trustworthy and are willing to do it, I would love you forever; I can't offer any payment bc I'm on permanent disability and can barely afford things like rent and food, but I would be eternally grateful and willing to draw you stuff and give you a shoutout if that's worth anything to you.