Home Creators Posts Import Register Favorites Logout

Poll

Emily voice, redo it with newer tech?

  • Yeah, it's totally worth it, especially with this specific example of voice 31
  • Yeah totally, but not this specific example of voice 0
  • No, I'm satisfied with the current voice and its emotional range 4
  • 2026-10-19
  • 35 votes
{'title': 'Emily voice, redo it with newer tech?', 'choices': [{'text': "Yeah, it's totally worth it, especially with this specific example of voice", 'votes': 31}, {'text': 'Yeah totally, but not this specific example of voice', 'votes': 0}, {'text': "No, I'm satisfied with the current voice and its emotional range", 'votes': 4}], 'closes_at': None, 'created_at': datetime.datetime(2026, 10, 19, 10, 33, 41, tzinfo=datetime.timezone.utc), 'description': None, 'allow_multiple': False, 'total_votes': 35}

Content

Hello everyone,We're going through multiple evolution regarding AI, and today, compared to when Emily has started, there have been large improvements regarding voice generation.Here is a quick sample:

Current one for comparison (best sample out of 30 generations):

It's generally easier to handle as well, as it's on spot quicker.It would take probably 2 weeks to redo everything with the new voice, but it could also be a long term quality improvement.So my question is this:Is it worth it to you? Would it be an important enough improvement to justify a smaller update next month, in november?(this is unrelated to the october's version that is almost done and ready)PS: this is something that had already happened with Zoe, she had an older, much more robotic AI voice, and it went through an overhaul to use a more recent technology

Comments

Drew

OMG, that new one gave me a tiny tent in my pants!

Hyde

100%, likewise you should fix Zoe's voice lines too. Not sure if you're using QWEN-3 but if you haven't you should look into that too. I'd honestly script out fixing all the voice lines.

Femdom Wife Game

Yeah if it works mostly as-is, I'll probably take some time to fix Zoe's voice too. The problem is not the scripting itself, it's already automated, but with the current model, I have to review every single line because there is a 80% failure rate where the voice is not using the correct ton or just bugs entirely when it involves exclamations, silences or "ahh" sounds. But so far, the new model has a near 0% failure rate, so if I don't have to spend time to review individually every voice line, it can be done. (I use ElevenLabs' model)