Join our next testing phase for Voxta.

We’ve made some updates based on what we've learned. Here's a quick rundown:

• We've got new guides and videos at https://doc.voxta.ai/
• We made our Virt-A-Mate scene better with more actions, details, and improved animations.
• We’ve included Python. This lets us use things like the faster_whisper package right on your device. This is good for saving money and keeping things private.
• Prompts are now better and will give you good results straight away.
• We made action guessing better, especially for smaller models. There are two new things: silent actions (where the AI can do stuff but not talk) and a way to ask the AI to do things (like looking up stuff online) - but that's not in the beta yet.
• Characters can now use a backup voice, so sharing them should work better for everyone.
• We now work with Deepgram. It's less costly than Azure and better than Vosk. Plus, we’ve added a way to detect silence to help lower the cost of speech-to-text.
• You can now share Tavern V2 character cards with pictures!

There's more to see on our Discord release channels. We're still working hard on the basics but will share our plans soon. There's so much we want to achieve!

Also, check out our new thread where we’ve put together some videos showcasing what Voxta can do. Watch them here: https://hub.virtamate.com/threads/voxta-ai.44874/

Thanks so much to everyone helping us, whether it’s through Patreon, sharing thoughts, or testing our beta. We really appreciate the support and are excited about what we’re creating!

P.S. We really need more testers for Voxta. If you're interested in helping us out, please reach out! Your feedback would be invaluable.