Paid Scenes vamX AI Chat - RESPONDS TO TOUCH + GPT4 + NSFW AIs + 100 Personalities + 1000s of Actions - 1.52

Fixed an issue where when reconnecting to the vamX Chat AI servers, AI Chat voice data stopped playing in vamX (she would no longer speak the AI's responses).


Fixed issue where AI Chat initial dialog was spoken twice.


Numerous compatibility fixes for AI Chat custom button actions to work with additional Virt-a-Mate scenes (depending on certain types of Atoms or buttons which were breaking AI Chat action links).

CHANGES / IMPROVEMENTS
Fixed major issue where audio would stop playing if you interrupted the AI (spoke or replied before all audio had been played) or if the audio arrived out of order (shorter clips finishing generation first) or when using ElevenLabs audio.

Fixed Silicon Maid 7B (this was not working, or working only with major delays). This was especially problematic using the default NSFW Random AI, where SiliconMaid can be selected automatically leading to long delays.

Fixed issue where audio would fail to generate if the main voice was an ElevenLabs voice but the Narrator or 2nd Person voices were not ElevenLabs voices.

Extended free access to advanced OpenRouter models (up to 20 hours per month free with your vamX subscription). Choosing OpenRouter in the vamX NSFW Model drop down gives you free access to many top-of-the-line AI models like Google Gemini 2, DeepSeek, or Claude for vamX AI Chat responses.

ElevenLabs.io - Added the ability to choose faster or more human-like ElevenLabs voices. This is especially useful for getting extremely fast, high quality voices from ElevenLabs, as the Turbo 2.5 model is still very high quality, but much faster than ElevenLabs v2/v3.

Automatic Edging - Using vamX Connect with a heart rate monitor, you can now automatically stop or slow toy thrusting when your heart rate rises past a set point.

More stable connection between Virt-a-Mate and the vamX AI Chat servers (less reloads / conversation restarts).

OpenRouter.ai support for amazing chat models including Google Gemini 2.0 Flash, DeepSeek, and MANY others. Many of these models work VERY WELL for NSFW role play. You can use our shared OpenRouter API Key for limited vamX use.

Much faster, more stable connection to Elevenlabs.ai. You can now use your own paid Elevenlabs.ai key to generate realistic, near real time, voices with vamX.

Nomi.ai support for long term relationship building. Nomis have their own personalities and conversation history. When speaking to your Nomi, you will always continue the conversation from where you stopped last time. You will need to create your own account at Nomi.ai to build a relationship with your Nomi in vamX.

Grok 2 + Grok 3 are now provided as NSFW AI Chat Models.

You can now say "asterisk" to create a "*". This can be used to speak something that happened, so the AI model knows it happened, not just that you want it to happen. For example you can say "Asterisk Jenny slaps Jack asterisk ouch." and it will turn into *Jenny slaps Jack* Ouch.

Additional Improvements, Fixes & Changes
Many stability improvements (but please report any ongoing issues you still find). Fixed various possible crashes that would cause a server restart, or cause the voices to stop working.

When a Personality specifies a voice, it now actually changes the drop down to that voice, so you can see the change (if you are using a Verbatik or ElevenLabs voice it doesn't change it).

You can now set Female, Male, and Narrator Voices to "Auto" voice sets without errors (in the past using multiple Auto voice sets conflicted with each other and resulted in using the same voice multiple times).

"Continue as me" now works if you don't have a voice selected for the male/user voice. Fixed issue where "Continue as me" could generate text that would be partially removed, for example it would say "Hi Emily! Let's go." and then "Hi Emily!" would disappear.

By default, you can talk in vamX AI Chat only while holding down a button (mouse or VR controller button).

If you would like to talk to the AI without pressing any buttons, you can use our separate vamX Deepgram App (free for all vamX AI Chat users).

The updated vamX Deepgram App allows you to use a single keyword as a "send now" command.

If you choose the keyword "Jenny" and say "Let's go play Jenny" the AI responds to "Let's go play" (the keyword is removed, so you can use any word).

Keyword sending prevents your speech from being cut off if you pause momentarily, without creating long delays by ensuring you've finished speaking by waiting for a couple of seconds of silence.

You can also use normal send delays, so that Deepgram sends after 0.5 seconds (or other amounts of silence), or send by whichever comes first (0.5 seconds of silence or the keyword).

The large set of Non-English (plus some English) voices from Verbatik are now much faster, leading to a better experience for speakers of other languages.

The vamX servers were updated to fix an issue with the Verbatik voices due to API changes. During this update we improved the way we handle Verbatik voices, and also added new Verbatik voices, including voices for additional languages.

Due to the faster nature of the voices, you can also use them for chats in English, as additional voices. To do this you would need to scroll down in the voice selection to find English in alphabetical order by language under the Verbatik voices section.

Choose from 3 main servers, including the new Fast Global Server.

Use the new vamX AI Chat -> AI Music area where you can generate music or discover community made music.

Automatic Reconnection
The server now automatically reconnects if the audio connection is lost.

Quicker Voice Recognition (speech recognition)
This makes the overall AI response speed faster when you are talking to the AI (using your voice).

Added new Improved Quick Voices
To use these you need to switch to them in the voice drop down. These have better voice quality than the Quick Voices but are much faster than the Improved voices. These voices work in English only, but when used can speed things up dramatically (compared with using Improved Voices, or other non-quick voices).

Added X 3 personality (from the video demo) to the main personalities list.

Added the new Many Slow Unreliable Models item when choosing a NSFW model. This is a library of over 100 language models, most of them 70B (large). These models generally have a 10-15 second delay before each response starts, and if the wait is > 15 seconds vamX switches to responding with Lunaris. These models will not always work, but it can be fun or interesting to play with this huge variety of possible responses. There are also some models that should work well specifically in Japanese (although we haven't tested this) including EZO, Swallow, Himeyuri and Karakuri.

GENERAL VAMX (NON-AI CHAT) FIXES

Fixed issue where vamX VaMSync didn't work for non-vamX scenes where the people weren't named Person (female) and Person#2 (male).

See more fixes and new features on the Download vamX AI Chat 1.44 page.

Coming Soon - Improved Response Speed with New English Voices - Funny Outtakes and Unused Clips Video Demo (this video shows voice quality only, speed improvements video coming soon).

These voices have a pretty high voice quality for being generated almost instantly (for short phrases the delay is < 0.5 seconds for voice generation, instead of ~ 2 seconds).