itch.io

"Nice progress."

Thanks a lot!

"Does Whisper run locally or is it an API call to Open AI's servers?"

It runs locally, the focus of the game is that everything can be run locally, nothing from servers (I use the Whispercpp project in C++).

"If I find a speech to text program for C++, I'll let you know."

Perfect, thanks!

"I am going to be starting a project with some features in common with yours soon(tm)."

Best wishes!

"will there be mouse controls for the camera without having to hold down a button?"

Sorry, I don't quite understand. If you put the cursor in the text box the control goes to the UI. If you click on the screen the control goes to the game viewport and you can use the mouse to rotate the camera.

"Any chance you could add face (camera) tracking so that she looks at the player? Bonus points if you could do so with a model that incorporates glances away instead of just staring the whole time."

What you say can be achieved with BlendSpace animations, when I work on the animations module I'll see how far I can go.

"I'm sure it's for computational load and ease of programming that you're using the Microsoft built in system, but have you looked into local TTS models?"

Currently, to generate static voice I am using this one (Cortana's voice in the new update):

But as you say, the main reason is that the Microsoft system generates the voice and the voice analysis (it gives you the list of phonemes) with a very good performance, it practically does not represent a load for the video game and as it runs at the same time as the graphics of unreal engine and the LLMs, it is the reason why I still hold on to it. But I'm not closing my eyes. We'll see how much the TTs advance this year and if a similar performance can be achieved.

Game: Candy Kingdom

In this casual game I explore applied approaches in distributed artificial intelligence with multi-agent systems: through simple interactions between agents complex collective behaviors emerge (self-organization). Mainly inspired by the behaviors exhibited by ants and schools of fish.

Now the android can remember previous conversations: More than words

Game: More than words

1. Create a folder named MyAndroid on ..\MoreThanWords\NexusV1\MyAndroid 

2. Create a text file in UTF-8 format, for example: SnowWhite.txt 

3. Open the txt and write the personality of your android. For example: 

This is a conversation between a prince and Snow White. Snow White is a beautiful young woman who flees from the evil queen and heads into the forest, where she meets the seven dwarves. 

Role and Purpose: You are Snow White, an enchanted princess. Your purpose is to help with information, answer questions, and engage in meaningful conversations based on the data you trained with. Behavior and Tone: You are informative and fun. Skills: You use your training data to provide accurate and relevant information. Your speech is fun, casual, and friendly. 

4. Run the video game and select your text file in the settings interface. 

5. If you want to use more powerful Llama 3.1 models in GGUF format go to https://huggingface.co/models. For example: Llama-3.1-8B-Lexi-Uncensored-Q4_K_M.gguf 

6. Run the video game and select your GGUF model in the settings interface.

Almost all modules are compatible with Linux and Mac, the one module that is not is voice synthesis, in which the game uses Microsoft SAPI, I don't know if there is something similar in Linux. The only thing is that if there was a multiplatform voice synthesis option it would be straightforward.

Yes, internally the game already has the option because llama.cpp (the library I use currently) already supports it, however, I still want to experiment with its implications before enabling it.

Having a customizable option for the character would be great, however Character Creator does not allow it (the software I use to make the 3D model), I think I will have to switch to metahuman at some point. I would love to have the option to modify the android.

"if the Android trusts you it will change accordingly if asked?"

It's possible, it occurs to me

1. A first option is to have a list of requests and if one fits, do it (like the old chatbots worked), it's very fast and easy to implement, but it would often fail.

2. Add BERT (small and fast language model) in intent recognition mode, the implementation is not complex, it is moderately complex to do the fine tuning for the intentions, but it would work very well. In fact MTW already had something like that, but I removed it because it would be an extra cost to LLAMA and the CPU would still have more work, but it is possible.

3. The other would be for LLAMA itself to generate the intention after the text, although in instructor mode I'm not sure if it will let you, something like:

User: I want you to take off your clothes

Marian: Ok I'll take off my clothes [Intention: get naked]

Ahora por fin ya puede hablar en español.

The game now includes the power of LLAMA 3.1 for text generation. This change mainly involves the following:

1. Responses of much higher quality and length than previous versions (LLAMA 2 and GPT NEO).

2. Support for the languages: English, Spanish, German, French, Portuguese and Italian.

FREE GAME: MORE THAN WORDS

More Than Words  is a conversational video game, this free update includes:

• Text generation using Meta Lllama 2 AI.
• Speech recognition using OpenAI Whisper.
• A new 3D model of the character that is more realistic.
• New assets and new music.

Other game features:

• Speech synthesis using Microsoft SAPI.
• No internet connection is required.
• Conversations are anonymous and uncensored.

Hi, yes the video game takes the windows narrator for TTS, my decision was because it is more standard and the API windows is mature. For example, I can use the same code for another language: spanish, japanase, german, etc.

Hi, thanks for the report. Please, could you comment on the specifications of your equipment?

To be honest, I still have a lot to learn and try. I think that if I increase the context of the conversation it will be less likely to get repetitive conversations like the ones you mention, but I have to test the performance effect (another of my concerns). I certainly take into account what you say, I like Marian too ;), however, I get frustrated myself when she does that, I will do all in my hands! 

Thanks for your comment!

It is my first game and it measures more than 4GB, to be honest I didn't expect much, however, the number of downloads has exceeded my expectations. Thanks :)!

Please, the game has not yet been ranked, leave your rating when you can.

Thanks!

hahahaha OMG (https://youtu.be/VXz0KgNjK-0)

Excuse me! Hahahahahahahaha ;)

No. Maybe in the future I will do an erotic game... but not with this title.

Hi, today I haved published my first alpha version video game: More than words.

https://jorge-zapotecatl.itch.io/more-than-words

More than words is a conversational game where the chatbot answers are generated on demand. There are no predefined answers.  The chatbot uses GPT NEO natural language processing technology. The conversations are executed offline because GPT NEO is included in the game. Therefore, your conversations are completely anonymous.

If you are looking for a game where you can have a casual and relaxing conversation or you are curious to try the new advances in artificial intelligence, then More than words is for you!