My Journey to implement 'voiced' character Avatars, live response chatbots, and a livestream together

View this thread on: d.buzz | hive.blog | peakd.com | ecency.com
·@asynckronism·
0.000 HBD
My Journey to implement 'voiced' character Avatars, live response chatbots, and a livestream together
## Introduction & Backstory

Participants who view my account or have read my various posts are probably aware that I enjoy fishing in Multiplayer Video Games. It takes me to places I normally wouldn't go and see areas from a different point of view that can be cool. Fishing is also an activity I can feel I am making progress (Either making gold or materials for cooking/alchemy) in the game without much input. It's also the chance of random items being reeled in that are unique or odd from the norm. Fishing in a Video Game has this element to it that sometimes incites conversation or interaction from nearby players also fishing. There's not much else to do while waiting for your line to pull and conversation is a good way to pass the time both in real life and digital.

I've created a series of posts([1](https://peakd.com/asynckronism/@asynckronism/making-a-virtual-fishing-journal), [2](https://peakd.com/asynckronism/@asynckronism/virtual-fishing-journal-project-zomboid-beginning-entries), [3](https://peakd.com/game/@asynckronism/the-gaming-fisherman-warframe-fish-tank-edition), [4](https://peakd.com/hivegaming/@asynckronism/reviewing-pc-gamer-article-and-my-thoughts-about-fishing-in-games), [5](https://peakd.com/myblog/@asynckronism/open-world-vs-mmorpg-fishing), with many more in my blog history) with this hobby/passion in mind and began cataloging my ventures out in the virtual waters. What I caught, weird events that happened, and player interaction if any occurred. I then started making posts about my adventures with a few requirements, Had to have a fish to catch as the goal, had to enter the zone from an entrance, Used a 'crafted' Generative AI as a fishing guide and that was fun for a bit. It added some depth and story to the fishing journey and wasn't just another post where I am suddenly at a location, complete the point of the post objective, and end the post with results. There was a story to read along and I felt that added to both the quality of the post and the chance of more interesting photos to add to the post. I spent a lot of time on this concept, even sinking hours into the grind that is Warframe to unlock the Fishing Aquarium!

## The Idea

I was feeling pretty happy with what I had created and wanted to go further with the idea: A Podcast! Low maintenance, not too taxing on my hardware, no requirements much more than what I already had, The unique thing would be the sounds of fishing in the game with my narrative and an AI guide to pitch in on occasion. I would have to edit parts of it but was looking forward to keep it a mostly raw audio file to upload.

I had set up pretty much everything and was ready to set off when I was suddenly thrown off course with the idea of 'Live Social Shopping' being the popular thing (Gary Vaynerchuk). People live streaming objects ready to sell to viewers on the livestream. I was heavily resistant to the idea because I am not a 'social' person, I had to really get over the sound of hearing my voice and that was a challenge already, and I did not want to get involved with the world of streaming. 

I researched the concept though and pondered my orb for a bit and talked about  it with others, before I was told that other creators might be like me, wanting to participate with the newest trends but not wanting to show hide or hair of themselves. I am (unfortunately) the type of person that will work their fingers to the bone for somebody else but will do a minimum amount of effort for myself. If there were others that wanted to be 'faceless' creators while still participating with the newest trends to make a living than I could at least provide a guide to help them out as my setup I worked on for my podcast idea might come in handy.

I looked at what I had already created and thought about twisting the concept of "Live Social Selling' for myself, I could create a guide but I needed motivation that I would happily participate with, and that's when the idea came to me: Live Social VIRTUAL Experiences. I would create a character showing off virtual destinations in multiplayer video games. "But isn't that just live streaming?" yes, most definitely. The main pull or defining feature of the stream would be the character, their goal, and the ability for others to join in on the story being told.
 

I had already built up everything for a podcast, how much harder would it be for a livestream?

## Beginning Research

Turns out pretty difficult.

I did not want my face or voice to be a part of the livestream, I did not want to be the character or put on a 'fake persona', I wanted an 'Avatar' to be the 'social' person on screen. Creating an Avatar seemed daunting as I am also not savvy with 'coding' or 'model rigging' and I am no artist (As shown by my crappy MSPaint art). The tools required would have to be simple enough to implement but be able to achieve exactly what I had in mind, An Avatar created and displayed, Live generative Text-To-Speech that could listen to me speak and generate a reply in a unique voice. Talk about specific and super niche!  

Creating an avatar takes a core design (MSPAINT), a name, an attitude and a story to go along with it, and since it would be a livestream avatar it would need a voice.

I delved further into 'Generative AI' by using my crappy MSpaint to generate a character and looked at my 'AI Fishing Guide' for what I set its responses to be: Gruff, Superstitous, weathered, Storyteller but generally helpful. While crafting this 'Gaming Fisherman' character I thought of Captain Quint from Jaws and his demeanor, responses towards events, and that general attitude towards fishing. Cool, so my avatar would have a generated voice similar to the characteristics of Captain Quint.

Here is where my first issue began, Many of the Generative AI Voice tools I found were 'generated' into an audio clip. Not really helpful for a Livestream as I would need to write out a script, wait for it to generate, then play the audio. It would just be me but with a different accent. Not really helpful for my goal here so I searched and searched before coming across 'Character.AI', an AI Chatbot that could create a voice from an audio clip. I created the character, filled the instructions so the chatbot would understand how it should respond and the format the prompts being given before giving a proper reply. Character.AI also featured a Live Text-To-Speech type option which was exactly what I was looking for because I could describe events ingame(Fish caught, what the scenery looked like, if any players interacted nearby or if enemies attacked) to the character through my microphone before it would reply in its 'voice'.

Perfect! I had an Avatar image, a voice, and the name.

Now I needed to be able to display the avatar and have it's voice come through the livestream from an audio source instead of being apart of the computer background. I again struggled for a bit as many tools were for actual people and not for an AI TTS/Voice. Turns out Avatars have a certain aesthetic surrounding it because I searched through a lot of near-NSFW videos, still full of useful information, before discovering Reactive.Fugitech, a tool which uses Discord to display the avatar with 2 images of the avatar, idle and talking, and uses the microphone input as an audio source which activates the displayed avatar by lighting the image up to 'talk' and darkening the image during 'idle'.

Perfect! I could download a virtual Microphone Input and direct the audio source to go through that.

3 weeks of research, testing tools, and jury-rigging programs and applications to work for me and I now had a working avatar display, voice, and microphone set up to talk to the 'character' and hear its responses on stream WHILE I was playing the game.

I didn't know this would all stem from my 'Virtual Fishing Journal' concept. What an evolution from the 'Virtual Fishing Journal' idea!

## What About the Selling?

Still a work in Progress for me but I'm sure others will have a greater understanding of how to implement this.

So interaction with the character could be a big pull for participation with potential viewers, part of the 'Live Social' element of 'Live Social virtual Expeditions / experiences'

This wasn't just a livestream to stream but to 'sell' items. What would I sell though? Well, the livestream would revolve around the characters virtual adventures with fishing so the items should revolve around that 'virtual' aspect. Physical Merchandise would be the novelty, digital merchandise could be mystical and fantastical as fishing always has a 'mystique' element to it and I could even add things from the characters adventures as a 'digital' item with some flavor/lore text.

The big ticket item would be to go on an 'expedition' with the character to go fishing in a multiplayer game and explore or learn of 'hidden' fishing ponds in the game. These items would be the 'selling' part of the live social aspect.


Will this part of the project work? I don't know. But the tools needed to get to this point will help others who want to build up a project or venture they wanted to create but weren't looking to be the face of it.

## The Completed Process

Now that everything is said and done Here was my process to create this odd 

**Avatar**: MSpaint (Or whatever art application professionals use)to sketch out an attempt at a character > Feed to Generative AI Image generator to 'inspire it' > Now have a generative avatar > Take design and generated image to REAL Artist

**Voice**: Think up of a character voice that would best fit avatar > Find out voice traits, style, and accent > Use Elevenlabs to create a sample of the voice > Feed into Character.AI

**Character**: Live TTS / Avatar Response: Create Character in Character.AI > Input specific instructions so chatbot understands you are the handler describing events to it > Test and tweak until good enough >

**Audio**: Download Virtual Audio Input/Output > Set up separate browser audio source towards Virtual audio Output > set up Discord to direct audio input towards Virtual audio Input > test by talking to character.ai and seeing if the discord sound notification lights up

**Display**: Download Reactive.Fugitech > Connect to 2nd Discord Account > Input 2 Avatar Images for idle and speaking > Copy Browser source link onto OBS Source

**Livestream**: Download OBS > Set up all audio, display, and monitor/screen settings > Paste Fugitech browser link into 'Browser' format > Test discord to see if Avatar lights up / shifts image to 'talking' > Record to see if avatar is shown and live TTS can be heard through normal audio methods > 

**Marketplace**: Create idea of logo / merchandise related to character/venture > use printify or ko-fi as place to sell items > feature links or virtual goodies on personal character website > 

I've done all the legwork and the process is here for anyone to utilize now until forever (Because Blockchain is supposedly infinite).

'But what about the generated image for the Avatar?'

I WAS using a generated image as the 'avatar' during the stream and tool stream test / set up but generated images for use on a monetizatible venture is generally considered a 'no-no' with creator content as I have no means of actually claiming ownership over the image. I used a crappy MSpaint to 'inspire' the generation but that probably won't hold up if anybody used something close to my 'exact' prompt or even copied the avatar style and input it to a image generator.

So I took my crappy MSpaint character, my written specifics about how it should look, the generated image and hired an artist to create (Notice I used 'Create' and not 'Generate'? Words have meaning!) the avatar!

So while I continue to set up all the other aspects (Marketplace, Main Website, Character names on platforms, ETC) I have begun posting the videos of recordings/streams I have done on Youtube, sharing Game Photography on Twitter/X to help build up a library, and start connecting all the things together.

I started this project with EXTREME reluctance and still would've preferred to create a podcast of my Virtual Fishing Adventures(Probably will still do) but knowing that the time and effort I've spent on this project could help someone else create was a big driving force about why I continued to participate with this project.

Creation should never be stifled or limited. If there is a comfortable way for a creator to begin then they should utilize any and every tool to make it happen.

## End Of The Post

Anyway, that is the main reason why I haven't been on Hive in a few months (If anybody was missing my paragraphs of comments). Because I've been building a project that'll hopefully help others.

That's all for now. Thank you for reading! Don't be hesitant to participate in these tools, they are used for a multiple of things (Some bordering on NSFW) but are available for mostly anyone to utilize for their (and others) benefit.

Happy Creating!
👍 , , , , , , , , , ,