Creating an AI Digital Avatar and Voice Clone of Myself with Free and Low Cost AI Tools

Over the years I’ve checked out varied methods of making and animating a digital avatar, from merely creating a character for Cartoon Animator and voicing and animating it myself, to making a live motion capture ready Vtuber avatar puppeted and voiced by me in real-time.

Within the final yr or so, making photographic photographs discuss, utilizing AI and AI voice cloning has actually progressed. To the purpose the place I questioned if I may create a photographic AI avatar of myself, full with my cloned voice, that I’d use on a few of my movies?

Creating My Avatar: Artflow

To create my avatar picture, so far as I am involved, Artflow.ai is the perfect worth AI web site for creating constant digital characters (or ‘actors’ as Artflow calls them). That is just about their total focus.

Artflow’s actor mannequin coaching person interface.

You get your first actor at no cost, 100 free credit per thirty days (which equates to 100 nonetheless photographs per thirty days – get a further 50 credit in the event you sign up with my link), 4 minutes of video from their Story Studio (put your characters into video tales), and a couple of minutes of video from their Video Studio (sort of like Story Studio however you may add dialogue and lip sync your characters to it).

As properly you should purchase extra actors for a small, one time charge, and put a number of actors into the identical picture utilizing their actor director mode. The one unfavourable for me is you want a credit score or debit card.

Coaching a mannequin takes a minimal of 5 photographs however I’d suggest utilizing as many as the location will allow you to in the event you can.

As soon as skilled, Artflow will present you a collection of pattern photographs along with your actor in numerous conditions. Textual content immediate any state of affairs or use one in every of Artflow’s templates to get you began. It is fairly addictive, particularly in case your first actor is of your self or somebody you recognize. It is simply enjoyable imagining totally different scenes to see what you get.

Voice Cloning: ElevenLabs/Speechify

I regarded round for a great, free voice cloner and could not discover any that really delivered a voice that sounded something near mine. Most gave the impression of dangerous textual content to speech (TTS) voices from 5-8 years in the past.

For those who actually do not need to spend any cash then Speechify is the one web site I got here throughout that will allow you to practice the AI by yourself voice at no cost and create a high quality mannequin. They provide you 1000 characters of TTS to check out your skilled mannequin. Use them properly as a result of as soon as they’re gone, there isn’t any topping up subsequent month, they’re gone!

ElevenLabs Voice Coaching UI. It truly is this straightforward.

Sadly Speechify could be very pricy to pay by the month, however virtually a 3rd of the month-to-month value in the event you’re ready to pay yearly.

ElevenLabs however, when you cannot practice a mannequin in your voice at no cost, is extraordinarily low-cost for a single month of their Starter plan (once more credit score/debit card or financial institution switch solely). At the moment they’re thought of the Gold Normal in AI voices.

I made a decision to take the gamble and paid for a month on the Starter plan – completely value it! I uploaded lower than two minutes of pattern audio and bought a close to excellent copy of my voice in seconds, and now I’ve 40,000 characters of TTS to burn by.

Animating My Avatar: Hedra

Artflow’s Video Studio really does have the flexibility to make your photographs discuss however not in a totally convincing means. You will get some head motion however principally simply mouth motion. On the plus facet you can also make a couple of character converse in a single picture.

Hedra, which is at the moment free to make use of, by comparability, is particularly designed to animate an avatar head in a extra pure means. It is not excellent however briefly bursts, it virtually passes for precise video footage of the individual. It is in a position so as to add not simply lip syncing however head and higher physique motion too.

Sadly Hedra does not actually offer you any management over the output. You get what you get. However the result’s typically fairly good. It does restrict you to thirty seconds of video every era as a result of by about that time the magic is beginning to crumble just a bit.

Hedra is a 3 step course of. 1. Add your audio. 2. Add your picture (or immediate for a personality). 3. Click on Generate video.

Hedra additionally tends to crop comparatively near your picture’s face. For those who wished the complete picture, there isn’t any option to inform Hedra. Most likely as a result of Hedra’s animation is centered principally across the head and shoulders.

The Completed Clone Video

You may see my completed demonstration video of my cloned avatar under. 

I did not need my completed video to be all one digicam shot so I made an try and sew Hedra’s cropped video picture again into the complete picture of me in an workplace that does not exist outdoors of a pc (so far as I do know).

You may see the extensive pictures do not precisely work in the event you look too intently. I am certain I may make it seamless  with a little bit extra work. Nonetheless in the event you take a look at the shut up shot, the place I’ve stitched the edges of the picture again into the shot so it fills the body, that part is fairly convincing that it may very well be the true me.

For essentially the most half I feel it is virtually adequate to idiot individuals into pondering it truly is me. Although I’ll say, I do not transfer my head round almost that a lot in my movies. It will be good if Hedra would allow you to tone down the top motion a little bit.

What do you suppose? It might or will not be convincing now however simply keep in mind the AI mantra “That is as dangerous as AI will ever be. It is solely going to get higher from right here,” and I feel that is fairly good.