Advertised Features

Instant Avatar

"Utilizing the power of Artificial Intelligence (AI), HeyGen allows you to create hyper-realistic avatars that mirror your physical appearance and mimic your speech. In just 5 minutes, your personalized instant avatar will be ready to greet you."

Impressions: The stock avatars HeyGen provides are impressive and highly advanced. But the ones the app creates with your content are really not that good, and do not mimic your speech in the way you'd think.

• In one case - it changed the accent from British English to American, without being told to.
• The voice clone option is also emotionless most of the time and sounds like any voice-generator you can find out there, though it does at least clone the way you sound accurately.
• The "Instant Avatar" feature or any other plan prior to "Finetune" do not allow you to remove any background elements from your video - even a simple green screen remover, it forces you to upgrade your current payment plan to the "Instant Avatar Finetune" in order to do that.
• Movements and mannerisms of your avatar look unnatural and puppet-like.
• The image generator they're using is also very bad and produces really badly composed images.

• At the time of signing up and paying for the plans, I was not aware of the existence of something called "The Avatar Plan". I only discovered it by accident, buried in the FAQ at the bottom of one of the features pages.

Before being forced to upgrade our plan once again, we were trying to figure out how to buy more credits, and I even contacted support:

to which they replied

but apprently, the Avatar Plan lets you buy credits as you need them, although the language isn't 100% clear - probably to avoid lawsuits.

Did their support agent not know of the existence of this Avatar Plan? The language used is not clear.

Voice Clone

I was going to describe this feature until I visited the feature page to add some of their own quotes in this article (see below).

The one thing I will say is that if you want a voice that isn't an emotionless robotic husk of yourself - you will discover it after you've already paid for a plan and created your footage - by being told to purchase the Pro Voice upgrade.

Honestly, you can look at the page for this for three hours and understand nothing. See it for yourself here, or look at the images and comments below.

The language in this page is either very vague, or outright incorrect.

"Create your ai voice using video footage for customising your own avatar"? What a confusing sentence, I've used these features and know what they are - and I still don't understand this.

By the time whoever wrote this got to number 3 they just gave up. "English".

The FAQ part of that page follows the same patterns.

Conclusion

HeyGen came out as an example in the AI video creation space that shows some impressive technological advances. It brings exciting possibilities to the table with its AI avatars and voice cloning features and makes you think that this is only the beginning. However, the reality of these features often falls short of expectations, with avatars that can appear unnatural and voice clones that lack emotion.

Something that really concerns me with HeyGen is its complex and opaque business model. The use of 'fake currencies' like credits or tokens, similar to tactics used in video games, adds a layer of confusion for users. This approach obscures the real cost of services, making it difficult for users to understand how much they're actually spending. This strategy, often seen in video games to encourage more spending on microtransactions, makes me significantly doubt the company's transparency and fairness. I get it, it's a business, but this is not a good look or feel.

Moreover, the unclear explanations of features and inconsistent information from customer support further muddy the waters for users. This lack of clarity and straightforwardness left me, a customer, feeling misled and undervalued.

Lastly, HeyGen's journey in the AI video production world reinforced my opinion that when it comes to AI technology, we need honest business practices and rules and laws to keep companies in check and consumers safe. It is vital to demand transparency and accountability.

In my opinion, HeyGen has the potential to be a real contender in AI-driven video/audio creation, but getting this status will require user-friendly and ethical business strategies.

A bit of fun -  Questioning Ethan Mollick's Assessment

Make sure to read his full article, found here, it's a great read!

I can confidently say that I disagree with his assessment. Right now, these avatars are not that good. Unless you pay hundreds of dollars in tokens whatever you generate is wonky, looks and sounds fake, and barely works as advertised. Sure, you can "make it say anything", but unless you have the most expensive add-ons to your already subscribed payment plan - it sounds like one of those automated voice generators we had in the early 2000s. And it moves the same way their visual counterparts did.

Also, In order to generate anything of yourself, HeyGen asks you submit a 2 minute video of yourself in a studio talking clearly in front of a green screen. I am not certain which option allowed him to upload a 30 second video and generate such a high quality avatar of himself - but it is possible that it's buried somewhere where I simply haven't discovered it yet, similarly to their other hidden pricing additions and plans.