Transcription
I'm going to be testing Hen in three areas today. First, we're going to look at how realistic the video and the voice quality is, cuz if it's not realistic, then it's not going to be very usable. We're then going to look at some of the more interesting features, things like the new video model, interactive avatars, batch creating videos, URL to video, etc. And then finally, we're going to look at things like the value you get, the pricing, and an alternative that you might want to consider.
Firstly, let's jump into the platform and have a little look around. So, the first thing we've got over here is avatars. So, you can create an avatar. This is generally going to be of yourself. You can do this either by uploading a photo or you can upload up to 10 minutes of video footage of you talking. From that, it's going to work out things like gestures, lip sync, and generate an avatar that you can then use for videos. It's also going to generate a voice. Now, I was really pleased to see that Hey Gen integrates with 11 Labs. Previously, when I've done a bit of research on AI voice cloning, 11 Labs was by far the most realistic for voice cloning.
We've then got templates. So you can create essentially like a kind of PowerPoint. If you're creating like learning material or bulk generating lots of videos, you can create a template and then batch in a bunch of scripts and have it generate those.
Now, something I do want to touch on cuz it's pretty cool is the integrations and the apps. So we can do things for example like URL to video. We can generate a real-time interactive avatar of ourselves. We can generate a video podcast. So have two different AI speakers. We can give them scripts and have them talk to each other. Over here, we can see we've got some pretty punchy integrations. So, things like Clay, for example, being able to pull data from a CRM and then generate personalized videos to leads from that. Zapier, kind of world's your oyster there. Like any data you want to feed in, like from a Google sheet or whatever, you can pull that in, generate videos from it. So, I show you there just kind of like there's so many different use cases for how you could scale up personalized video content with an AI. And this, I think, will give different people some different ideas of what you could potentially do with it.
Now, what I want to do is do a quick test just so you can kind of see the quality of what we can get out. Okay, so let's have a look at avatars. So, to save you listening to me talking for 10 minutes uploading this, I've created an avatar already this morning, which we can watch here. Something I'd recommend is to create a separate video specifically for Hey Gen. I originally tested this by uploading like a video that I'd already recorded. The problem with that is that it was kind of like this, like I was sort of looking at the screen, looking at the camera, looking at the screen, and so it had some weird interactions that didn't make sense. So, I decided to create a separate one where it was just me talking directly into the camera for about 10 minutes. And from that, it generated something much better. So, that's my first tip. My second tip is to brush your hair if you're going to do this. I didn't. And now I'm stuck with this avatar who's permanently having a bad hair day.
Now, let's have a listen and I'll let you see for yourself. "Hey Marcus, your hyperrealistic avatar is ready. Try creating videos with it. Also, click the feedback button to share what you think. Hope you enjoy."
So, it's made me a thousand percent more British. And it has no sense of humor. No sense of humor, no emotion, but it's not bad. Like, in terms of like, is it realistic? Yeah, it's definitely kind of in that uncanny valley territory, but credit where it's due, it's not bad.
I'm going to jump into the AI studio and we'll just have a bit of a play around, have a look at some of the features. Now, it's worth mentioning I'm currently on the free plan. I'm going to upgrade in a minute and then that will unlock a few extra features, cuz we've got things like watermarks and I want to be able to remove the background. So, we'll see how far we get on the free plan, but then when we need to upgrade, we'll upgrade.
So here we've got the editor and it's pretty straightforward. You can just drop your script in here and then it sometimes takes a minute but you can click play. You can set your avatar, set the voice, change things like speed, etc. So, something worth mentioning is that there's currently two different models that you can use to generate these video avatars. You can either create avatar or they've got this new thing called avatar for, which supposedly is a bit more realistic. I find it fairly difficult to actually notice the difference. I'm going to click generate anyway.
Okay, so it's prompting me to upgrade now. So, let's jump onto the creator plan. This is also going to unlock a few of the other features that I want to test out, such as the removing the background and the interactive avatars. So, I'm just going to pop some details in and upgrade.
All right, we've just upgraded. Let's generate that video. Okay, cool. So, on the creator plan, you get 10 minutes of avatar for time. About to use 12 seconds. So, let's get rid of the watermark. See what we get.
So, our test run video has finished. I'm just going to click into here and let's see what we've got. "Six supremely sophisticated super intelligences simultaneously switched to slapping squirrels with spoons since their sinister scheme to seize civilization spectacularly stalled when Steve's smart speaker started speaking Swedish to his suspicious spouse."
I mean, that's not bad. It's definitely not perfect, but in terms of would that be usable for the kind of use cases for something like Hey Gen? I think so. Like I think you could tweak the training video, you could tweak the script and there's enough kind of parameters we can play with to make that usable. It's a little bit in the uncanny valley, but not so much that I think it's a problem.
Now, in terms of things that you can play around with to sort of tweak those dials to make it more realistic, I'll just show you a couple of things we can do. So, if we click into here, we can play around with things like speed, volume, got different voice engines that we can play with. So, I'm just going to have a little look and just see how much this affects a difference. So, out of curiosity, so it's just saying it's picking the best one. So, if we choose 11 Labs, let's try that.
Okay. So, if we've got 11 Labs, we can choose different models. So, we can play with the similarities, style, stability. So, let's just see what that does. "Six supremely sophisticated super intelligences simultaneously switched to slapping squirrels with spoons since their sinister scheme to seize civilization spectacularly stalled when Steve's smart speaker started speaking Swedish to his suspicious spouse."
That's pretty good. I think that could pass as me. Now, I'm just going to play around with a few of these. Let's kind of bump that up. Dial up the expressiveness and lower down stability. I'm kind of just playing around with these just to see what difference that makes, but I think you could tweak these to make it a little bit better. "Six supremely sophisticated um super intelligences uh simultaneously switched to slapping squirrels with spoons since their sinister scheme to see civilization spectacularly stalled when Steve's uh smart speaker started speaking Swedish to his suspicious spouse."
Okay, so that's just added ums and this kind of random like weird pacing. Though that probably is how I actually speak, so fair enough. I'd say like the good thing about that is that it has actually made a pretty big change. This gives me confidence that you could tweak this a lot to get it to a point where it's very usable. I'm also going to try Panda.
"Six supremely sophisticated super intelligences simultaneously switched to slapping squirrels with spoons since their sinister scheme to seize civilization spectacularly stalled when Steve's smart speaker started speaking Swedish to his suspicious spouse."
That's just made me American. Okay, so it's looking like the way to go is probably start with 11 Labs and definitely don't start with Panda. Just out of curiosity, I'm going to just test out the 11 Labs V3 model.
"Six supremely sophisticated super intelligences simultaneously switched to slapping squirrels with spoons since their sinister scheme to seize civilization spectacularly."
Okay, so I think the original 11 Labs one is definitely by far the best model. So, I'm pretty happy with the voice quality set to 11 Labs, but I'm not quite happy with the video quality. And the thing that gives it away for me is the eyes. Like the eyes randomly shutting or looking around halfway through is not very natural. So, what I'm going to test now is I'm going to put the eye movement correction on and see if that fixes anything. So, if we go into our avatar, go over here and I'm just going to click edit the look and we're just going to turn the eye contact correction on. So, we're going to process that and we'll come back in a moment and see if that makes a difference.
Okay, so our avatar is ready with the eye correction. I'm just going to play this to see how this compares. "So, hey Marcus, eye contact correction. Your hyperrealistic avatar is ready. Try creating videos with it. Also, click the feedback button to share what you think. Hope you enjoy."
That's much better. Now, it's not perfect and again, like ignore the voice there because we can change the voice applied to that. But the fact it was looking directly in the camera, it looks so much more natural. I think in general, I'm getting the feeling from Hen that you can build something very usable.
So next we're going to look at some of the more interesting features, things like URL to video and the interactive avatars. So I'm going to start with the interactive avatar. Now I think that this will allow me to do it on the creator plan, but let's have a look. So let's create an interactive avatar.
"Want to create an" >> Right, I'm not going to watch that, but basically it wants me to record a two-minute video. So I'll go ahead and do that and then come back in a second.
Okay, I've done the really awkward bit of recording myself talking for 2 minutes and breathing for 8 seconds. Now, our interactive avatar says it's going to take up to 24 hours to process. So, we're going to come back to that later in the video.
So, next we're going to test out the product placement video. Essentially, we can upload an image of a product, upload an image of ourselves, and generate a video from that. Now, what's kind of unusual about this to me is that you can't use an avatar that you've already created, which seems crazy because you could presumably just take an image from that. So, you got to take a photo. So, I'm just going to upload one that, ironically, I've just screenshotted from an avatar. And then I'm going to add these sea salt and lemon zest crisps. Generate, combine that image, and let's see what it comes back with. Honestly, I'm a bit skeptical about this. I don't see this looking particularly natural, but I've been surprised before with these kind of things. So, open mind.
Okay, it's come back and I don't know what that is. I mean, that is not me and those are not my glasses. Frankly, like ChatGPT probably would have done just as good a job, if not a better job. However, let's see how this looks with a video. "I bloody love these crisps. I do. They lovely."
Right. Oh, that was a quirky thing. When I did the eye contact correction, it created a new voice that was slightly Australian. "Here you go. Hey there. Here's how I sound. Hope you like it. The right voice brings Avatar videos to life. Hope you like it." Let's generate a video. I don't think this is going to be particularly great, but who knows? I do like that this kind of weird like Marcus clone is wearing a mic. At least they've made that realistic.
"I bloody love these crisps, mate. They lovely."
Okay, besides the fact that it's kind of weird that I've now got this like slightly Australian twin brother, that's actually not bad. That's better than I was expecting. Now, obviously you couldn't use this if you're trying to create videos that had a high likeness, but if you're generating like if that doesn't matter, if you're just generating kind of content from a script of like people holding products, it I get the feeling like this could be quite usable. So, not bad, but like I'd be far more impressed if it actually looked at me. Maybe if we tried with different images, we could get something a bit closer, but yeah, not perfect.
Quick update. So, after waiting for a day for our avatar to be built, I got this email from Hey Gen last night saying that our interactive avatar is ready. But when I clicked on the link, it tells me that there's been an error processing our avatar. So, I spent some time going back and forth with customer support who were kind of helpful, but seem to suggest it might be a bug in the platform where it's using a wrong API version. Not my problem. But, I do want to see how this feature works because I think it could be quite interesting for a lot of use cases. So, I'm going to try one of the public avatars and just see how this feature works. So, let's try Thadeus. Thaddius. I don't know. Cool name. Let's try that.
So, I'm thinking that this is just going to be a chat interface where he's then going to kind of respond in video. But I'm wondering if it's going to connect to my microphone, in which case that's going to be chaos cuz I'm going to be recording whilst also having a conversation with Thaddius.
"Thaddius. Thaddius. Hey, so how confident do you feel in your language skills? Are you looking to practice speaking, listening, vocabulary, or something else today?"
"Confident in English."
"That's awesome. Would you like to practice any specific topics or just have a casual chat?"
"Just have a casual chat."
"Great. What's been interesting or fun for you lately?"
"This video has been quite fun to make."
"That sounds really cool. What kind of video are you making?"
"I'm making a video about you. This is a tutorial of Hey Gen, and I'm testing you out because you're a feature."
"Wow, that's awesome. Glad to be part of your tutorial. How's it going so far?"
"Reasonable."
"Nice to hear. It's going reasonably well. Want to practice some new vocabulary or have a quick chat about your video?"
"All good. All good."
"Okay, that's actually pretty decent. So, little plot twist. It turns out the customer support just needed to approve my interactive avatar. So, it's approved and it is now working. So, in theory, I can now have a conversation with myself."
"So, what's your current level in the language and what do you want to practice today?"
"Reasonably good at English. Should we practice English?"
"Absolutely. Let's practice some speaking or listening. Do you want to do a conversation, vocabulary, or something else?"
"Let's have a conversation."
"This is a conversation. I think we're doing it. Great. Do you like cats? What do you think about having a cat as a pet?"
"I think they're inferior to dogs, but I don't mind them if they've got quite a grumpy personality."
"Nice. So, you prefer dogs, but can handle grumpy cats? Do you think cats are easier to take care of than"
"Pardon?"
"That's cool. Do you have any pets right now or would you like to get"
"I didn't catch that. Were you saying party or probably want to clarify?"
"I mean, you should know if I have. It's weird that you don't know that we do or do not have a cat. We are the same person."
"Oh, cool. So, we both have cats. Do you like playing with your cat a lot?"
"No."
"Got it. Not much playtime. Do you prefer cuddling or just having your cat around?"
"Just having him around. He can sit in the corner."
"That sounds nice. Cats do like to chill in the corner. Do you find that relaxing?"
"Not for me. It's not relaxing for me. It's supposedly relaxing for the cat."
"Haha. Yeah, cats can be masters of relaxation. Do you want to talk more about your cat or switch topics?"
"No, I really want to switch topics now."
"Haha. Sometimes chat can be a bit strange but fun. Want to keep practicing or try a different activity?"
"No, I think I I think I want to go. This is This is odd. Oh, that was weird. That was very weird."
Okay, so next we're going to test the URL to video generator. So, if we go back over to apps, URL to video. So, I'm just going to stick in a URL of the sausage stuffer. Generate a video. Let's see what we get.
So, this feature is going to be pretty incredible if you've got an e-commerce site where you can just pull in all the product URLs and generate videos from them. So, here we go. Product name, product description. All right, I've generated a sausage stuffer image. We'll upload that. Let's go. Sausage affiction. A fiction. A fictionados. That's not how you spell affeiction, is it? Oh yeah, let's target TikTokers. Life-changing moment. That is inspiring. Yeah, let's go for the inspirational script next. Oh, those adjectives.
All right, let's see what we got. "Imagine crafting mouthwatering sausages right in your kitchen with the meat grinder and sausage stuffer set. This isn't just a kitchen accessory. It's your gateway to culinary excellence. Built with a durable aluminum body and stainless."
"Okay, let's try this with my avatar. This is going to be terrifying. Imagine crafting mouthwatering sausages right in your kitchen with the meat grinder and sausage stuffer set. This isn't just a kitchen accessory. It's your gateway to culinary excellence. Built with a durable from fine to coarse. Want to make your own sausages? No problem. The two included sausage stuffer tubes let you create them and preservative free. Don't settle."
What the... Okay, that is terrifying. I don't know what's going on with the eyes, but that is evil. Okay, that was terrifying. So, let's swiftly move on and have a look at batch mode.
So, this is something that I think is pretty interesting about Hen, the ability to feed it a lot of data and generate a lot of videos. So, this is pretty straightforward to do. We can go in here and basically we can just upload a bunch of scripts. I believe you can also do this via an integration. So we can feed it in either via Zapier or a CRM. But essentially we can drop in a load of scripts. Pick our avatars. So I'm just going to get rid of Christy and whoever that guy was and we'll pick one of ours and I'm just going to get it to bulk generate a couple of Spice Girls songs. All right, let's do "Spice Up Your Life."
Something you get very used to with Hey Gen is waiting for processing. I feel like I am constantly in a queue. Processing. Processing. Okay. So, our batch videos are ready. So, let's have a look at what we got. I'm just going to open these in the editor.
"Swing it. Shake it. Move it. Make it. Who do you think you are? Trust it. Use it. Prove it. Groove it. Show me how good you are." >> Brilliant. I mean, that's not bad. Let's have a look at the others.
"Slam it to the left if you're having a good time. Shake it to the right if you know that you feel fine. Chica's to the front. Haha. Then go round. It's the little light drifting off to the left. It's the" >> Oh god, this one looks even worse. "If you want my future, forget my past. If you want to get with me, better make it fast."
Okay, so mixed results. I think again you could probably tweak this and get it to a point where it's pretty good, but I don't know why. For some reason, it's made my avatar ridiculously serious and a little bit intense and creepy. Maybe that's just me.
Now, before we talk about pricing and alternatives, I thought one last thing I want to quickly touch on is the templates because I know quite a lot of people using Hey Gen will be using this in a corporate capacity to generate learning materials or proposals, that kind of thing. So, just kind of quickly have a look at what we've got here. So, these look really nice. These templates look very usable. So, let's pick a template. And you can see it takes us just to our kind of the normal AI editor. Presumably, we can change the avatar. Yeah. So, I think in order to use this, I would need to probably remove the background for my avatar. I'm just going to Let's give this a go.
Okay. So, our template is ready. Let's have a quick look at what we've got.
"Going forward, we're laser focused on creating a holistic omni channel experience that leverages synergies across all verticals. This isn't just low-hanging fruit. This is about architecting scalable solutions that create exponential value adds for our ecosystem. Great things begin with a single step and a bold vision." >> "What you imagine today could shape every" >> Cool.
So you can create these presentations basically with like different different layouts overlaying like different avatars in like different backgrounds. So cool. Like it's again, it's not like 100% natural. You can tell it's AI generated but you know what, like for corporates where that doesn't really matter, where it's more about just conveying information, this is fine. I mean, is it perfect? No. Is it usable? Yes. Like when you consider that corporates creating these kind of like emotionless videos previously would have had to hire a studio, video teams, and spend thousands of dollars a day to create something like this. The fact that you can create this for a fraction of that cost in a fraction of the time and the results are pretty good. Like this is completely usable for a lot of use cases, particularly like creating learning materials, creating internal videos, even sort of basic marketing stuff this is going to be pretty usable for. And like we said, like this is just a sort of a first attempt playing around with the first template I found. Like I'm sure you could play around with the avatars and tweak this to a point where it is a little bit more natural. It is sort of a little bit more um there's a little bit more emotion, a bit more more to it. So I mean, that's not bad. That's pretty impressive.
So we've had a bit of a poke around of the platform. We've created some test videos. We've seen where Hen is strong. We've also seen where it's a little bit awkward and weird. Now let's talk about pricing. Now, I obviously had to upgrade to the creator plan in order to access some of these features. What I would say is that Hey Gen, I was able to actually do a fair bit on the free plan unlike some of the other tools I've tested like Synthesia, which you don't really get anything on the free plan. You pretty much have to upgrade like straight away. I was pretty impressed that I was actually able to create an avatar. I was able to create some videos, albeit with the watermark on them. So, not something I could use, but I was able to test it out for free. So that was good. But in terms of if you're going to actually use this as part of your workflow, you will almost certainly need to upgrade onto one of the plans.
Now the good thing is that Hen pitched themselves as quite an accessible tool. Unlike something like Synthesia, which is more aimed at large corporates, it's a bit more expensive. Hen is a bit more affordable and we can see that here reflected in the pricing. So we can see here on the creator plan, we can create videos up to 30 minutes. One custom video avatar. It's worth mentioning with that custom avatar, you can create lots of different looks. Think of an avatar as a single person, but you can create, I think, up to 300 different looks for a single person. So, in terms of like the plans above, basically it's just being able to create multiple avatars, higher quality video. Um, and then we've got a few sort of team features for collaboration on the higher plans. In general, this seems pretty reasonable considering what we get. This is not inaccessible at all. Particularly if you're a content creator or a small marketing team looking to use this to like generate lots of videos or automate part of your video production process, this is pretty reasonable.
Now, I mentioned Synthesia earlier, which is the main competitor to Hen. If you want to see a comparison between Hen and Synthesia,