Hedra, the popular AI character maker, has just announced a new voice cloning feature. You can now grab a short clip of your voice to clone, store and then use for any character you create or upload to the platform.
The new tool is only available to paid subscribers, so you’ll have to stump up a minimum of $10 a month to take advantage, but I’ve got to say it’s worth it if you've got the need.
One of the great things about Hedra is how delightfully simple it is to use. It beat out Runway and Kling in our lip-sync battle and produces consistently accurate mouth movements.
How Hedra voice cloning works
Sign up, log in and head to the create page, where you’ll find three main panels. The first deals with the text and voice audio you want to use, the second manages the image, and the final panel is for the output once the two are combined. So far so sweet.
The first task for voice cloners is to click on the New Voice button in the first panel, and record their voice in the clone voice box. The three lines of text you use is enough for the platform to build your clone, and it’s really very accurate.
Of course you can also upload or live record an audio script at this stage, instead of using your voice clone, or you can select from one of the off the shelf voices.
Once the audio has been sorted out, it’s time to upload or create a character image. The hot tip at this stage is to experiment with the image generator, because it’s very good.
The basic model used is StableDiffusion, but if you want stunningly realistic characters you’ll upgrade to a premium plan and go for the Flux Realism setting.
The results are superb, and hard to tell apart from real human faces, although the video resolution does lack some clarity. The lip and face movements however are excellent.
How well Hedra works
I would avoid trying to use animal characters at this time because the lip-syncing fails spectacularly. I tried a lion and a panda, and both times the head just wobbled and no lips moved at all. Shame.
The Basic premium plan allows for 20 minutes of video a month, videos up to 1 minute long and the voice cloning function with no watermark.
That seems like a fair price, although it would be nice if the company offered at least a short voice clone function on the free watermarked plan. But that’s a small quibble.
I’m also still struggling to find a truly compelling use case for these video avatar tools outside of cheesy marketing promos, but maybe the enthusiast story-telling, music and game market will step up and prove me wrong.
Overall it seems like this kind of total clone functionality is going to grow in quality and popularity as the models increase in power. Even Runway is entering this space with Act-1, where your own face controls the motion of the character.
Whether you want to create more business-like characters with tools like HeyGen or more fun characters with Hedra, it’s great to have a variety of tool choice for all aspects of the AI avatar process.
More from Tom's Guide
- Apple is bringing iPhone Mirroring to macOS Sequoia — here’s what we know
- iOS 18 supported devices: Here are all the compatible iPhones
- Apple Intelligence unveiled — all the new AI features coming to iOS 18, iPadOS 18 and macOS Sequoia
