Replica Studios is making giant leaps with AI generated voices in the video games development and animation space. Our plugin for Unreal Engine, currently in beta, has received a tremendous response from the game development community, and we're excited to see what new uses might emerge for this amazing technology.
Using artificial intelligence and neural networks, Replica processes recorded scripts from our voice actors and integrates the output into a TTS (Text-to-Speech) interface. Our AI learns to mimic the pitch, timbre and pacing of a given voice recording, and from this our customers are able to generate their own lines with similar characteristics.
If you're interested in joining the Replica Studios suite of AI voice actors, please send us your details for consideration.
We are looking for experienced voice actors to create Replica Voices to be featured on our platform. Each Replica Voice needs to be trained for a specific role.
It's common for voice actors to have more than five or so voices in their repertoire. Each of these voices, or characters, can become a Replica Voice on our platform.
For example, a voice actor could train a Replica Voice in the style of a superhero. The same voice actor could train another Replica Voice that's great for narrating a story.
You may have heard of the voice actor behind the voice of Siri. She trained Apple's AI voice by recording 50 hours of her voice. Like Apple, we also require voice data; however, unlike Apple, we only need two hours of voice data to train a realistic-sounding AI voice.
We're creating a range of AI voices for use in all kinds of projects, and we're looking for professional voice actors to help.
Frequently asked questions by voice actors
Will AI voices put me out of business?
No. This is an experimental technology, and isn’t going to replace human voice actors any time soon. Just as when Jim Henson created The Muppets, those adorable felt and foam characters didn’t render human actors obsolete. The voices generated by our AI do not have the quality, nuance or flexibility of a human voice actor, and we don’t see Replica Voices competing directly with human voice actors in the foreseeable future. What we do see is a bright and exciting opportunity for voice actors to generate revenue streams from areas of creative production that, until now, have generated none. We are aiming to work with voice actors to help them transition in the long term so they don't miss out on this kind of opportunity.
In large, dialogue-heavy video games, there is rarely a budget for placeholder voices. The prototyping phase is a very important part of development, where scripts for cut scenes can be tested for timing, and overall dialogue can be checked and heard to see if it flows. Until now, placeholder dialogue was typically done by whoever was available in-house, or a poor quality text-to-speech engine, or not at all. Instead of recruiting programmers to moonlight as reluctant voice talent, studios can now use Replica Studios to generate compelling, emotive placeholder voices and save time and money in the process.
How can voice actors benefit from creating Replica Voices?
We have found that creatives from many different industries are looking for a more cost-effective and efficient way to add voice performances to their projects. These projects range from indie games and animation to software products and also narration for audiobooks.
Since releasing Replica free in early access to the public in 2020, we have received a lot of great feedback and insights and have had many customers requesting the ability to pay for more speech credit. This feedback confirms that buyers are looking and willing to pay for unique AI voices to use in their projects.
In the not too distant future, we believe voice actors will be paid to create AI voices, rather than spend long hours in the recording booth.
Imagine being paid each time your Replica Voice (which only took a couple of hours to create) was used by one of our customers. Now imagine if you have 10 Replica Voices on our platform.
One of the most common requests we get is for more diverse voices, which is why we are looking for more voice actors to help us create the next wave of Replica Voices. Replica is currently in early access, and you can try our product out by signing up at replicastudios.com
Do we own your voice forever?
You will always own your voice. The Replica Voice you create is a version of one of your performance styles. For example a hard sell advertising performance, or a soft children's book narration performance. We've added more information about the contract here.
Who will be using my Replica Voice?
So far, most demand has come from industries or products who don't work with traditional voice actors (think Siri on iPhone, dynamic voice advertising, dynamic games, etc).
We also have interest from game development studios who are looking for placeholder audio while they develop their games. They'll bring in traditional voice actors once they have finalised the script.
Can I control how the Replica Voices I create are used?
How long does it take to create a Replica Voice?
The AI needs a minimum of 1 hour of 'finished audio' to be properly trained. This may take several hours recording and editing. Depending on the emotional range of the character you perform, this can be up to 4 hours.
We'll provide you with a guide that contains the character brief, scripts, and instructions to help you get it right the first time. We'll then use the audio you record, to train the AI to speak like the character you perform. Training can be done in parts, to add additional styles and emotions to the Replica Voice. On average, it can take one week from the beginning of the process to having your Replica Voice listed on the platform.
Is the quality of a Replica Voice as good as a real voice?
No. Replica Voices suffer from lower quality, pronunciation issues, and limits to how expressive they can be when compared to a real voice. Currently, all Replica Voices have a maximum resolution limit of 16bit 22050Hz. With EQ and post-processing, some of the defects can be masked. Some voice actors (like Hank Azaria from The Simpsons) can perform several ‘character impressions’, but a Replica Voice only ‘sounds like’ one character at a time. Voice actors can create several Replica Voices, each with an individual character from your repertoire of impressions.
Will my Replica Voice replace my regular work?
No, at least not in the foreseeable future. We think Replica Voices will create new revenue streams for voice talent and co-exist alongside your regular work. As technology improves over the coming decades, we imagine voice actors will be paid to create character voices, rather than spend physical hours in the recording booth. However, we cannot predict when this will happen. While this technology is young, it is starting to gain traction with creatives and there are numerous other companies that provide this service. Replica wants to work with voice actors to help them adapt to a changing technological landscape.
What can my Replica Voice do that my physical voice cannot?
Your Replica Voice can work at any time of the day and work multiple jobs at the same time. Your Replica Voice can work for long periods of time without breaks, or on-demand to fit the schedules of creatives while still earning a set hourly rate (no commute and coffee expense!). Your Replica Voice can do the kinds of jobs your physical voice can never do (eg : dynamic game content, chatbots, voice assistants).
What limitations exist for Replica Voices?
Replica Voices, and AI voices generally, cannot match a real human voice actor in many ways. They can’t act, or react to anything other than the narrow input they are given. They don’t really transition between emotions in any smooth sense. They can’t laugh, cry, scream or even breathe. If the range of emotions and nuances delivered in a scene by a human actor were colours on a canvas, an AI voice would be a single primary colour drawn in crayon, instructed by remote control.
Can my Replica Voice be stolen?
Your Replica Voice is stored securely on our servers - think of it as “voice data” composed of ones and zeroes. Even if someone were to steal your “voice data” it is not possible to use it for anything, since that is just one tiny component that needs to work with several other complex AI algorithms owned and operated by Replica. Replica will do its utmost to protect your Replica Voices and your data. You can choose to permanently delete all your Replica Voices at any time after the licence period has elapsed.
New opportunities for voice actors with Replica Studios
We are in the process of building relationships with some outstanding voice talent, and look forward to meeting more of the voice acting community as we explore this emerging industry together. This is a new technology and it is enabling the development of entirely new game mechanics and genres, many of which haven't been imagined yet. The use of AI voices will likely be dominant in areas where traditionally recorded voices would be impractical or impossible. Get in touch if you’d like to be a part of this exciting venture!
If this sounds like something you'd be interested in being a part of, please send us your details for consideration.