All the data needed to create these voices have been generated with the help of English-speaking professional actors, so speaking in English will definitely give you the best results.

That being said, other languages can still work. But be aware that our AI voices might have difficulty picking some of them up, due to the uniqueness of their sounds and articulations.

For an overall better experience, try these tips when using our voices:

🎙  Speak naturally and avoid any rushing or slurring.
🚑  Make sure there’s no background noise.
🛠  Try to avoid filler words and expressions like “um,” “uh,” “er,” and “ah.”
💨  Avoid breathing into your microphone.

Transforming one voice into an entirely new voice in real-time requires a powerful engine and a whole lot of data. All the key components that make up the human voice must be considered in the process, like harmonics, formants, resonance, inflection, articulation, and more.

This new approach still contains elements of creative sound design, but at its core is a quite different process. It’s important to recognize that these AI voices are not just “voice cloning.” Far from it. They are a fusion of technology and creativity that allow you to explore new use cases and be inspired like never before.

To deliver superior quality throughout the process, we work with professional voice actors whose vocal characteristics match the voice we want to recreate.

We then give them scripts to read from and record, which include specific combinations of sounds and words that we use to feed our AI-based models. The more data we feed to the model, the more accurate the final results.

But feeding the data to the engine is just one part of the equation. Once that part is done, we think about the creative use. And that’s when our sound designers take these voices and turn them into characters that you can impersonate.

This step involves adding things like dynamic effects, filters, and music beds. Throw it all together with our soundboards, and these characters come to life.