What's the difference between a Voiceprint and a Replica Voice and what are their purposes?
A Voiceprint is a small amount of audio of you speaking. Its purpose is to confirm the voice you are replicating is your own and no one else's.
Voiceprints can be used to generate speech, but because they are not trained on much audio data, they do not sound very good.
Once created, you can upload more audio files of your voice. Using both your Voiceprint and this new audio, we confirm that both voices are in fact your voice and your Replica Voice is created.
Because Voiceprints are created by recording directly into Replica, no one but you can create it which means your voice cannot be stolen.
Notice how the Voiceprint tone, prosody and pronunciation is not very good? This is due to minimal training time for as mentioned the purpose of the Voiceprint is not to perform, it is a security measure.
A Replica Voice is a fully trained AI voice and its purpose is to perform.
It is created by combining your Voiceprint with additional audio of your voice, essentially levelling it up. This is what we call training. Ideally you upload several hours of recorded speech, then your Voiceprint becomes a Replica Voice.
Because of this additional audio data, Replica Voices sound far better than Voice prints.
In this example you will notice a leap in quality of the compared the Voiceprint due to the Replica Voice being trained on far more data.
Got more questions? Message us in the app or email us at firstname.lastname@example.org