Sometimes our digital learning needs a voice, and not just any voice. In building towards a more authentic learning experience sometimes we need to say things using audio, in our own words, using our own voices. But correctly recording large amounts of audio for a learning project can be time-consuming, it can be inconsistent if not all recorded at the same time on the same day, and can sometimes be a logistical nightmare to get your chosen vocal talent to become available for the recording session (eg. having an executive-level staff member taking half a day out to record audio can be tricky). This is where new AI-based voice technologies come in.
What is AI voice replication?
AI voice replication is a form of technology that allows you to train a computer to be able to replicate a specific person’s voice, by training the computer to understand an individual’s vocal patterns. If I need to sample my own voice, as an example, I simply jump into a voice replication tool, record myself saying specific phrases and then the computer can go away and create a virtual facsimile. The amazing thing about this is, once a voice training session is successfully completed, you no longer need me to record audio the next time you need my voice. You simply jump into the voice replication application, type out the speech you would like me to say and the clever text-to- speech engine turns your words into audio, spoken by me!
Why use voice replication?
From a digital learning perspective voice replication opens up a world of possibilities; from having the CEO step through key training messages, saving valuable time by recording key voice talent in one short session and then utilising the voice sample throughout, and even arranging external experts/celebrities/people of interest to become the brand ambassadors and “voice” of your training programs.
By using the voices of real and known individuals, it also adds a layer of engagement and authenticity to the learning and can assist learners in connecting to the person and the content more deeply than the use of a generic voice artist or traditional text to speech technologies
But you could make me say anything?
Rightly, AI voice replication brings up valid concerns around how a person’s voice sampling will be used. Technically, you could use the voice sample to have an individual say anything you liked, including some not so nice or inflammatory things. This is where it is important to have protections and processes in place to minimise the risk of voice misappropriation. Protections and processes could consist of:
- Ensuring the person whose voice is being recorded understands the purpose and use of their voice (and where possible signs off on final audio before learning materials are utilised)
- The voice replication technology used is password protected so only authorised people can create audio files
- There are clear organisational policies in place around the misappropriation/misuse of a person’s voice sample.
Voice replication technologies are only in their early days, but there is already some engaging and interesting outcomes being achieved. If we wish to continue to innovate, increase engagement, and create efficiencies in our development processes, voice replication technologies should be considered for use within our digital learning programs.