There are various predictions about the affects of AI on the audio industry, they range from a fantastic revolution to us all losing our jobs. We consider one area that’s already being targeted, voice over and ADR work.

What’s In A Voice?

If you’ve spent any time on social media then you’ll soon spot them. They are AI generated voice over bots being used as a cheap option for narration on short media clips.

No one would be fooled by how bad they are. They are stilted, the nuance sounds like an automated announcment on the London Underground, and after a couple of days you realise these things are popping up everywhere.

Voice over artists can rest assured, their skill and experience is not going to be replaced by one of these cheap voice bots.

Then Came Alec Baldwin

We’ve already heard some of the incrdible work being done with cloning of famous voices. Let’s be clear, this isn’t some smart impressionist tricking us. Instead this is machine learning being able to take the character of a voice and from a line of TYPED text be able to replicate their voice.

If you are wondering how good this technology is, then check out this new clip with the actor Alec Baldwin, who is appearing with the real Alec Baldwin on a podcast about AI. It was posted by Jay LeBoeuf, Head Of Business & Corporate Development at Descript on Linkedin.

As you can hear, AI is doing a good job of replacing Alec Baldwin’s voice. Check out the podcast for an excellent discussion about the implications of AI.

Who Wants To Be Replaced?

It may seem like a no-brainer to ask the question, who wants to be replaced by AI? No one in their right mind would want that, it has implications to livelihood and many other things.

However, imagine you’re an actor and have been on a 6 month shoot for a TV show or movie. Some months later you are either on your next project or on the beach and the edit is in full swing. There’s a problem, it could be technical or creative, but it means you need to be in a voice over booth to re-record some lines.

On a recent podcast Hollywood audio professional Reid Caulfield spoke about this very issue. He explained how many top actors were refusing to go back and do ADR on a project they had worked on months before. Of course wild tracks are a possibility, but that’s assuming you know what dialogue needs replacing at the time.

You can listen to the entire discussion on the podcast.

However, what of you could be sent the lines of the script and approve that AI be used to do the job?

Glass Half Empty Or Half Full?

At first glance the idea that voice talent can be replaced by AI using typed text to generate the audio can appear horrifying. There are all sorts of legal issues around the use of a voice, those can be fixed with laws and well written contracts. There’s also issues about the change to work.

Do we need voice over talent if this technology exists? Do we need ADR professionals if a computer can generate the lines in a second?

We think in both cases there are advantages for both areas of the creative industry.

Even with the best AI technology, there’s nothing better than real talent in the room reading a script and giving their own creative take on the performance. Right now, AI voiceovers aren’t even close, but they will become so good you can’t tell the difference.

However, people don’t hire voiceover talent just to read the words, they want the voice, in some cases A voice, such as Judy Dench or Morgan Freeman. Or even if it’s not A list talent, there are plenty of voices in the voiceover world that are in big demand, both for the character of their voice and their ability to perform at the highest level for a client. So AI can’t replace that, if for no other reason for legal ones. However, the law needs to be strengthened fast to protect those who make a living with their unique voice.

And what about ADR professionals? If a computer can replace the words without the talent being in the room then what job do they have?

Let’s consider the opportunities available to both professions with this new technology?

Take voiceover talent. What if you have control of your voice, even in the world of AI. The AI gets so good that it becomes indistingishable from your real performance? This gives you, the owner of the likeness, the chance to do things like be in two places at once, to in effect, increase your productivity and income. You may be asking, isn’t Economics 101 based on supply and demand. If there’s more of me, then surely that means the price goes down. Not if you’re the only person able to use your voice, AI or otherwise. This is why it’s critical to get strong laws in place around the use of audio likeness and to do it urgently.

Now to ADR. Some think ADR is a technical job, in some ways it is. Record a high quality part and make it fit to the original take from set. We already have technology like Revoice Pro and VocAlign that can do this, but it hasn’t replaced ADR professionals, even after 20 years. This is because ADR is as much a creative task about getting the right performance as it is a technical one.

Now if AI made ADR a job where the engineer requested script approvals from the talent’s agent and lawyers, then used AI to create the lines, it would still require skill to ensure the words had the right intonation and were fitted correctly.

Anyone who has used technology in the audio world for any length of time will know that equipment, especially software, rarely lives up to the marketing promises. Things still need anything from nudging to sheer brute force workarounds to get the desired result.

So ADR without needing the talent in the room? What’s not to like for anyone involved?

Summary

AI generation of voices has the potential to damage the professions of those working in the creative audio sector. However, with strong laws to protect both the talent and those working with them, it may be a blessing in disguise if we are the masters of the robots.