AI voice improvements

According to the MIT Technology Review, the latest breakthroughs in the quality of AI voices are down to Deep Learning. With synthetic voices like Siri and Alexa, individual recorded words are combined to create sentences. The effect of which can often sound strange with a lack of cadence, pauses and breathing and adding in these elements needs to be completed manually.

Deep Learning is able to take a few hours of audio and learn to replicate the individual pattern of each voice for audio, through the use of an algorithm. Working this way means words can be used that are not present in the initial recordings, making these new AI voices much more flexible. This presents us with more opportunities as marketers.

But if we are all using AI voices, does this mean voiceover artists will become obsolete?

The impact on voiceover artists

A key part of creating synthetic voices via Deep Learning is having the original audio to begin with. Currently, a lot of the businesses offering these AI voices are working in collaboration with voiceover artists in order to create AI versions of their voice. This could lead to voiceover artists being paid for the time spent in doing the initial recordings and then potentially also getting royalties, each time their AI voice is used.

However, there is already previous evidence of voiceover artists being taken advantage of when it comes to synthetic voices. Susan Bennett, the original voice or Siri, was contracted by ScanSoft and didn’t know her recordings were going to be used by Apple. Although she gave ScanSoft permission to use her recordings anywhere, she was paid for the original recordings and not for their continued use.

Another voiceover artist Bev Standing is currently suing TikTok after alleging the company used her voice for its text-to-speech feature without compensation or consent. The worry in the industry is that contracts and fees have not yet caught up with the technology that is available. So, voiceover artists can unknowingly sign away use of their voice forever for a minimal fee.

There is also the added worry that with unclear consent and contracting, voices could be used in a negative way that could bring voiceover artists into disrepute. Whether that is being used in a sexualised video game, a context with increased swearing and bad language or simply for a company that they might not choose to work with.

Should we be using AI voices

But just because something is available does it mean we should be using it? More importantly, is it the best option? With synthetic voices there are a few things to think about:

  1. What do you need a voiceover for?
  2. Is it a supplementary or core aspect of your content?

For example, here at PinPoint Media we choose to use synthetic voices for the audio versions of our blogs. We do this because we want to quickly and easily provide an alternative option to reading. Because we produce so many blogs it wouldn’t be cost or time effective for us to record our own audio versions of these. So, a synthetic voice gives us an alternative option. For this particular scenario we’re excited that these voices may become more and more realistic in time.

In comparison, when we are creating audio voiceover for an internal video, authenticity and emotion are a key element of the process. Although these are areas where AI voices have improved, we can achieve a higher quality voiceover by recording this in-house with members of our team who can talk confidently about the subject.

Equally when we are adding a voiceover to a video or animation for our clients, emotion is often a key component in this process. Being able to direct a voiceover artist to get the right end result is crucial to ensure it blends seamlessly with the visual content. This is something that our clients depend on us to get right to ensure the project has the right level of emotional depth and engages with their target audience.

Senior Animation Producer Jessica Barder had this to say, “More often than not a voiceover is key to leading and driving our animated content. There are many occasions when I will jump on a call to direct the voiceover artist who will have been selected by our client. Call me traditional but I do not believe for a second that you would get the same bespoke quality of delivery from AI voiceovers. Working collaboratively with the talent, you are able to achieve a great result. There is definitely a place for AI voiceovers, but the project and its purpose would need to be carefully considered.”

Ultimately, like us, if you are thinking of using synthetic voices you need to weigh up whether they are right for your project. Although the cost and time impact of recording audio may well be a factor in your decision making, so should the importance of quality in your end result. We’ve written previously about why you should use professional voiceover artists and the many benefits that can be gained from working with professionals. So, it’s worth checking out our reasoning to help your decision making.

Are AI voices the future

So, after all that, are AI voices the future? Yes and no! There is already a place for synthetic voices, and we can see the contexts they are used in growing as AI voices improve over time. However, we believe there will always be a place for professional voiceover artists. Firstly, because they are needed to help ensure AI voices continue to improve and secondly because if you need a voiceover that engages on an emotional level with your target audience, you need a human.

We produce data-driven content from video and animation, to voice, audio and content strategy. Use the button below to enquire about our services.