Advances in AI technology and its applications are affecting every industry at the moment. In recent weeks, different music artists have had songs going viral with their voices 100% simulated by generative AI tools.
Meanwhile, scammers are using generative voice AI tools to simulate the voices of bosses, family members and public figures to extort money from unwitting victims. Does this spell the end for voice actors in an age of synthetic voice tools capable of creating or even replicating almost any voice?
To understand the strengths and limitations of generative AI voice tools, you need to know how the technology works. Algorithms can’t generate a voice from nothing. They need input data and, practically speaking, this takes the form of voiceover tracks produced by humans.
In essence, the algorithm collects data of the sound of the person’s voice and the way they talk. This includes intonation, pronunciation, tone and a depth of vocal features that can vary in many different ways.
People’s voices change as they get quieter or louder, express different emotions and talk about different subjects, based on an almost endless set of variables. AI algorithms need data to replicate all of these voice changes and nuances – so we are talking about a lot of data here.
Now, it is not realistic to expect a generative AI voice tool to replicate every characteristic of a person’s voice. The question is how much do these algorithms need to replicate?
Let’s say a company wants someone like Snoop Dog to say their brand logo in his signature voice. This is a far simpler task to generate with AI tools than a complete audiobook or a voiceover track for an animated movie – especially if range is required.
Generative AI tools are already doing an impressive job of replicating public figures’ voices. The caveat is that algorithms will have an easier job replicating someone as stylized as Snoop Dog and public figures, where large volumes of voice data is already accessible.
The simple answer is no – of course not. This doesn’t mean they can’t compete very well with professional human voice actors and voiceover artists though. For simple tasks like a brand slogan, the differences between human voice actors and AI-generated voices may not be too noticeable.
However, more challenging projects will reveal the limitations of the technology.
Here’s a quick summary of its main weaknesses at the moment:
- Range: Algorithms can’t replicate the full range of a person’s voice, tone, style or delivery.
- Consistency: Gaps in AI voice tools’ databases and their limited ability to fully replicate the entire range of an individual’s voice can make consistency challenging, depending on the nature and length of the project.
- Flexibility: Human voice actors can respond to instructions fast, understand the needs of the project and fine-tune their output based on what has been agreed for the project.
- Emphasis: Algorithms can’t (yet) emphasise specific words, phrases or lines with the subtlety of professional human voice actors.
- Variation: Human voice actors can convey a variety of emotions, tones, styles and other delivery features throughout a single piece.
- Copyright issues: The US Copyright Office has decided some AI-generated works may not be copyrighted and the rules around this are likely to change.
- Uncertainty: With governments and bodies responding to the surge of generative AI tools, the future remains uncertain.
However, generative AI voice tools aren’t inferior in every way. Given the right time and data to work with, they are very capable of producing results faster than professional human voice actors. In several cases, they can’t fully replace the voice artist, but they are able to handle basic tasks and projects such as corporate videos – and for these projects, their use will save time and money for the end client.
Let’s say for example that a generative AI voice tool can produce 20% of the audio for a project and the voice artist is required for the other 80%. That is still a 20% reduction in production time and, probably to a lesser extent, cost.
Not completely. The concern for voice artists (like many other professionals at the moment) is whether AI will take over enough work to affect their services, compensation and the amount of projects they deal with.
However, AI voice platforms shouldn’t be feared, but rather seen as almost assistants and alternatives for companies to use in simple projects and scenarios where a professional voice artist cannot be sourced. In general, they can be used to enhance the work of voice actors and not necessarily take it away from them.
If you need help finding the perfect voice for your brand (human or synthetic), our voiceover specialists team is here to help you navigate through different choices.
Get in touch with us about your voiceover project request by filling out the form on our contact page and we will get back to you.