AI Voice vs Human Voice: What Are Brands Actually Choosing?
AI voice is no longer a hypothetical part of the creative industry. It is already being used across advertising, branded content, localisation, internal communications and high-volume production. But that does not mean the human voice is disappearing.
In this episode of High Notes, BRAVA's Melissa Thom speaks with two people working at the point where creativity, production and technology meet: Mark Graeme, Head of Studio at OCX UK, Omnicom Production, and Mark Eaves, founder of Gravity Road.
Their conversations reveal a more nuanced picture than a simple choice between AI and human voice. For both, the more useful question is not "Which one wins?" but "What does this particular piece of work need?" Increasingly, the answer depends on scale, emotion, trust, authenticity and the role the voice is being asked to play.
AI voice is becoming part of the production toolkit
At OCX UK, AI voice sits alongside more traditional production methods. Mark Graeme describes teams testing and evaluating new tools, including voice and audio technologies, before deciding where they might fit into live projects. That process is not simply about what the technology can technically produce. Legal, moral and ethical considerations are part of the decision too.
Voice tools can offer access to library voices or, with permission, allow a voice to be cloned.
For Mark Graeme, these technologies are becoming another tool for producers, audio engineers, and creatives rather than an automatic replacement for existing craft. The question is how, and where, they should be used.
Sometimes voice is central to an idea from the beginning
Voiceover does not always enter an advertising campaign at the same stage. Sometimes it is considered later, when an edit is taking shape, and the creative team is deciding on music, sound design, tone and delivery. Other times, voice is part of the idea from the beginning.
Mark Graeme points to the emotional effect of hearing the right voice attached to a piece of creative work. A particular performance can help a client understand an idea, establish its tone and sometimes become inseparable from the campaign itself. That is one reason the conversation around AI voice can't be only about producing an acceptable audio file. A voice can help shape how an idea feels.
A human performer brings more than the finished recording
One of the clearest distinctions in the conversation is between generating audio and working with a performer. Mark Graeme talks about recent campaigns where characterful casting added something that could not easily be separated from the creative idea itself. A human voice actor can understand the project around the script, react to direction, respond to feedback in the moment and offer something unexpected.
Melissa makes a similar point from her experience as a performer. A voice session may begin with the assumption that someone will simply come in and read the script, but once the performer, creative team and client begin working together, the session can become collaborative. Choices change. A line is interpreted differently. A performance takes the creative somewhere nobody quite anticipated beforehand.
Mark Eaves agrees that this craft has sometimes been underappreciated in advertising and argues that it needs more space. That distinction matters when thinking about AI. If the task is simply to produce understandable speech, technology may increasingly be able to do that. If the task is to interpret, collaborate and perform, the decision becomes rather different.
Where AI voice starts to make practical sense
Neither conversation dismisses AI voice. In fact, both Mark Graeme and Mark Eaves identify situations where it can be genuinely useful.
For Mark Graeme, one of the clearest is high-volume content. Brands today may need hundreds or even thousands of assets across platforms, countries and languages, while production budgets have not necessarily increased at the same rate.
A premium television commercial, emotionally driven campaign or piece of high-profile creative work may warrant a carefully cast human performer. But when a brand is producing large volumes of instructional videos, product explainers or lower-profile content, a properly licensed AI voice or a model based on a performer's voice, with their permission, may offer a practical way to meet that scale.
Mark Eaves reaches a similar conclusion from a slightly different direction. At Gravity Road, the starting question is always: why are we using AI? Voice technology may let a creative team do things that were previously difficult or impossible, including rapid translation or dynamic voice experiences. What concerns him is AI becoming the default simply because it is quicker, easier or cheaper. That is a very different justification.
It may not be human or AI. We could see more of both.
Mark Eaves makes an important observation about where all this could be heading. There will probably be far more AI-generated content, but at the same time, audiences may increasingly care about knowing when something is real, who made it and where it came from.
So rather than one replacing the other, he expects both tendencies to grow. More synthetic content, and more interest in authenticity. That possibility is particularly relevant to voice because hearing somebody speak can carry a strong sense of identity, presence and trust. As synthetic media becomes more commonplace, knowing that a voice genuinely belongs to the person speaking may become more valuable, not less.
Reuters: when authenticity becomes part of the creative decision
A project Mark Eaves discusses for Reuters shows what that looks like in practice. Gravity Road has been working with Reuters as the organisation develops a stronger direct relationship with consumers, particularly in North America. The positioning centres on receiving news from the source.
That created an interesting decision around voice. Technically, AI could have made it easier to create versions of the campaign in different languages. But for a news organisation whose identity is strongly connected to truth and authenticity, the team decided to use human voices throughout. More specifically, they used the voices of Reuters journalists.
The choice of voice therefore became part of the message itself. The audience was not merely hearing information delivered in a convincing voice. They were hearing people genuinely connected to the organisation producing the journalism. Mark believes this wider story of how creative work is made will become increasingly important. Audiences may not only judge the final piece. They may also care about who created it, how it was created and what choices were made along the way.
Consent has to remain fundamental
Both the opportunities and concerns around AI voice become particularly important when cloning enters the conversation. For Mark Graeme, the principle is straightforward: consent is consent.
It should be built into the production process whether the work is a major public campaign or an internal training video. Cloning or mimicking a voice without permission is not acceptable. The same scrutiny should apply to the tools being used. Creative teams need to understand where training data comes from and how library voices are created.
A piece of content might have a relatively small audience, but that doesn't remove those responsibilities. This matters even more as AI-generated voice moves beyond advertising into training, corporate communications, research material, and other forms of everyday business content.
Ode: using AI to bring people closer to human voices
Perhaps the most interesting example in the episode is Ode, a project developed by Gravity Road, William Sieghart, and Microsoft. Ode is an AI poetry companion. A user speaks to it about how they are feeling, the system asks questions, and it then recommends a poem that might resonate with them.
AI therefore plays an important role in the experience, but a deliberate boundary defines what it does. The poem itself is a real poem written by a human poet, with the appropriate rights cleared. And when the poem is performed, the listener hears a human actor. Those performers include Stephen Fry, Indira Varma and Dominic West.
Mark describes that boundary as one of the project's most important parts. The AI helps the listener find the poem. It does not perform it.
Technology can support human performance without replacing it
Ode also provides a useful example of a less obvious role for AI in audio production. Some existing recordings of Poetry Pharmacy performances had been made at live events and were not of sufficient quality to use. AI audio tools helped clean those recordings and make material usable that might otherwise have been lost.
For Mark Eaves, that is an example of the technology doing something genuinely valuable. AI was not replacing the actor. It was helping preserve and improve a human performance.
The team could have created the entire Ode experience with synthetic voices, and Mark believes it would have taken a fraction of the time. But it would also have changed the experience. The intention was for Ode to use technology to bring somebody closer to a human voice, not further away.
So what does the future look like for human voice?
Both guests ultimately arrive at an optimistic view, although for slightly different reasons.
Mark Graeme believes human voice has a healthy future wherever communication depends on emotion, feeling and high-quality creative work. For premium advertising, he sees the craft of voice performance as something it would be short-sighted to overlook after investing so much thought and planning in the rest of a campaign. But he also sees opportunities for voice actors as technology changes.
With the right permission, remuneration and agreements in place, performers may choose to license or extend the use of their voices across larger volumes of content. Different performers will have different levels of comfort with that. What matters is that they retain a meaningful role in those decisions.
Mark Eaves sees an opportunity for human voice to become more expansive. Historically, voice in advertising has often been confined to a short commercial, a radio spot or the final element added to a campaign. Podcasting has already demonstrated that audiences have an appetite for much richer forms of audio storytelling. Combine that appetite with emerging technology, and you can imagine more immersive, ambitious voice experiences.
Technology may make those experiences possible, but their emotional value may still depend on hearing another human being at their centre.
The question is not simply AI or human
Perhaps the clearest conclusion from these two conversations is that AI voice and human voice are unlikely to settle into a simple winner-and-loser relationship.
AI has clear advantages where production requires scale, speed, localisation or large volumes of functional content. Human performers offer something different: interpretation, spontaneity, emotional connection, collaboration and authenticity.
Sometimes technology may replace a task previously performed by a person. Sometimes it may allow a human voice to be used in entirely new ways. Sometimes the strongest use of AI is to support human creativity rather than imitate it.
The important question for brands, agencies and creatives is therefore less "Can we use AI voice?" and more:
Why are we using it, and is it the right choice for this particular piece of work?
Listen to the full conversation with Mark Graeme and Mark Eaves in this episode of High Notes, the BRAVA podcast on the art and business of voice.
Comments