Voices.com Promoting AI Voices Over Human Ones

Voices.com participated in a 2019 Voicebot.ai survey that said 71% preferred a human voice-over AI. Yes, AI voices have come a long way since 2019, but even the best ones are still annoying, even in brief samples. Voices.com did their own survey of 336 registered client users of Voices.com which came out in April, 2023. 60% of respondents said they haven’t used AI voices and 30% said they were still “too robotic”. Now, Voices.com has just acquired Voices.ai. In the press release announcing this acquisition, Voices.com CEO David Ciccarelli acknowledges the need for a human voice-over for eLearning, with a condition:

“While there will always be a place for voice actors who bring their unique interpretation to a script and create character voices, there are many applications that don’t require artistic interpretation,”

Then the press release says that most industrial videos are best suited to AI voices, and eLearning falls in that category. That statement is immediately followed by this quote from Ciccarelli:

“The reality is that many applications would benefit from AI generated voices in real-time rather than rely on recordings done by voice talent. Especially when the content changes frequently, having access to voices created instantly offers a benefit not previously realized.“

Human Voice-Over More Reliable than Legacy AI

I understand he has a service to sell here, but there is no such thing as “voices created instantly”. In Voices.com “AI in the Creative Industry” report, half of respondents cited a need for instantly downloaded voices as the most compelling reason to use AI voices. But a developer still has to generate the right voice for the project. It will need editing, have the voice approved. Then edit and process it into the eLearning program.

If you need me, a human voice-over for industrials or eLearning, to re-record a sentence or paragraph, it’s no problem. I have the original recording sessions saved for at least two years. I can instantly match the style and recording settings of the section that needs updating and deliver it to you in fifteen minutes. That is, if I’m not booked at the time. If I am, I will definitely deliver it to you in a few hours at the latest.

However, here is an important question to those of you who might still prioritize speed over a human voice-over for eLearning. Can you guarantee that your AI will be able to provide the same voice two years from now, considering the way technology is changing? Perhaps, but I wouldn’t count on it. You might be able to have better AI voices and have to redo the whole module to have it match. There is no guarantee whatsoever that AI providers will maintain support for legacy voices. However, I guarantee that I can match the voice of my previously recorded eLearning modules, and do it quickly and affordably. And, I guarantee that the spiffy AI voice that you have convinced yourself sounds “almost human” today will sound dated in two years.