I still remember the first time I tried to clone my own voice. It was a few years ago, and the technology was still in its infancy. I spent hours recording myself, reading endless paragraphs of text, and the result was… well, let’s just say it was less “me” and more “a robot who had a very bad day.” The pitch was wrong, the cadence was off, and it had this weird, metallic echo that made my skin crawl. My wife said it sounded like a ghost in a machine, and she wasn’t wrong.
That initial failure could have been the end of my journey with AI voice. As an entrepreneur and investor, I’m constantly looking for the next big thing, but I’m also a pragmatist. If a technology doesn’t work, I’m not one to waste my time. But something about AI voice stuck with me. I couldn’t shake the feeling that we were on the cusp of something big, something that could change the way we interact with technology forever.
The Early Days of Robotic Voices
My first real foray into the world of AI voice wasn’t as an investor, but as a frustrated user. I was trying to build an automated podcasting tool, something that could take a written article and turn it into a natural-sounding audio segment. The idea was simple, but the execution was a nightmare. The available text-to-speech (TTS) engines were clunky, expensive, and the voices they produced were so robotic that they were almost comical.
I remember one particular incident where we were trying to generate a voiceover for an article about a new programming language. The AI voice pronounced “Python” as “Pith-on,” and no matter what we did, we couldn’t get it to say the word correctly. We spent days trying to fix it, tweaking the pronunciation guides, and even trying to spell the word phonetically. In the end, we had to manually record the word and splice it into the audio. It was a painful reminder of how far the technology still had to go.
But even in those early days, I saw glimpses of the future. I met with a small startup in a cramped office in Palo Alto that was working on a new approach to voice synthesis. They were using a deep learning model to generate voices from scratch, and the results were unlike anything I had ever heard before. The voices were still a bit rough around the edges, but they had a naturalness and a warmth that was missing from the existing technology. I invested in that company on the spot, and it was one of the best decisions I’ve ever made.
The Turning Point: When AI Found Its Voice
That little startup, which is now a major player in the AI voice space, was my first real taste of what was possible. They were one of the first companies to crack the code of natural-sounding AI voice, and they did it by focusing on one thing: data. They had amassed a massive dataset of high-quality voice recordings, and they used it to train a model that could generate voices that were almost indistinguishable from a human’s.
This was a revelation for me. I had always thought of AI as being about algorithms and processing power, but I was wrong. It’s about data. The more high-quality data you have, the better your model will be. It’s a simple concept, but it’s one that many people in the AI world still don’t fully grasp.
Another key learning for me was the importance of the “uncanny valley.” This is a term that’s often used in robotics and computer graphics to describe the feeling of unease that people experience when they encounter a robot or an animation that is almost, but not quite, human. The same thing happens with AI voices. If a voice is too perfect, too polished, it can sound fake and untrustworthy. The best AI voices are the ones that have a little bit of imperfection, a little bit of humanity.
My Adventures in Voice Cloning
After my initial failed attempt, I decided to give voice cloning another try. This time, I used a much more sophisticated tool, one that was powered by the same technology as the startup I had invested in. The process was still a bit tedious—I had to record myself for about an hour—but the result was mind-blowing. It was my voice, but it was also… more. It was a perfect, idealized version of my voice, without any of the stumbles, pauses, or imperfections of my natural speech.
I started using my cloned voice for all sorts of things. I used it to record the audiobooks for my books, to create personalized video messages for my friends and family, and even to order a pizza. The pizza part was a bit of a disaster—the AI voice couldn’t understand the guy on the other end of the line—but it was a fun experiment nonetheless.
But my adventures in voice cloning also opened my eyes to the potential for misuse. I realized that this technology could be used to create fake audio recordings, to spread misinformation, and to impersonate people without their consent. It’s a scary thought, and it’s one that we as a society need to grapple with as this technology becomes more widespread.
The Future is a Conversation
So, where is all of this heading? I believe that we are on the verge of a new era of computing, one that is built on the foundation of AI voice. In the near future, I believe that we will all have our own personal AI assistants, ones that we can talk to and interact with in a natural, conversational way. These assistants will be more than just a voice on a speaker; they will be our companions, our collaborators, and our guides to the digital world.
I’m putting my money where my mouth is. I’ve invested in over a dozen companies that are working on different aspects of this future, from the hardware that will power our AI assistants to the software that will give them their personalities. It’s a risky bet, but it’s one that I’m willing to take. Because I believe that the future of technology is not about screens and keyboards; it’s about conversations.
And who knows, maybe one day my AI assistant will be able to order a pizza without any help from me. A guy can dream, right?
Frequently Asked Questions
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.