Interview

How did you end up in voice?

I studied AI at Carnegie Mellon, which is where I got inspired to take up voice research and applications, largely by looking at the great research and work coming out of WavLab, led by Shinji Watanabe. From there, I went on to work at Apple's Siri group, at a time when it was being led by Ian Goodfellow.

Safe to say, my core academic background and early professional experience launched me very deep into speech synthesis and voice AI as a whole.

How should someone start in voice?

I think it starts by asking the right questions on what makes you curious about Voice AI. Why do you care about Voice AI? Is it the thrill of building voice-first applications? If so, a great way to begin is to build or prototype those applications.

There are excellent open-source and closed-source providers today that can be leveraged to create very solid prototypes. As you go deeper, the natural limitations of Voice AI may motivate you to move in a more technical, research-first direction, solving problems in prosody, pronunciation, and more. The best place to begin is to build.

Where do you think voice is going?

All text will become voice. I do not believe typing or reading will remain the primary medium of the future. We will interact with computers the way we interact with each other: through voice.

If voice is to become the primary interface for "everything," then its future is essentially being omnipresent. At CAMB, we're focused on building small voice models that can fit anywhere, be deployed everywhere, and ultimately be available to all 8 billion people.