The Southeast Asian playbook for winning at voice AI

To stay competitive in voice AI, Miles Wen, co-founder and CEO of Fano, trains his AI on a lot of data that few people would recognize as “voice.”

He’s trained his models on data that includes human screaming, dogs barking, glass shattering, and even explosions, datasets with a “wide range of acoustic conditions,” he tells Tech in Asia.

Image credit : Ulla

While this may sound counter-intuitive, Wen says it enables the software to pick up on anomalies in the background, which helps it with one of its primary functions of detecting fraud. For instance, it can tell when a person is trying to change their voice using a voice changer device or if they’re speaking with their mouths covered.

This, Wen says, helps Fano stay ahead of the larger voice AI model providers, such as ElevenLabs, in a market that is becoming one of the hottest verticals in Southeast Asia.

Globally, the industry is projected to see a growth rate of between 10% to 30% by 2031, driven by the explosion of short-form video content, new audio trends in digital publishing, and the automation of global customer service.

Fano is just one of a growing number of voice AI startups across Southeast Asia. Other regional players include Singapore-based Wiz.ai, which specializes in models that are fluent in regional languages, and Indonesia-based Kata.ai, both of which have deployed their voice AI in banks and other financial institutions.

Unlike Fano, they provide more mainstream services, such as telemarketing, credit collection, outbound know-your-customer calls, and answering customers’ questions.

But these startups will also have to contend with players such as ElevenLabs, a US-based AI voice technology company known for its high-fidelity voice cloning and text-to-speech platform. It is the most highly valued startup in voice AI with an US$11 billion valuation.

Other voice AI specialists include Deepgram, UK-based Speechmatics, and France-based Gradium, which came out of stealth late last year with a US$70 million seed round.

Pricing in a hot market

Kata.ai co-founder and CEO Irzan Raditya says he’s seeing first hand the increased demand for voice AI solutions in Southeast Asia this year, especially from the financial services sector, where companies have “real pain points,” such as scaling up operations to handle more customer queries.

Kata.ai’s business revolves mostly around text-based solutions, with voice currently making up less than 10% of its total revenue.

Irzan Raditya, co-founder and CEO of Kata.ai / Photo credit: Tech in Asia

The defensibility question

All-in for data