Hello Betamax, Have you been to a party entirely soundtracked by AI-generated music? I wasn’t quite sure what to expect when I spotted it on the agenda for last week’s ElevenLabs Summit in Bengaluru. But it turned out to be not that bad. The London-based AI voice startup hosted its post-event party on the terrace of one of Bengaluru’s tallest skyscrapers, with free-flowing wine and beer for close to 1,000 attendees. Fittingly enough, ElevenLabs has had plenty of reasons to celebrate. The company raised US$500 million in February, and its top leadership arrived in India fresh off a US$300 million employee tender offer that doubled its valuation to US$22 billion. It has also just opened an office in Singapore. But of all the places to throw a party of this scale, India is a particularly interesting choice.
Image credit: Made by Ulla/Tech in Asia with the help of AI
The country is a linguistic maze. I have long wondered how AI, whose most powerful models are built largely around English and Chinese, would fare here. CEO Mati Staniszewski believes that if ElevenLabs can crack “how people interact with AI” in India, it can “solve it anywhere in the world,” pointing to the country’s 22 officially recognized languages and thousands of dialects and accents. The company plans to invest “hundreds of millions of dollars” here as well, Reuters reported. India is already ElevenLabs’ largest market outside the US, according to Staniszewski. The company now wants to expand its support from 14 Indian languages to all 22 officially recognized languages in the country. Karthik Rajaram, ElevenLabs’ India head, told me the goal is to also tackle the country’s countless dialects. For now, the company appears to have a head start. ElevenLabs said it has signed up more than 250 enterprises. Several developers at the event told me they use multiple AI voice models but found ElevenLabs the most consistent, especially for local languages. That’s pretty much what I’ve heard, too. Over the past three months, I have spoken to several executives in the voice AI industry, and ElevenLabs has almost always been among the first names to come up in conversations, alongside competitors such as Sarvam, Gnani.ai, and Cartesia. ElevenLabs’ latest v4 model is intended to push that advantage further. On a benchmark published by Humyn Labs, even ElevenLabs’ older v2 model ranks ahead of competing models across several Indic languages. “They have the first-mover advantage in India,” a voice AI founder told me at the event. “They started collecting data and improving their models before many of the local players.” That advantage aside – and to the firm’s credit – the summit steered clear of the usual narrative about AI replacing human workers. At most other voice AI events I’ve attended, people boast about how many calls an AI agent can handle. Instead, ElevenLabs set up stalls where attendees could clone their voices in 20 seconds and talk to their AI twins. They could also snap a few pictures, turn them into reels, or make their own music using ElevenMusic. I can’t say whether ElevenLabs will successfully crack the Indian market. But at least I had fun. Meanwhile, here are some of our recent AI coverage worth checking out:
Pranav Balakrishnan, AI reporter
Catch up on past issues of the newsletter here. |
