Everyone's shipping voice agents this year. Receptionists, collections bots, lead qualifiers, support lines. The demos look great.
Then they hit a real phone call.
I lead growth at a voice AI company, so I sit in on a lot of these conversations. And the thing that surprised me is how rarely the LLM is the problem. It's almost always the voice layer underneath it. The latency that was 90ms in the demo, suddenly climbs to 400ms on a real phone call. The voice that sounds great in English and falls apart the moment someone switches to another language mid-sentence. The TTS that reads a pincode like a phone number. Or the per-minute bill that was fine at 10,000 minutes and became the biggest line item at a million.
Murf AI
Hey Product Hunt 👋 I'm Ankur, co-founder and CEO of Murf AI.
Here's the thing about real-time text-to-speech: you almost always end up trading something away. Want it to sound natural? Latency goes up. Want it fast? Your bill goes up. Want it cheap? It starts sounding like a robot reading a phone menu.
We've been building voice AI for six years now, and honestly, every stack we've come across makes you pay for it somewhere.
So we built Falcon 2 to stop making people choose. It's our fastest and most natural TTS model yet, and it runs at $0.01/minute.
A few things I'm proud of:
It sounds real. Falcon 2 ranks among the top real-time TTS models in independent benchmarks across several platforms. Being fast was the easy part for us — the hard part was staying fast without it sounding fast.
The speed holds up everywhere. We ran 1,500+ identical streaming requests from 32 edge locations and measured time-to-first-byte. Falcon 2 came back under 100ms in almost every region — 68ms in Mumbai, 74ms in London and Ohio.
It's actually multilingual. 150+ voices, 35+ languages, and it handles mixed-language text without the awkward stumble you usually get when switching mid-sentence. That matters a lot if you're building voice agents for real users.
Instant voice cloning (for our enterprise folks). Clone a brand voice from just 10 seconds of audio — always with the speaker's explicit consent — so your agents get a consistent identity you can ship.
Built to scale. Up to 10,000 concurrent calls, data residency across 11 regions, and on-prem if your team needs it.
And all of that is roughly 1/5th the cost of comparable models. One cent a minute.
Sign up today and you'll get 1 million free characters, credited every month. We've also got live agents running on Falcon 2 — go talk to them and tell me what you think.
Thanks for checking us out. I'll be hanging around in the comments today, so hit me with questions and feedback.
Murf AI
Hey PH Community! I'm Sneha, Co-founder and COO of @Murf AI. Really excited to get your feedback on the Falcon 2 launch. Do give it a try, and I'll be around all day to answer anything - pricing, roadmap, languages, whatever's on your mind.