The Most Natural Falcon is Here
More human prosody. High-fidelity speech. Fastest time-to-first-audio.
Murf Falcon 2 delivers 10x better naturalness than the previous Falcon model and reduces time-to-first-audio
to under 100ms, while retaining the same cost-efficient pricing at 1 cent per minute.
Hear Falcon 2 At Work
Four conversations, start to finish. Hear the code-mixing,
pronunciation, latency and natural delivery hold up across real-world scenarios.
What's Different About Falcon 2
The faster and better voice AI for enterprise agents.
More Natural and Conversational
In blind evaluations, Falcon 2 was preferred 10x more often than the previous Falcon model. It better understands sentence context to deliver speech that feels more conversational, with more human pacing, clearer emphasis, and cleaner output. The side-by-side sample below lets you hear Falcon 2 compared with another TTS provider.
.svg)
Fastest Sub -100ms Time-to-First-Audio
Using apiping.io, a geo-distributed API relay, Falcon 2 was benchmarked against leading real-time TTS models on latency. The tests covered 1,500+ pings from multiple global locations. Falcon 2 ranked #1 across every region, delivering the fastest TTFA and staying consistently under 100ms in each.
New: Instant Voice Cloning
Available for enterprises, Falcon 2 lets you clone a custom brand voice with less than a minute of recording, giving your agents a distinct voice identity built for production use.
.webp)
Improved Multilingual and Code-mixed Delivery
Falcon 2 supports 150+ voices across 35+ languages and delivers mixed-language text with smoother transitions across supported languages, making it better suited for real-world multilingual voice agents.
What Makes Falcon 2 More Natural
Falcon 2 builds on Falcon's lightweight real-time architecture and is optimized for
deeper context understanding, more natural prosody, and higher-fidelity speech delivery.
.svg){kind=icon}
Deeper Semantic Understanding
An advanced semantic layer reads sentence meaning at a finer grain, helping Falcon 2 shape pacing, emphasis, and enunciation around the context of what is being said.
Expanded Conversational Training
Trained on real customer calls, support flows, and booking conversations, Falcon 2 delivers speech that sounds closer to live interaction than scripted narration.
.svg){kind=icon}
Advanced Text Normalization
Improved text normalization helps Falcon 2 pronounce and render real-world details more accurately, including pincodes, numbers, currency, addresses, dates, and vehicle numbers.
Real-Time TTS Without the Trade-Offs
Until now, every voice stack forced builders to compromise on latency, naturalness or cost. Falcon 2 breaks that
cycle by bringing together human-like delivery, sub-100ms time-to-first-audio, and 1 cent per minute pricing.
Enterprise Ready Deployment
Falcon 2 lets you scale and deploy without limits.
.webp)
10,000 Concurrent Calls with Stable Latency
Falcon’s efficient architecture supports up to 10,000 concurrent calls without compromising latency.
.webp)
Data Residency in 11 Geographies
With data residency in 11 geographies, Falcon ensures your data stays local and secure. Edge deployment also ensures consistent latency everywhere you operate.
On-Premise Deployment
Falcon is engineered for flexibility, supporting on-premise deployment for enterprises that need full control and security.
Multi-Layered
Security Infrastructure
Ready to Build with Falcon 2 ?
Falcon 2 brings more natural speech, sub-100ms time-to-first-audio, enterprise voice
cloning, and 1 cent per minute pricing to production voice agents.
