←
Murf AI573 posts
Murf AI avatar

Murf AI

@MurfAIStudio

Introducing Murf Falcon - the fastest, most efficient TTS built for Voice Agents.

◷ Joined August 2020▧ 225 Photos and videos
Murf AI avatar
Murf AI@MurfAIStudio·20 Nov 2025

We’re proud to launch Murf Falcon: the fastest, most efficient text-to-speech API, built for the next generation of voice-first experiences. Here’s what sets Falcon apart: - Fastest in production: 55 ms model latency and fastest 130 ms time-to-first-audio, consistent across geographies - Extensive language support: 35+ languages with expressive, natural voices - Best in code mixing: Fluent and accurate multilingual capability - Compliance-ready: Data residency across 10+ regions for privacy and regulatory needs - Flexible deployment: On-premise support for enterprise environments All at just 1 cent per minute. Try Falcon today: murf.ai/falcon

2:15
Murf AI avatar
Murf AI@MurfAIStudio·Nov 13

Falcon 2.1 is #3 among real-time Hindi TTS models on Voice Arena. Hindi brings its own challenges in pronunciation, phonetic variation and natural speech delivery, and we're glad to be making giant strides! More to come soon. Shoutout to the folks at @voicearena_ai for these independent benchmarks that let us measure how our models perform against real listener preferences, identify areas for improvement, and keep raising the bar on speech quality. #murffalcon2 #voiceai

Murf AI's Falcon 2.1 enters the Voice Arena Hindi 🇮🇳 TTS Leaderboard at #6 of 22 with 1065 Elo, seven places and 52 Elo above Falcon 2. Falcon 2.1 is Murf's text-to-speech model built for real-time conversational AI, and it ranks #3 of 11 among real-time models on the Hindi board. Every ranking on Voice Arena comes from blind, head-to-head votes by vetted native speakers, scored with Bradley-Terry Elo. Results backed by 770 blind, head-to-head listener votes from native Hindi speakers. We're evaluating Falcon 2.1 on our other boards too. Results will follow soon. Congrats to the whole @MurfAIStudio team 👏 Explore the board → voicearena.com/tts-leaderboa…

Murf AI avatar
Murf AI@MurfAIStudio·Nov 11

Got an idea for a voice-powered application? It's time to build it. Build with @AgoraIO is bringing together 250+ builders for a one-day hackathon focused on creating real-time AI applications. We at Murf AI are the Exclusive Partners, bringing voice technology into the build. Experiment with the most natural, real-time AI voices, create conversational experiences and put voice at the centre of your application. Build alongside fellow developers, get expert mentorship and compete for $3,000 in cash prizes and gadgets. Happening on 21st November. Register now: luma.com/2g5vcgrr #buildwithagora #murffalcon2 #voiceai #hackathon

Post attachment
Murf AI avatar
Murf AI@MurfAIStudio·Nov 11

We're going live today! Join us for the first Murf Voice Circle Masterclass led by Gaurav Singh. We'll get into the engineering decisions behind production voice agents: turn-taking, interruptions, prompt caching, tool calls, latency, infrastructure and reliability. If you're building voice agents or working on conversational AI applications, this session is for you. Register Now: luma.com/falcon-83vx #murfvoicecircle #voiceai #community #agents

Post attachment
Murf AI avatar
Murf AI@MurfAIStudio·Oct 28

Falcon 2 is now faster with improved accuracy. We’re constantly improving our models, refining everything from how speech is generated to how quickly it reaches the listener. In streaming voice, the first audio bytes don’t always mean the first sound. Models can return audio that begins with leading silence, meaning the listener waits longer than a basic latency measurement suggests. Reducing that gap is key to making voice interactions feel more responsive. With Falcon 2, model training and refinements to audio delivery have helped reduce leading silence, bringing median time to first audio down to 131 ms. These improvements focus on how quickly the listener actually hears speech. Infrastructure optimizations have also helped limit latency spikes during peak demand, supporting more consistent response times. Falcon 2 also delivers a 1.5% word error rate, placing it among the most accurate voice models evaluated. These results are based on data from the last seven days and include p95 latency. Shoutout to the team at @covaldev for the benchmarks and for digging into what goes into measuring real-world voice performance. Their measurement of leading silence helped us refine and improve our models.

Post attachmentPost attachment
Indian Data Club avatar
⟳ Murf AI retweeted
Indian Data Club@indiandataclub·Oct 23

We’re happy to collaborate with @MurfAIStudio for our upcoming Hyderabad event, The Model Wars. Join us for conversations around AI models, real-world applications, and what’s actually working in today’s rapidly evolving AI landscape. 🎟️ Register: luma.com/54iu9u3t

Post attachment
Murf AI avatar
Murf AI@MurfAIStudio·Oct 22

A big step forward for Falcon 2! We're hard at work to make natural voice generation faster and more responsive, powering real-time voice experiences for businesses across the world. Thanks to the folks at Voice Arena for putting our latest progress to the test. More updates coming soon as we continue to push the boundaries of Voice AI. #murffalcon2 #voiceai

Murf AI's Falcon 2 moves from #11 to #2 of 14 on the Voice Arena US English 🇺🇸 latency board, with 102 ms to first audio at the median. That is more than five times faster than its own earlier 528 ms, and puts Falcon 2 ahead of SpeechifyAI's Simba 3.2 at 123 ms and Inworld TTS 2 at 168.5 ms. We measure time to first audio the way a listener hears it: leading silence counts, and no exact text is ever sent twice. We're also evaluating Murf's latest Falcon 2 release in US English and Hindi, with results coming soon. Congratulations to @MurfAIStudio on the update! Explore the board → voicearena.com/tts-leaderboa…

Voice Arena avatar
⟳ Murf AI retweeted
Voice Arena@voicearena_ai·Oct 22

Murf AI's Falcon 2 moves from #11 to #2 of 14 on the Voice Arena US English 🇺🇸 latency board, with 102 ms to first audio at the median. That is more than five times faster than its own earlier 528 ms, and puts Falcon 2 ahead of SpeechifyAI's Simba 3.2 at 123 ms and Inworld TTS 2 at 168.5 ms. We measure time to first audio the way a listener hears it: leading silence counts, and no exact text is ever sent twice. We're also evaluating Murf's latest Falcon 2 release in US English and Hindi, with results coming soon. Congratulations to @MurfAIStudio on the update! Explore the board → voicearena.com/tts-leaderboa…

Murf AI avatar
Murf AI@MurfAIStudio·Oct 15

Congratulations to our partners at @revragai, on raising $2.5 million in seed funding! RevRag AI is building multilingual AI agents that act as virtual relationship managers within financial services apps, helping customers navigate onboarding, understand financial products and access support. Congratulations to Ashutosh Singh, Pankaj Gupta, Neeraj Gupta, Swapnil Yadav and the entire RevRag AI team. We're glad to be part of your journey and look forward to seeing what you build next! #voiceai

Murf AI avatar
Murf AI@MurfAIStudio·Oct 9

Day 2 at @Analyticsindiam Cypher 2026 and things were electric! Our co-founder Sneha Roy took the stage today to talk about building Voice AI from India, for India, and for the world, and the room was buzzing with ideas and incredible questions. Back at Booth No. 44, our retro telephone booth kept ringing, with plenty of people stopping by to experience the magic of Falcon 2 for themselves. But we're not done yet! One more day of Cypher remains and we'll be there. Come pick up the phone. #voiceai #murffalcon2 #cypher2026

Post attachmentPost attachmentPost attachmentPost attachment
Murf AI avatar
Murf AI@MurfAIStudio·Oct 6

Our co-founder Sneha Roy is part of this year's @BwDisrupt 40 Under 40 and this accompanying magazine feature explores Sneha’s journey at Murf and the challenges of building voice AI that works across languages, accents, latency constraints and global markets. Here’s to the journey, and to everything still to build. @BWBusinessworld #voiceai #murffalcon2

Post attachment
Murf AI avatar
Murf AI@MurfAIStudio·Oct 6

Something a little unexpected is waiting at @Analyticsindiam Cypher 2026. We’re bringing the Murf telephone booth to Booth No. 44, where you can step in, pick up the phone and experience Voice AI like you haven't before. Come say hello, meet the team and explore what we’re building. #murffalcon2 #voiceai #cypher2026

Post attachment
Murf AI avatar
Murf AI@MurfAIStudio·Sep 17

All of that comes with the speed and naturalness required for real-time voice experiences, including sub-100ms time-to-first-audio. Because the voice can be natural, the response can be fast, and the data can still stay where it needs to. #voiceai #murffalcon2 #enterpriseai

Murf AI avatar
Murf AI@MurfAIStudio·Sep 17

For enterprises, these are fundamental requirements, especially in regulated industries, where voice AI may handle sensitive customer information or proprietary business data. That is why Falcon 2 gives enterprises the flexibility to deploy voice AI while keeping complete control over where their data is processed.

Murf AI avatar
Murf AI@MurfAIStudio·Sep 17

A voice AI model can be fast, natural and incredibly capable. But for an enterprise, that is only half the equation. The other half is control. Where is the voice data processed, and who controls the infrastructure it runs on?