Bengaluru start-up's new AI voice model outperforms rival from OpenAI
What's the story
Bengaluru-based start-up, Murf AI, has launched a new artificial intelligence (AI) voice model called Falcon 2. The innovative text-to-speech system is publicly available from August 20. It has already outperformed some platforms from industry giants like OpenAI and ElevenLabs in benchmarks on Artificial Analysis, an independent platform that tracks AI performance.
Economic advantage
Falcon 2 comes at a competitive price
Falcon 2 comes at a competitive price of $0.01 per generated minute, making it a cost-effective choice for businesses like call centers, banks, and airlines looking to handle large volumes of customer communications.
The model is also faster than many competitors, responding in under 100 milliseconds.
This combination of speed and affordability makes Falcon 2 an attractive option for conversational AI and real-time voice agents.
Clientele
Murf AI has over 10M users
Founded in 2020 by IIT Kharagpur alumni Sneha Roy, Ankur Edkie, and Divyanshu Pandey, Murf AI is backed by Matrix Partners and Elevation Capital.
The start-up already has over 10 million users across 300 companies, including Honeywell International, Cisco, Pfizer, VMware, Nestle SA and Air France-KLM.
Falcon 2 was first released to select clients before its wider launch.
Advanced features
Falcon 2 trained on real-world interactions
Falcon 2's training involved real-world interactions like live customer calls and booking flows, not just scripted narration.
It can understand the meaning of a sentence while adjusting pace, emphasis, and pronunciation accordingly.
The model also improves voice delivery on details like currencies, dates, addresses, and license plate numbers that often confuse many AI systems.
Ethical practices
Commitment to ethical practices
Murf AI has stressed its commitment to ethical practices by working with real voice actors who consent to their voices being used.
The company also offers transparent contracts, ongoing royalties, and data protection.
For added security, it will provide cloud support for data storage in 11 geographies and deployment within in-house clouds.