Voice AI startups are increasingly competing to handle enterprise calls in sales, marketing, and support. Large organisations are offloading these tasks to model developers like ElevenLabs and Deepgram, infrastructure providers such as Vapi, Retell, and LiveKit, and dedicated support shops including Decagon and Sierra.
Rime, based in San Francisco, aims to stand out in this crowded field. The company trains its voice AI models on conversational data it records itself. This strategy aims to reduce the customisation burden for its clients.
Founded in 2022 by Lily Clifford, Brooke Larson, and Ares Geovanos, the team includes a former Stanford PhD student, an ex-Amazon Alexa engineer, and a Stanford engineer. Rime has built a recording studio in San Francisco to gather its own audio. It does not rely on scraping the web.
The startup tunes its voice models to handle specific brand names and industry terms correctly. It uses a phoneme-based architecture to adapt to different pronunciations. This means customers do not need to retrain models for their specific sector.
Rime announced on Wednesday that it has raised $24 million in a Series A round. M13 Ventures led the investment. Twilio Ventures, Corazon Capital, Unusual Ventures, and other existing investors also participated.
Clifford noted that despite progress in voice AI, enterprises still prefer legacy IVR systems. AI voice technology cannot yet match their effectiveness.
“The voice technology is still not there to automate the vast majority of enterprise phone calls,” Clifford said. “LLMs have made it a lot easier to build voice applications that work, but they haven’t changed how it feels to interact. Talking with a voice AI agent is not the most compelling experience for the end user. It’s kinda like a new IVR, but with a better voice.”
The company originally used a pipeline of separate models for speech-to-text, text-to-speech, and a large language model. It is now shifting focus to develop better speech-to-speech models. This change aims to reduce latency, improve turn-taking, and handle background noise. The new approach will also decrease reliance on complex orchestration, so the company does not have to manage many different models.
Rime reports customers in food service, healthcare, airlines, and fintech. The company claims its training data and model positioning result in longer call durations. This has helped it win enterprise contracts from Mayo Clinic, Dialpad, Upstart, and Asurion.
With the new funding, Rime plans to expand its team of 35 people. Hiring will focus on model development, engineering, and partnerships. The company recently appointed Rafael Valle as Chief Scientist. He previously worked on audio understanding at Meta Superintelligence Labs and NVIDIA’s applied deep learning audio research team.
“Companies like ElevenLabs have moved into being an orchestration and the application layer, going head to head with the Sierras and Decagons of the world,” M13’s Morgan Blumberg told TechCrunch. “I think there’s just so much more to be done technically, and Rime’s approach of pushing forward on the best model with low latency and high reliability in a regulated environment stands out.”
The startup previously raised $5.5 million in a seed round last May. Blumberg is joining the board as part of the funding round.
What it means
For businesses relying on phone support, the shift from stitching together separate models to a unified speech-to-speech approach could reduce latency and make interactions feel more natural. However, the current reality remains that voice AI still struggles to fully replace legacy IVR systems, as the user experience lacks the necessary polish for widespread adoption.
