Rime AI makes voice models for real-time conversations. Its models turn text into speech, transcribe phone audio, and handle speech-to-speech conversations. The site says its voice library has 600+ voices and 50+ languages, and that its Mist model starts sending audio in under 100 milliseconds.
The product line has three parts. Prism is a speech-to-speech model that listens to a call, reasons over it, and replies. It generates text before audio, so guardrails and disclosures can be checked before a caller hears them. Coda and Mist are text-to-speech models with control over accent, pace and tone, and custom pronunciation for names and terms. Rime STT is speech-to-text for live calls.
The page aims the product at businesses that run voice agents, including banking, healthcare, hospitality and food ordering. Teams connect through an API and can use frameworks such as LiveKit and Pipecat. Rime cloud is the default deployment. Enterprise plans add VPC and on-prem options, and the site states HIPAA BAA and SOC 2 Type II for that tier.
Pricing starts with a free plan that includes $50 of usage a month. The Flex plan is billed per minute with no monthly fee. The Scale plan costs $100 a month and lists rates 20% below Flex. Enterprise pricing is custom and invoiced annually. On the Flex plan, Prism is listed at $0.080 per minute, and speech-to-text is listed at $0.002 per minute and free until November 19.
