tools
Rime Mist
Rime Mist is an API-based text-to-speech model for real-time voice agents, customer calls, IVR systems, and other spoken applications.
Rime Mist converts written text into conversational speech for voice applications. It supports streaming audio, pronunciation controls, speed adjustment, spelling, multiple languages, and deployment through cloud or self-hosted environments.
It is used by developers and enterprises building customer support agents, healthcare systems, food-ordering services, financial applications, and outbound calling workflows. Mist is not a general-purpose chatbot or text-generation tool; usage is billed by characters, while enterprise deployment and support are custom-priced.
Features
- Mist v3 text-to-speech model for low-latency voice applications
- HTTP and WebSockets streaming
- Inline pronunciation control and automatic text normalization
- Explicit character-by-character spelling
- Continuous playback speed adjustment
- Custom pauses and controllable phrasing intervals
- Cloud, VPC, and on-premises deployment
- Framework integrations for LiveKit, Pipecat, and Twilio
Use cases
- Build real-time customer support voice agents
- Create healthcare patient and member engagement calls
- Automate restaurant and fast-food ordering
- Read account IDs, addresses, dates, and phone numbers accurately
- Run multilingual IVR and outbound calling systems
- Add spoken responses to financial and hospitality applications
Pros
Cons
Latest updates
- Introducing Prism: the first composable speech-to-speech model
End the tradeoff between control and quality with a speech-to-speech model built for enterprise use.
- Coda v2: Getting every word right (v2)
Phoneme control for Rime’s voice, in six languages, checked before it ever reaches a call.
- Announcing speech-to-text: Rime is now listening
Speech-to-text from the team behind your agent's voice.
- Product Updates
Model update with native coverage of natural sounding number sequences, word spelling, and pausing; users can add custom pauses and pronunciations.
- We've gone live with a model update and some new tools for tailoring your speech!
Native coverage of number sequences, word spelling, and pausing.
Capabilities
- Text to speech — “Today we're releasing Mist v3 , our fastest and most efficient text-to-speech model to date, built for the latency and throughput demands of enterprise production voice deployments.” source
- Voice cloning — “Unlimited custom TTS voice clones” source
- Many languages — “Real fluency across 50+ languages, plus regional dialects and accents — with the same pronunciation controls to get names and terms right every time.” source
- API — “Yes. Drop Rime in through a simple API, or self-host when compliance demands it.” source
Get it
Security
- SOC 2 Type II — “As of May 2025, we are SOC 2 Type 2 compliant .” source
- SOC 2 — “As of May 2025, we are SOC 2 Type 2 compliant .” source
- GDPR — “work through MSAs, BAAs, and data processing agreements as part of enterprise onboarding” source
- HIPAA — “As of February 2024, we are HIPAA compliant .” source
- Not trained on your data — “We never use customer text or audio to train our models” source
Pricing
- Prices checked
- 2026-10-25
- Read
- by web search
Pay as you go - Coda v2
- $0.05 / minute
- Most natural expression TTS
- Metered by characters (~1K per minute)
Pay as you go - Mist v3
- $0.03 / minute
- Lowest latency TTS
- Metered by characters (~1K per minute)
Pay as you go - Rime STT
- $0.002 / minute
- Speech to text, free until Nov 19 2026
Enterprise
Price on request
- Volume rates, annual commitment
- Cloud, VPC, or on-prem
- SLAs, dedicated support
- BAA (HIPAA), SOC 2 Type II
- Provisioned concurrency, custom voices