Rime Labs
Low-latency text-to-speech for phone agents, built by linguists
Overview
Rime Labs is a San Francisco text-to-speech company whose Coda and Mist v3 models are built for real-time voice agents and contact centres. Coda serves ten languages through the cloud API, and both models can be self-hosted on premises or in a VPC. Pricing starts at $0.03 per 1,000 characters, with custom Enterprise plans.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 19 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Mist v3 $0.03 per 1K characters
- Coda $0.05 per 1K characters
- 20 concurrent TTS generations
- Public Slack support
- Unlimited concurrent TTS generations
- Unlimited custom TTS voice clones
- SLAs and dedicated support
- Cloud, on-prem or VPC
- BAA (HIPAA) and SOC 2 reports
Use cases
What you can produce with Rime Labs
- Streaming speech for live voice agents over HTTP and WebSockets, around 100 ms to first audio
- Coda voices in ten languages through the cloud API, plus faster Mist v3 voices in four
- Word-level timestamps for interruption handling (Coda)
- Careful reading of codes, IDs, numbers and names, with spelling and pronunciation controls
- Self-hosted deployment on premises or in a VPC via Docker Compose or Kubernetes
- Ready integrations for LiveKit, Pipecat, Twilio and a hosted MCP server
- Enterprise voice clones, SLAs, a HIPAA BAA and SOC 2 Type II reports
ASEAN Perspective
Rime Labs in Southeast Asia
Rime Labs is US-headquartered and English-first, with no current support for Malay, Indonesian, Thai, Vietnamese, or other core ASEAN languages. Japanese is supported via the Coda and Arcana models, making Rime viable for Japan-facing voice agent deployments. ASEAN enterprises in English-heavy verticals such as BPO, international customer service, or hospitality technology can benefit from Rime's conversational realism and low latency, but teams requiring native-language support across Southeast Asia will need to supplement or choose a broader multilingual provider.
Rime is a strong choice for English-first voice agents on the phone: Coda and Mist v3 stream fast enough for live conversation, handle codes and numbers carefully, and can run inside your own network with SOC 2 Type II reports and a HIPAA BAA. Per-character pricing from $0.03 per 1,000 characters is easy to model, and Dialpad's August 2026 choice of Rime is a meaningful reference.
The gaps are language and consistency. Coda covers ten languages but no Mandarin, Korean, Malay or Indonesian, which limits it for ASEAN deployments, and Rime's own pages disagree on free minutes, voice counts and language totals, so confirm figures with sales before you budget.
What we found
Rime Labs is a San Francisco text-to-speech company set up in 2022 by Lily Clifford, Brooke Larson and Ares Geovanos, a team built around linguists rather than audio engineers. Its pitch is voices trained on recordings of real, unscripted conversation captured in its own studio, aimed at phone agents where a voice has to read back a confirmation code or a refund amount correctly the first time.
The model line-up has narrowed. Coda, released in May 2026, is the flagship and Mist v3 the faster, cheaper option. Arcana was retired from the cloud API on 15 August 2026, with requests now served by Coda, so older descriptions of Arcana no longer describe anything Rime sells in the cloud. Rime's docs list ten languages for Coda (English, Arabic, Dutch, French, German, Hindi, Italian, Japanese, Portuguese and Spanish) and four for Mist v3; Mandarin, Korean, Malay and Indonesian are absent. Its pricing page gives a median time to first audio of 96 ms for Coda and 37 ms for Mist v3.
Pricing is simple: $0.03 per 1,000 characters for Mist v3 and $0.05 for Coda, with custom volume pricing on Enterprise, which adds unlimited concurrency and voice clones, on-premises or VPC deployment, a HIPAA BAA and SOC 2 Type II reports. Rime's own pages disagree on details: the free allowance appears as about 800 minutes and as 3,000, the voice count as 184, 313 and 600-plus, and languages as 8, 10 and 50-plus.
The company raised a $24 million Series A in July 2026, led by M13 with Twilio Ventures taking part, and told TechCrunch it had 35 staff. In August Dialpad named Rime the voice behind its agentic products. A September benchmark that Rime ran itself found listeners preferred it over four rivals on support calls, and it published its losses to ElevenLabs and Deepgram on narration. That is vendor research, but candid.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Notable facts
- Rime built its own in-house recording studio to capture a proprietary dataset of spontaneous, full-duplex conversational speech — including interruptions, laughter, and verbal stumbles — rather than relying on read-aloud recordings like most TTS providers.
- CEO Lily Clifford dropped out of a Stanford NLP PhD program to co-found Rime with a PhD linguist who previously worked on Amazon Alexa.
- The Arcana model can infer emotion from context and spontaneously produce laughter, sighs, audible breathing, and verbal false-starts without explicit markup — a capability the team traces directly to their sociolinguistics-grounded training approach.
- Rime is the only next-generation voice AI provider (as of mid-2026) that offers fully on-premises deployment, a critical differentiator for healthcare and government customers with strict data-residency requirements.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including Rime Labs's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Rime Labs unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Rime Labs directly →
Spotted something out of date? Suggest an update →
Alternatives to Rime Labs
More in Video & Audio
