Rime
Category: Voice / Speech AI
Enterprise voice AI platform building ultra-realistic, linguistics-first conversational speech models for high-volume phone interactions. Rime was founded in 2022. The company is led by Lily Clifford. Based in San Francisco, California, USA. Team size: 11-50. Total funding raised: $29.5M. Latest round: Series A. Key investors include M13 Ventures, Twilio Ventures, Unusual Ventures, Corazon Capital, Cadenza Capital, Founders You Should Know.
- Founded
- 2022
- Headquarters
- San Francisco, California, USA
- Team size
- 11-50
- Total funding
- $29.5M
Value proposition
Rime provides enterprise-grade voice AI models trained on proprietary conversational speech data (not scraped audiobooks), delivering sub-100ms latency, 600+ voices across 50+ languages, and deterministic pronunciation control — purpose-built for real-time customer calls in regulated industries.
Products and solutions
Mist v3 (production TTS engine, ~40ms p90 TTFB, deterministic pronunciation), Arcana v2 (expressive conversational model, 40+ voices, multilingual), Coda (flagship balanced model, 600+ voices, 50+ languages, dual-decoder), Rimecaster (open-source speech model), On-premise deployment option, HIPAA & SOC 2 Type II compliant
Unique value
Linguistics-first approach with proprietary in-studio recorded conversational speech dataset (not web-scraped); phoneme-based architecture for deterministic pronunciation; sub-100ms latency; on-premise deployment for regulated industries; HIPAA & SOC 2 Type II compliance.
Target customer
Enterprise product and engineering leaders in telecom, healthcare, financial services, food service/restaurants, and airlines who need high-volume, compliance-driven voice deployments.
Industries served
Healthcare, Financial Services/Fintech, Food Service/Restaurants, Airlines, Telecom, Customer Support
Technology advantage
Proprietary dataset of full-duplex spontaneous speech (interruptions, laughter, disfluencies) recorded in-house; phoneme-based architecture enabling deterministic pronunciation without retraining; dual-decoder Coda model balancing expressiveness and speed; on-premise deployment capability unique among next-gen voice AI providers; sub-100ms latency in production.
How they differentiate
Unlike ElevenLabs (which has moved into orchestration/application layer) and Deepgram (modular stack), Rime focuses exclusively on the voice model layer with a linguistics-first approach, proprietary conversational training data, deterministic pronunciation control, and on-premise deployment for regulated industries. Rime's phoneme-based architecture reduces customer customization load vs competitors who require retraining for industry-specific terms.
Main competitors
ElevenLabs, Deepgram, Cartesia
Key partnerships
Together AI (native model hosting, sub-700ms end-to-end voice pipeline), Oracle Cloud Infrastructure (3x price-performance on healthcare workloads), SLNG (low-latency voice AI in underserved global markets)
Notable customers
Mayo Clinic, Dialpad, Upstart, Asurion, Domino's, Wingstop, ConverseNow, SigmaMind AI, Trillet AI, Attune, LiveKit
Major milestones
2022: Company founded, 2024: Mist model ships, 2025-05: $5.5M Seed led by Unusual Ventures, 2025: Arcana v2 launches (multilingual, on-prem), Rimecaster open-sourced, 2026-03: Partnership with Together AI for native model hosting, 2026-04: Mist v3 released (~40ms p90 TTFB), 2026-06: Coda flagship model launched (600+ voices, 50+ languages), 2026-07: $24M Series A led by M13 Ventures, Rafael Valle joins as Chief Scientist (ex-Meta, ex-NVIDIA)
Growth metrics
Powers 100M+ phone conversations per month; 1.5M+ minutes of conversation monthly; 600+ voices; 50+ languages; sub-100ms latency in production
Market positioning
Enterprise-focused voice AI model provider competing with ElevenLabs and Deepgram in the voice model layer, differentiated by linguistics expertise, proprietary training data, and compliance-ready deployment (on-prem, HIPAA, SOC 2). Positioned as the "trusted enterprise voice AI platform" for regulated, high-volume phone interactions.
Geographic focus
Primarily US-based (San Francisco), with multilingual capabilities (50+ languages) enabling global enterprise deployments.
About Lily Clifford
Stanford NLP PhD dropout; Computational Linguistics researcher. Co-founded Rime in 2022 after leaving Stanford PhD program.
Latest news about Rime
More Voice / Speech AI companies
Official website: https://rime.ai