MisoTTS is an open-source 8-billion-parameter text-to-speech model that generates emotive, low-latency speech for conversational voice agents.
MisoTTS is Miso Labs' foundation text-to-speech model. Built on a hierarchical RVQ transformer, it produces expressive speech and supports one-shot voice cloning from a short audio sample. The model runs locally for on-premises deployment and is available with open-source weights via Hugging Face, with API access planned.
It targets developers building voice interfaces and enterprise teams that need data sovereignty over their voice layer.
| Attribute | MisoTTS | ElevenAPI | Sesame CSM | |
|---|---|---|---|---|
| Provider | |  |  | |
| Founded | 2025 | 2022 | 2022 | |
| Sells To | B2B, Developers | B2B, B2C, Enterprises | — | |
| Pricing Model | B2B Subscription | B2B Subscription, Recurring, Usage-based | — | |
| Ownership | — | Private, Venture-backed | — |
Provider
Founded 2025
Sells To B2B, Developers
Pricing Model B2B Subscription
Ownership —
Provider
Founded 2022
Sells To B2B, B2C, Enterprises
Pricing Model B2B Subscription, Recurring, Usage-based
Ownership Private, Venture-backed
Provider
Founded 2022
Sells To —
Pricing Model —
Ownership —
Outspeed Voice Platform provides realtime voice infrastructure for AI applications with natural prosody and ultra-low latency.
AI voice generation platform for text-to-speech voiceovers, voice agents, and AI dubbing.
Foundation Models are Mirage's multimodal audio and video models that generate realistic talking videos from text, audio, or images.
Bland Speech provides developers with a text-to-speech studio and speech-generation API.
Developer platform providing real-time and batch APIs for speech-to-text, text-to-speech, and voice agents.
Outspeed Voice Platform provides realtime voice infrastructure for AI applications with natural prosody and ultra-low latency.
AI voice generation platform for text-to-speech voiceovers, voice agents, and AI dubbing.
Foundation Models are Mirage's multimodal audio and video models that generate realistic talking videos from text, audio, or images.
Bland Speech provides developers with a text-to-speech studio and speech-generation API.
Developer platform providing real-time and batch APIs for speech-to-text, text-to-speech, and voice agents.