Voice developers can download the weights and clone voices from ten seconds of audio, but hosted API pricing is unpublished.
Kenning's take
What it does
MisoTTS is an 8 billion parameter speech model that generates voice from both text and the audio of the person speaking, so replies can match their tone. Miso Labs says it responds in about 110 milliseconds and can clone a voice from a ten-second clip. The weights are free on Hugging Face with code on GitHub, and hosted API access was listed as coming soon at launch.
- 8B parameter model conditioned on text and audio
- One-shot voice cloning from a ten-second clip
- About 110 ms response latency (maker's figure)
- Open weights for local deployment
- On-premises hosting and support contracts for enterprises
Who it suits. Developers building voice agents who want an expressive speech model they can run themselves.
Key facts
- Category
- AI models
- Made by
- Miso Labs
- Released
- 3 June 2026
- How to get it
- free open weights on Hugging Face and code on GitHub; API access by request from misolabs.ai
- Works on
- self-hosted, API
- Free plan
- open weights free on Hugging Face
FAQ
What is Miso One?
Miso One is how The Rundown lists Miso Labs' first speech model, which the company calls MisoTTS (miso-tts v1).
Is MisoTTS open source?
Its weights are public on Hugging Face and the code is on GitHub.
When was MisoTTS released?
Miso Labs announced it on 3 June 2026.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
ACE-Step 1.5
Open-source music generation model that writes full songs with lyrics and runs on ordinary consumer GPUs.
AlphaGenome Atlas
Google DeepMind database of AI-predicted effects for every possible single-letter change in human DNA.
Amazon Nova 2
Amazon's second family of Nova AI models, covering reasoning, speech conversation and combined text-and-image generation.
Cisco Antares
Cisco's small open-weight models that find where known security vulnerabilities sit in a codebase, running locally.