All models
controllable

Maya1

Apache-2.0

Design a voice from a text description, not a reference clip

Maya1 from Maya Research skips voice cloning entirely: you describe the voice you want in natural language and it designs one from scratch, with over 20 inline emotion tags — laugh, sigh, whisper, and more — for expressive delivery. Served through vLLM it hits sub-100ms latency, making it viable for live, interactive use.

Hear Maya1

Standardized samples are pending for this release. See the official sources below for the author’s demo.

Maya1 · Reference-based · Same three scripts across the directory.

Best for: Design a voice from a text description, not a reference clip

Before you choose: Compare the samples with your own use case. Check the linked documentation for code, weight, and voice license terms.

Details awaiting a fresh review. Hardware figures are estimates unless a benchmark is linked.

No samples yet

Every model in this directory is read against the same three scripts so voices can be compared honestly — Maya1’s samples just haven’t been generated yet.

Contribute samples on GitHub →

Run Maya1

git clone https://github.com/MayaResearch/maya1-fastapi

Maya1 FAQ

Do I need a GPU to run Maya1?

The directory lists roughly 12 GB of VRAM as a historical estimate. Check the model documentation for your checkpoint and runtime.

Can Maya1 clone voices?

No — Maya1 uses preset voices. Browse the voice-cloning category for models that clone from reference audio.

What license is Maya1 released under?

Maya1 is released under the Apache-2.0 license. Always check the repository for the exact terms — some models license code and weights separately.

What languages does Maya1 support?

Maya1 supports English.

Is there a hosted Maya1 API?

OpenSpeech Cloud is accepting production-access interest for Maya1, but this model is not hosted in the beta. Use the official repository for current deployment options.