Today, we’re launching Orpheus, an open-source TTS model that exceeds the capabilities of both open and closed-source models such as ElevenLabs and OpenAI! (1/6)

Orpheus is capable of expressing empathy, consistent with the emotional intelligence of a human, being able to produce non-textual cues like sighing, laughing, chuckling, etc. Open-source TTS models have not been competitive with closed-source and we’re changing that today! (2/6)

Orpheus is a family of pretrained and fine-tuned models with 3B parameters, and will release smaller models in 1B, 500M, and 150M in the next few days. We demonstrate extremely high-quality, aesthetically pleasing speech generation even through very tiny model sizes. (3/6)

Our fine-tuned models can be used for conversational use cases, and our pretrained model can be used for a variety of downstream tasks, like voice cloning or classification. (4/6)

This is just one of the many pieces we've built at Canopy Labs. We're on a mission to build 3D digital humans that are indistinguishable from real humans. We see a future where every AI application will have a “human” that you can interact with. (5/6)

If any of this sounds interesting, we’d love to hear from you! Additionally, we will pay $5,000, for any successful referral! (6/6)

Check out all the details at

Discover the future of AI investing. AIS delivers exposure to the companies driving the next wave of innovation—semiconductors, data centers, and AI applications. Explore the supercycle today.

Can you find tune it to talk 3x faster than it already does?

We'll do that soon – can probably do that with like 50 training examples

did you manage to offer what @sesame was promising for an open-source model ?

@sesame Yep, but with full code and weights open-sourced!