Model
Inference & Serving
Orpheus TTS
1Llama-based 2speech 3model 4with 5emotive 6tags
the six Ws · specification
W1
Who
Canopy Labs released Orpheus TTS.
W2
What
Orpheus TTS is a Llama architecture based text-to-speech model that supports zero-shot voice cloning and inline emotive tags like laughter or sighs, with low-latency streaming.
W3
Where
Hosted on GitHub at canopyai/Orpheus-TTS and on Hugging Face under canopylabs.
W4
When
Released in March 2025.
W5
Why
It shows that a repurposed LLM backbone can deliver natural, emotionally expressive, streamable speech synthesis in real time.
W6
With
Built on a Llama based transformer with a SNAC audio codec, served via vLLM for fast inference.
W7
Watch
Released under the Apache 2.0 license by Canopy Labs; fully open weights.
alternatives