Elias Fizesan co-founded Canopy Labs with Amu Varma. His public work focuses on digital humans: combining a conversational interface with expressive speech, rather than treating text-to-speech as an isolated voice output.[2][3]
Start with Orpheus TTS →
Career
Contributions
Orpheus provides model weights and practical inference examples, including streaming output. Its repository also supplies data-processing and fine-tuning material, allowing developers to inspect the training path as well as try the resulting voices. Fizesan is associated with this team release through his Canopy role.[4]
Ideas
Fizesan describes speech as one part of a broader digital-human interface. In his Orpheus launch post, he points to expressive speech and conversational cues as ingredients of a more natural interaction. This is a founder’s product direction, rather than independent evidence that a system is indistinguishable from a person.[2]
Tools, writing & talks
Tools
Start here
Orpheus TTS
github.comThe company’s speech models and examples for streaming synthesis and fine-tuning.
Begin with the streaming example and model choices. Running the code needs a compatible Python environment and model-serving hardware or a hosted endpoint. Compare voice quality and latency on your own input.
Company release; co-founder[4]
Writing, research & talks
- Digital humans and Orpheuslinkedin.com
Fizesan’s first-person explanation of how speech fits Canopy’s broader direction.
Author[2]