Open-weight model
Orpheus TTS Online
Speech studio
34 voices · 18 languages
Standard+ (Orpheus) speaks at a fixed speed — the speed control applies to the other voices.
Hear Canopy Labs’ Orpheus without touching a GPU — the Llama-based 3B speech model, running in the cloud. Type text, choose Tara or Dan, and download the audio.
How it works
Three steps, about a minute
- 1
Paste or write your text
Drop in a script, a paragraph, or a whole article. Longer text is split into parts automatically and joined back into one file.
- 2
Pick a voice and preview it
Almost every voice has a preview you can play before spending anything, paid ones included. Filter by language, then hear two or three before committing.
- 3
Generate and download
Audio comes back in seconds as MP3 or WAV, with a commercial licence on every tier — including the free one.
Examples
Hear it on real scripts
Every clip below was generated here on the free tier — no account, no credits. Play one to hear the voice on that kind of writing, or load the script into the generator and make your own version of it.
Podcast opener
Cold open that sets up the episode without a sponsor read.
“This week: a bank that lost four hundred million dollars because of a rounding error, the intern who found it, and the eight months it took anyone to believe her. It's a story about arithmetic, and it is somehow also a story about how hard it is to be right early.”
Audiobook passage
Narrative prose, to hear how a voice handles description and pace.
“The house had been empty for eleven years, and it had the particular stillness of a place that has stopped expecting anyone. Dust lay on the banister in a way that suggested nobody had run a hand along it in a very long time. She did, once, on the way up, and felt slightly guilty about it.”
YouTube intro
Opens a video without the "hey guys, welcome back" throat-clearing.
“Three years ago, this cost four thousand dollars and a studio. Today it costs nothing and takes about a minute. In this video I'll show you exactly how — no jargon, no upsell, and every step is one you can copy while you watch.”
Included
What you get
- Free without an account
- Thousands of characters a day as a guest, no sign-up and nothing to install. An account raises the limit; it is not a gate on the tool.
- Commercial use included
- Monetised videos, client work and ads are all fine on every tier. You own what you generate.
- MP3 and WAV export
- MP3 for editors and uploads, WAV where you need uncompressed audio. Speed is adjustable from 0.75× to 1.5× on engines that support it.
- Long text, one file
- Text past the per-request limit is split at sentence boundaries, generated in order, and stitched back into a single download.
- Voices across 18 languages
- Several languages are read by voices trained on that language rather than an English voice reading foreign words — the difference is audible immediately.
- Credits that never expire
- Premium and ultra voices are billed in credits. One-time packs never expire; monthly plans refill every billing cycle instead.
A language model that speaks
Orpheus takes an unusual route to speech: instead of a purpose-built acoustic architecture, it starts from Llama-3.2-3B and teaches it to emit audio tokens. Delivery inherits the fluency of the underlying language model — pacing, hesitation, emphasis — which is why Orpheus clips tend to sound spoken rather than narrated.
Canopy Labs published the English fine-tune with eight named voices and an Apache-2.0 license, and the model quickly became one of the most-watched open TTS releases. As with its peers, actually running it means a GPU, a serving stack, and patience — which is the step this page removes.
The real model, two of its voices
The generator above serves the actual Orpheus checkpoint with its published Tara and Dan voices — the same names you will find in the model card. If you have seen Orpheus demos and wondered whether they were representative, the honest test is running your own text through it here.
Because it shares a free tier with Kokoro, Chatterbox, and Zonos, you can rerun the same sentence across four open models in a couple of minutes. That side-by-side beats any spec sheet for deciding which one fits your project.
When Orpheus is the right pick
From running the open models side by side:
- Pick Orpheus for conversational lines — dialogue, assistant replies, anything that should sound spoken, not read
- Pick Kokoro for long narration: faster, steadier, and 1,000–3,000 characters per part against Orpheus’s 300
- Self-host Orpheus if you need its emotive tags or the six voices not deployed here
Questions
Frequently asked
Keep going