Text-to-speech
Turn any script or passage into natural spoken narration or voiceover.
Audio Generation
Auteryn agents can produce audio — not just text and images. Turn a script into natural-sounding narration, or create a multi-speaker, podcast-style conversation, all as part of a normal task.
What you can create
Section titled “What you can create”Multi-speaker audio
Create a podcast-style conversation between multiple distinct voices from a script.
Slide narration
Add spoken voiceover to presentation slides for a narrated deck.
How it works
Section titled “How it works”- Describe the audio you want — the text to speak, a voice or style, or the speakers in a conversation.
- The agent generates the audio and uploads it to secure cloud storage.
- The result appears as a playable artifact in the workspace, ready to download or reuse.
For narrated slide decks, audio narration is added directly to your presentation — see the Presentations deep dive.
Text-to-speech costs 50 credits per 1,000 characters in a preset, designed or cloned voice. A voice you cloned before the consent line was added costs 60 credits per 1,000 characters, per the pricing page. Costs are deducted from your org’s credit balance automatically.
Audio your agent listens to
Section titled “Audio your agent listens to”Audio isn’t only an output. When you attach a recording to a chat, or send your agent a voice note through a connected channel, the agent works from what you actually said — models that can’t listen to audio directly are given an automatic transcript first, so the choice of model no longer decides whether a voice note is heard. A transcript is billed at 10 credits per minute of audio, and the same attachment isn’t transcribed again when a later turn reads it.
Voice conversations vs. audio generation
Section titled “Voice conversations vs. audio generation”Audio generation produces a file from a script. If instead you want a real-time, back-and-forth spoken conversation with an agent — where you talk and it replies live — that’s the separate Voice capability.
| Audio generation | Voice conversation | |
|---|---|---|
| Output | A generated audio file | Live two-way conversation |
| Best for | Narration, voiceover, podcasts | Talking to your agent in real time |
| Where | Internal agents with Audio generation on | Voice-enabled agents |
- Be specific about tone and pacing in your instructions (“warm and conversational”, “brisk and professional”).
- Review generated audio before publishing — models can mispronounce names or unusual terms.
- For narrated presentations, generate the deck first, then add narration in one pass.