S2.1 Pro
Fish Audio · Available for text to speech
Natural speech powered by leading AI models
Turn text into a voice people want to hear
Create natural, expressive narration for videos, products, assistants, and stories. Choose a voice or clone one you have permission to use, then shape the pace and export production-ready audio.
Natural deliveryInstant voice cloningFour audio formats
Voice studio
Text to speech preview
Your script
Every idea has a voice. Give yours the clarity, warmth, and rhythm it deserves.
Natural voiceMP3 · WAV · PCM · Opus
Speed
1.0×
Generation mode
Fast and balanced
Supported models
AI speech models available on HeyMarmot
New speech models appear here as they become available, so you can choose the right voice technology for each project.
Built for real production
Natural speech with practical control
Move from a finished script to usable audio without recording gear or a complicated editing workflow.
Lifelike delivery
Advanced speech models preserve phrasing, pauses, and expression so long narration stays clear and engaging.
Instant voice cloning
Upload a clean voice sample and its transcript to reproduce an authorized voice without a training step.
Pace and quality control
Adjust speaking speed and choose a generation mode that favors responsiveness or stability.
Flexible audio formats
Export MP3, WAV, PCM, or Opus for editing, streaming, apps, and telephony workflows.
Three simple steps
From script to speech in minutes
Start with text, pick the right voice, and generate audio that is ready to use.
Write or paste your script
Use punctuation and paragraph breaks to guide pauses, pacing, and emphasis naturally.
Choose or clone a voice
Select a saved voice ID or upload an authorized 10 to 30 second reference sample with its transcript.
Generate and download
Choose a supported model, speed, and format, then preview or download the result.
Voice cloning
Keep the voice that makes your content recognizable
A short, clean recording can become a reusable voice for narration, product demos, lessons, and branded content. An accurate transcript helps preserve pronunciation and consistency.
Only clone your own voice or a voice you have explicit permission to use.
Reference voice sample
Voice ready
Generate new speech in the authorized voice
One voice workflow, many ways to publish
Create consistent speech for the channels your audience already uses.
Voiceovers and narration
Produce narration for videos, explainers, podcasts, courses, and audiobooks.
Conversational AI
Give assistants and product experiences a more natural, recognizable voice.
Accessibility
Turn written information into clear spoken content for more people.
IVR and notifications
Create phone prompts, alerts, updates, and reusable product messages.
FAQ
Text to Speech FAQ
Answers about supported models, formats, voice cloning, and production use.
What is AI text to speech?
AI text to speech converts written text into spoken audio. HeyMarmot uses advanced speech models to create natural phrasing, pauses, and expression for narration and product experiences.
Which models does HeyMarmot support for speech generation?
The speech models currently available on HeyMarmot are: S2.1 Pro.
Can I clone a voice?
Yes. Upload a clean 10 to 30 second recording and the exact transcript. You must own the voice or have explicit permission from the voice owner.
Which audio formats can I download?
You can generate MP3, WAV, PCM, and Opus audio. MP3 is convenient for everyday publishing, WAV suits high quality editing, and PCM or Opus work well in real-time pipelines.
How do I make generated speech sound more natural?
Use clear punctuation, paragraph breaks, and a well-written script. You can also adjust speed and use a clean reference recording with an accurate transcript when cloning a voice.
Give your next idea a voice
Create natural speech with the model that fits your project and keep every voiceover in your HeyMarmot workspace.
Explore more creation guides