Cartesia Sonic 3.6 Speech Generator
Turn text into natural speech with Cartesia Sonic 3.6 using a stock voice or a saved custom voice
No card required
Inputs
Loading workflow structure...
Overview
Cartesia Sonic 3.6 Speech Generator turns a script into natural speech audio in a stock Cartesia voice or a custom voice your Organization saved with Cartesia Instant Voice Clone. Use it for voiceover drafts, narration, ad reads, podcast segments, and product-demo audio when you want a quick, natural read in one of 44 languages.
Use cases
- Voice a product demo, explainer, or launch video script with a stock voice.
- Read an ad or podcast segment in your Organization's saved spokesperson voice.
- Produce the same script in several languages for regional campaigns.
- Audition voices and pacing before booking a studio session.
Input tips
- Keep the script within 10,000 characters; split longer scripts into sections.
- Choose a stock voice, or pick a custom voice saved by your Organization from the voice picker.
- Set language to the language the script is written in.
- Adjust speed between 0.6 and 1.5 and volume between 0.5 and 2 only when the default read is too fast, slow, quiet, or loud.
- Keep MP3 for sharing and review; choose WAV when an editor or audio tool needs uncompressed audio.
- Spell out numbers, acronyms, and unusual names the way they should be spoken.
Expected output
The AI Tool returns one speech audio file you can play and download, with its duration, file type, and file size. The result also shows the voice ID, language, and the number of characters the cost estimate is based on.
Caveats
- Cost is estimated from the length of the script, because exact usage is not reported for each request.
- Listen for pronunciation, emphasis, and pacing before using the audio in published work.
- Custom voices are private to the Organization that saved them.
- Only use custom voices for purposes the speaker has approved.
Related AI Tools

Cartesia Instant Voice Clone
Create a reusable private voice from a short recording with Cartesia Instant Voice Cloning

ElevenLabs TTS v3
Generate high-quality speech from text with character-level timing using Turbo v3 model. Fast generation with 74 language support.

Minimax Speech v2.8
Generate high-quality natural speech audio from text using Minimax Speech v2.8 models with expressive voice options and emotion control (up to 10K characters)
