Cartesia Sonic

What it does

Cartesia Sonic turns text into streamed speech for voice applications. Developers send a script, select a voice and optionally supply pronunciation entries or non-verbal cues such as laughter. The service returns audio that can begin playing before the full response is generated. Voice cloning and multilingual delivery support applications that need a consistent spoken identity.

Teams use Sonic as the speech layer in conversational agents, support systems and narrated experiences. The surrounding application supplies the response text and handles business actions; Sonic generates the voice. Custom pronunciation dictionaries help with product names and specialist vocabulary. Voice-cloning options and concurrency allowances depend on the selected plan.

Sonic suits developers building real-time voice products. Free access includes 20,000 monthly model credits; the Pro plan is $5 per month with 100,000 credits and commercial-use licensing. Sonic charges one credit per character, and paid plans can enable overages. Enterprise customers can discuss on-premises or private-cloud deployment with Cartesia. Test pronunciation and delivery with representative scripts before production use.

What it can do

• Stream speech from text for real-time playback
• Control delivery with emotional context and non-verbal cues
• Clone a voice from a short sample, subject to plan and consent requirements
• Generate multilingual speech and localized voices
• Specify custom pronunciations for names and specialist terms

Example workflows

Speak an agent response

Illustrative example. Supply a response from your application, send the text to Sonic and play the returning audio chunks. The output is a spoken reply; the application remains responsible for the conversation and business actions.

Prepare branded narration

Illustrative example. Supply a script and pronunciation entries for product names. Generate speech, listen for errors and revise the text or dictionary before using the resulting narration.

Before you choose

• Credit usage is about 1 credit per character for Sonic TTS, roughly 750 to 800 credits per minute of audio, so volume costs scale with text length.
• Voice cloning requires verified consent from the speaker; cloning voices without permission is prohibited.
• On-prem, VPC, and OEM deployments are available only under enterprise contracts with custom pricing.

Listed under

• Coding
• Customer support
• Developers
• Multilingual
• Real-time audio
• Text to speech
• Voice agents

Get started

Free plan includes 20K credits per month with access to Sonic and the API.

1. Create an account at play.cartesia.ai and open the playground to test Sonic.
2. Generate an API key from the playground.
3. Follow the Quickstart in the Cartesia docs to stream first audio in Python or JavaScript.

Credits are spent on TTS at 1 credit per character. Overages can be enabled on paid plans at per-1M-credit rates by tier (Pro $65, Startup $45, Scale $38). Monthly renewal on purchase date. The page shows no separate annual price for these plans.

Source: cartesia.ai/pricing, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Sonic AI voice: real-time TTS API with emotion | Cartesia ↗www.cartesia.ai
• Pricing: free plan, TTS, STT, and voice agents | Cartesia ↗www.cartesia.ai
• Welcome to Cartesia - Cartesia Docs ↗docs.cartesia.ai

Company website

cartesia.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.