What it is

Voice infrastructure company built on state-space models. Sonic 3 does text-to-speech in 40-plus languages with emotion controls, laughter tags and voice cloning from a 3-second sample. Ink handles speech-to-text and Line orchestrates phone-based voice agents.

Why it earns a slot

Around 90ms time-to-first-audio, and 40ms on Sonic Turbo - fast enough that a phone agent never sounds like it is thinking. That latency, not raw voice beauty, is why realtime builders pick it over ElevenLabs.

Know before you commit

The free tier is personal use only; any commercial deployment needs at least the $5 Pro plan. Pro voice cloning bills at 1.5 credits per character rather than 1, so cloned-voice costs run 50% above the headline rate.

Getting started

Low-latency streaming API with WebSocket support; SDKs for Python, Node and Rust.

Field notes

0 from the community

Sign in to add a field note. Reading is always open.

Loading notes…

Field check

Unverified

Sign in to confirm this entry is still accurate.

Pricing snapshot

Free 20K credits, personal use only. Pro $5/mo (100K), Startup $49 (1.25M), Scale $299 (8M). TTS is 1 credit/character, Ink STT 1 credit/second, Line voice agents $0.014/min.

Snapshots move fast — verify at checkout, and post a field note if it changed.

Drop into your Lovable chat

Evaluate Cartesia (https://cartesia.ai) for this project, then use it only if it fits what we are already building.
Context: Ultra-low-latency TTS for agents. Voice infrastructure company built on state-space models. Sonic 3 does text-to-speech in 40-plus languages with emotion controls, laughter tags and voice cloning from a 3-second sample. Ink handles speech-to-text and Line orchestrates phone-based voice agents.
Reviewer notes (general, not written for this codebase): Around 90ms time-to-first-audio, and 40ms on Sonic Turbo - fast enough that a phone agent never sounds like it is thinking. That latency, not raw voice beauty, is why realtime builders pick it over ElevenLabs.
Watch out for: The free tier is personal use only; any commercial deployment needs at least the $5 Pro plan. Pro voice cloning bills at 1.5 credits per character rather than 1, so cloned-voice costs run 50% above the headline rate.
Pricing: Free 20K credits, personal use only. Pro $5/mo (100K), Startup $49 (1.25M), Scale $299 (8M). TTS is 1 credit/character, Ink STT 1 credit/second, Line voice agents $0.014/min.
Setup: Low-latency streaming API with WebSocket support; SDKs for Python, Node and Rust.
Rules: treat the notes above as background only. Do not assume a theme, style, layout or feature that is not already in this project — inspect the existing code first and ask me before changing visual direction.

/api/public/tools/cartesia

Also in AI media generation

Adobe Firefly

firefly.adobe.com

Indemnified, commercially safe gen

Free tier + paidAI media generation

Free tier with limited watermarked credits. Standard $9.99/mo (2,000 credits), Pro $19.99 (4,000), Premium $199.99 (50,000). API is enterprise contract only.

Open dossier

Black Forest Labs (FLUX)

bfl.ai

Open-weight image models + API

Free tier + paidAI media generation

API per image: FLUX.2 klein from $0.014, pro from $0.03, flex $0.05, max $0.07. Kontext pro $0.04, max $0.08. Open weights free to self-host.

Open dossier

Descript

descript.com

Doc-style video and audio editor

Free tier + paidAI media generation

Free 1 hr/mo, 720p watermarked. Hobbyist $16/mo annual (10 hrs, 400 AI credits), Creator $24 (30 hrs, 800), Business $50 (40 hrs, 1,500). Monthly billing runs about 45% higher.

Open dossier

ElevenLabs

elevenlabs.io

Text-to-speech, voice cloning

Free tier + paidAI media generation

Free 10K credits/mo. Starter $6 (30K), Creator $11 first month then $22 (121K), Pro $99 (600K), Scale $299 (1.8M), Business $990 (6M). About 17% off annual.

Open dossier