CS

Game Voice Generators

Cartesia Sonic: Real-Time Text-to-Speech

cartesia.ai · checked 1 Oct 2026

Cartesia Sonic is a text-to-speech model for real-time voice applications.

How it plugs in

YesPublic API
NoOpen source
YesFree tier

Signal on Game Voice Generators7.4#14 of 45

Cartesia Sonic turns text into spoken audio for real-time voice applications. It supports 44 languages, sub-90ms latency, voice cloning, emotional cues, and audio localization that keeps a speaker’s identity. Sonic is available through an API and web tools, with free and paid plans. Deployment options include on-premise, customer cloud environments, and OEM licensing.

Cartesia Sonic home page

Review

In a line

Cartesia Sonic is a text-to-speech model for real-time voice applications. It is made by Cartesia, a company founded in 2023 and based in San Francisco.

What it does

Sonic converts written text into speech with latency below 90 milliseconds. It supports 44 languages and can read emotional subtext, adding non-verbal sounds such as laughter.

It can localize an audio clip into another language while keeping the emotion, tone, and speaker identity. Custom pronunciation dictionaries let you set how names and specialist terms are spoken.

Voice agents combine Sonic-3.6 text-to-speech with Ink-2 speech-to-text and a selected large language model. Managed Agents handle turn-taking and interruptions.

How it connects

Sonic offers API access and web use. Managed Agents can use phone numbers provided by Cartesia, import Twilio numbers, or connect a carrier through SIP, a standard way to link phone systems.

Cartesia says Sonic can run on your own premises, in a customer VPC on AWS, Google Cloud, or Microsoft Azure, or through OEM licensing. Cartesia lists HIPAA, SOC 2 Type 2, GDPR, and PCI on its Sonic page.

The free plan includes text-to-speech and speech-to-text. Help is available through Discord, where users can chat with the Cartesia team and community.

Good fit

Sonic suits teams building real-time voice applications, multilingual audio, or voice agents. Its voice cloning, pronunciation controls, and language localization support branded or localized speech.

It may also fit organizations that need commercial use rights or deployment choices such as on-premise or a customer cloud environment. See other text-to-speech software options if you are comparing tools.

Watch for

Check concurrency needs before choosing a plan. Free allows two concurrent Sonic text-to-speech requests, while Pro allows three; Scale lists high concurrency limits.

Voice cloning and agent use vary by plan. Confirm the credits, agent allowance, and security requirements that apply to your project, especially if you need custom concurrency, DPAs, BAAs, or single sign-on.

Pricing

Sonic has a free plan at $0/mo with 20K credits per month, $1 prepaid agents per month, text-to-speech, and speech-to-text.

Pro costs $5/mo and includes 100K credits per month, $5 prepaid agents per month, a commercial-use license, and instant voice cloning. Startup is $49/mo with 1.25M credits, $49 prepaid agents, pro voice cloning, and organizations.

Scale costs $299/mo with 8M credits, $299 prepaid agents, priority support, and high concurrency limits. Enterprise pricing is custom and includes custom credits and agent usage, volume pricing, custom concurrency limits, DPAs and BAAs, a shared Slack channel, SSO, and security questionnaires.

Verdict

Choose Cartesia Sonic if your project needs fast spoken responses, many languages, emotional delivery, or cloned voices. Its API, web access, phone connections, and multiple deployment models give teams several ways to build and run voice features.

The free tier is a way to start with limited monthly credits and concurrency. Paid plans add commercial rights, more credits, voice-cloning levels, support, and enterprise controls. Look elsewhere if your required usage, concurrency, or compliance terms do not fit these plans.

Our take

Cartesia Sonic is a practical option for developers and businesses creating live voice features, multilingual audio, or phone-based agents. The model offers sub-90ms latency, 44 languages, emotional expression, cloning, and pronunciation controls. A free tier lowers the entry cost, while paid plans add commercial rights, credits, and enterprise options. Review concurrency limits and agent allowances closely. Enterprise buyers should confirm whether the listed compliance documents, custom terms, and deployment choices meet their requirements before signing up.

  • Sub-90ms latency for real-time voice applications
  • Supports 44 languages and emotional expression
  • Voice cloning and custom pronunciation dictionaries
  • API, web access, and several deployment options
  • Free plan available
  • Free and Pro plans have limited concurrency
  • Credits and prepaid agent amounts vary by plan
  • Enterprise pricing is custom
  • Some voice-cloning features require paid plans
  • Pick it ifReal-time voice applications and voice agents; Multilingual audio with emotion and speaker identity; Teams needing API or private-cloud deployment
  • Skip it ifYou need concurrency beyond the limits listed for your plan; You require predictable enterprise pricing; You need features not covered by the listed plans

Pricing

A free $0/mo plan offers 20K monthly credits. Plan features and limits vary, including credits, prepaid agent amounts, voice cloning, concurrency, and support.

  • Pro$5/mo · 100K credits / month · $5 prepaid agents / month · Commercial use license · Instant voice cloning$5/mo
  • Startup$49/mo · 1.25M credits / month · $49 prepaid agents / month · Pro voice cloning · Organizations$49/mo
  • Scale$299/mo · 8M credits / month · $299 prepaid agents / month · Priority support · High concurrency limits$299/mo
  • Free$0/mo · 20K credits / month · $1 prepaid agents / month · Text to Speech · Speech to TextFree
  • EnterpriseCustom · Custom credits & agent usage · Volume pricing · Custom concurrency limits · DPAs and BAAs · Shared Slack channel · SSO · Security questionnairesOn request

As cartesia.ai lists them, checked 1 Oct 2026.

Questions

What is Cartesia Sonic?

Sonic is Cartesia’s text-to-speech model for real-time voice applications.

How fast is Sonic?

Cartesia lists sub-90ms latency.

How many languages does it support?

Sonic supports 44 languages.

Does Sonic clone voices?

Yes. Pro includes instant voice cloning, while Startup includes pro voice cloning.

Can Sonic connect to phone systems?

Managed Agents can use Cartesia numbers, import Twilio numbers, or connect a carrier over SIP.

Is there a free plan?

Yes. Free costs $0/mo and includes 20K credits per month, text-to-speech, and speech-to-text.

Where can Sonic run?

Cartesia says Sonic can run on-premise, in a customer VPC on AWS, Google Cloud, or Microsoft Azure, or through OEM licensing.