Our take: Soniox Text-to-Speech suits developers building multilingual voice agents who want streaming and REST speech generation with audio-tag control and voice cloning at roughly $0.70 per hour of generated speech.
Main limitation: It is a developer API with token-based billing that needs conversion to estimate cost, and cloned voices must be recomputed when a new model ships.
By the AI Tool Atlas editors.
Based on the vendor's official documentation; we have not tested this product hands-on. What “documented overview” means
Who should use Soniox Text-to-Speech
Soniox Text-to-Speech suits developers building multilingual voice agents who want streaming and REST speech generation with audio-tag control and voice cloning at roughly $0.70 per hour of generated speech. Based on the vendor's documentation, not on a hands-on test.
Best for: Developers building voice agents
Also a good fit for
Teams needing multilingual TTS
Products replacing several speech vendors with one API
Consider another tool if developer-oriented API; no consumer TTS app described
Or if
Token-based billing needs conversion to estimate cost per minute
Cloned voices must be recomputed when a new model ships
Soniox Text-to-Speech pricing and usage limits
Prices checked Oct 1, 2026 on the official US pricing page. Official pricing ↗ Prices and limits change often.
Vendor pricing page re-checked automatically 11 days ago: no changes detected.
The official Soniox Text-to-Speech pricing page as we saw it. soniox.com, captured Oct 1, 2026
Free access and trial conditions
Not stated on the pricing page
Not stated; console includes a browser demo
Paid plans: what changes as you upgrade
Pay as you go API
$0.7 per hour of generated speech (approx.); $4.00 per 1M input text tokens and $21.50 per 1M output audio tokens source (opens in a new tab)
Unknown, annual commitment
Enterprise discounts and committed-use contracts may differ
For: Voice agents and multilingual speech products
Limits
15,000 input text tokens is about 1 hour of speech
1 hour of speech is about 30,000 output audio tokens
What is Soniox Text-to-Speech, and how does it work?
Soniox Text-to-Speech is the speech generation part of the Soniox speech AI platform, which also offers speech-to-text and real-time translation APIs. The current TTS v2 model produces speech in 60+ languages, supports mid-sentence language switching, reads alphanumerics such as phone numbers and IDs accurately, and can be steered with inline audio tags like [warm] or [sighs].
The API is offered as WebSocket streaming for low-latency voice agents and as a REST endpoint for batch generation, with Python and Node SDKs and integrations for LiveKit and Pipecat. Custom voices can be cloned from a reference clip of up to 2 minutes in the Console or via the voice API. Billing is token-based pay-as-you-go.
Soniox Text-to-Speech's official site: how the product presents itself. soniox.com, captured Oct 1, 2026
What you provide and what you get back
What you give Soniox Text-to-Speech and what comes back, from its documentation
You provide
Text to speak
Language and voice choice
Inline audio tags such as [warm] or [sighs]
A short audio sample for cloning
You get back
Streaming audio over WebSocket
Single-request audio via REST
Cloned custom voices
Where you can use it
Platforms
Web, API
Account
Required
What people use Soniox Text-to-Speech for
Scenarios are worked out by our editors from the documentation; the actual result depends on your input and plan.
Voice agents that speak as the LLM response streams in
Not verified: Free credits; Privacy and retention; Maximum text length.
How we documented and maintain this page
Every fact about Soniox Text-to-Speech above comes from the 5 official pages listed here, last compared on Oct 1, 2026; we have not run Soniox Text-to-Speech ourselves, so it has no editorial score. How we check and re-check pages.
Article updated Oct 10, 2026; prices checked Oct 1, 2026. Outbound links to vendors are plain links; we do not earn commissions. How we review · Report an error
