Build with API

Speech-to-text for voice agents that understand people in real time

Give your voice agent fast, accurate speech understanding, natural turn-taking, and multilingual support across 60+ languages. Soniox streams speech as it happens and knows when the user is done speaking, so your agent can start thinking sooner and respond at the right moment.

Build with APIExplore docs

Trusted by teams building global voice products

Livekit
Krisp
Pipecat
Summary AI
Perplexity
Samsung
Wispr Flow
LG
Agora
Retell AI
Fireflies.ai
Skit.ai
Kindroid
Deliver Health
Truecaller
Journalia
Mobius
TranscribeMe
Vapi
Zomato
SLNG
Japan AI
Boost.ai
Convin
Genspark
HappyRobot
Uniscribe
Jamie.ai
InteractCX
The Plato
MobilApp
Onvego
Wonderful.ai
Manifone
Tana
Transync AI
SotaTek

What voice agents need from speech-to-text

Voice agents are only as good as their conversation loop. The user speaks. The agent understands. The agent responds. Every delay or recognition mistake in that loop makes the experience feel less natural.Soniox is built for real-time conversational AI, giving your agent accurate streaming speech recognition, semantic turn detection, multilingual understanding, and the speed needed to keep conversation moving.

Understand speech as it happens

Receive transcription continuously while the user is still talking, so your application can begin processing speech before the full utterance is complete.

012345678901234567890123456789ms

Know when to respond

Semantic endpoint detection uses more than silence to determine when someone has finished speaking, helping agents respond quickly without constantly cutting users off.

Understand real people

Handle accents, fast speech, background noise, names, numbers, specialized vocabulary, and conversations that switch between languages.

Deploy globally with one model

Support 60+ languages with one model. Pair with Soniox Text-to-Speech so your agent can understand and respond naturally across languages. Learn more

English

From speech to response, every millisecond matters

A voice agent is a real-time pipeline: it listens to the user, turns speech into text, sends that understanding to the LLM, and speaks the generated response back. Every step adds latency, so optimizing only one part of the conversation is not enough. Soniox is designed for the two places where your agent meets the user: understanding what they say and speaking back to them.

Hear

Soniox streams transcription tokens as audio arrives, giving your agent access to the conversation before the speaker has finished.

Know when the turn is over

Built-in semantic endpointing identifies when the user has completed a thought and emits a clear end-of-turn signal for your application.

Think

Send stable speech transcription directly into your LLM, tools, or agent framework while preserving the speed of a real-time interaction.

Speak

Pair Soniox Speech-to-Text with Soniox Text-to-Speech to stream the agent's response back as natural speech as soon as the LLM starts generating.

Build with Soniox API

Works with the voice agent stack you already use

Connect Soniox to LiveKit, Pipecat, Twilio, and other real-time voice frameworks without rebuilding your agent architecture.

Use Soniox in popular frameworks

Soniox integrates seamlessly with leading real-time communication platforms, AI frameworks, automation tools, and developer SDKs.

An open source framework and developer platform for building, testing, deploying, scaling, and observing agents in production.

Open source framework for voice and multimodal conversational AI.

Twilio is a cloud-based customer engagement platform (CPaaS) that provides APIs, allowing developers to integrate voice, messaging (SMS, WhatsApp), email, and authentication capabilities into applications.

Open-source development framework designed to build applications powered by large language models (LLMs).

The open-source AI toolkit designed to help developers build AI-powered applications and agents with React, Next.js, Vue, Svelte, Node.js, and more.

Open-source AI SDK with a unified interface across multiple providers. No vendor lock-in, no proprietary formats.

n8n is a powerful, low-code/pro-code workflow automation tool that connects various applications, APIs, and databases to automate tasks.

Build the next generation of voice products, from agents and wearables to dictation, translation, and real-time multilingual experiences.

Everything your voice agent needs to feel natural

Every part of Soniox APIs is designed to reduce latency, improve accuracy, and make your voice agent feel immediate.

Start understanding before the user stops talking

Soniox streams transcription as speech arrives, giving your agent access to what the user is saying before the full utterance is complete.

Downstream systems can start detecting intent, retrieving context, and preparing LLM or tool calls earlier, reducing the time between the end of the user's turn and the response.

Explore real-time transcription

Accuracy matters when software takes action

Voice agents act on what they hear. Names, dates, addresses, numbers, account details, commands, and product terminology need to arrive correctly before your application takes the next step.

Give Soniox relevant names, terminology, entities, and domain knowledge at request time to improve recognition without maintaining separately fine-tuned speech models.

Explore context customization

Complete the conversation with Text-to-Speech

Understanding the user is only half of the experience. The response also needs to arrive quickly and sound natural enough to keep the conversation moving.

Pair Soniox STT with Soniox TTS to stream LLM output into speech, generate critical information accurately, and stop the response cleanly when the user interrupts.

Explore Text-to-Speech for voice agents

Turn-taking that feels like conversation

Silence alone does not tell you when someone has finished speaking. Respond too early and the agent interrupts; wait too long and the conversation feels slow.

Soniox uses semantic endpoint detection to recognize conversational boundaries and emit a clear end-of-turn signal, helping your agent respond at the right moment.

Explore endpoint detection

One voice agent for every language

Real conversations cross language boundaries. Users switch languages, use foreign names, and mix terminology naturally within the same interaction.

Soniox handles 60+ languages and in-stream language switching with one unified model, so you can build one speech pipeline instead of routing every language separately.

See supported languages
Agora

“It just gets the words right, any language, any accent, any context. That’s what accuracy is supposed to look like.”

Tony Wang,
Cofounder & Chief Revenue Officer at Agora

Build every kind of voice agent

AI assistants

Build conversational assistants that understand users quickly, answer questions, and complete tasks through natural voice interaction.

Customer service agents

Handle customer conversations across languages, accents, background noise, and interruptions while resolving requests in real time.

Scheduling and booking agents

Capture names, dates, times, locations, and requests accurately so agents can schedule appointments and complete bookings reliably.

Sales and qualification agents

Understand prospects in real time, collect key information, answer questions, and trigger the right next action.

Call routing agents

Detect caller intent from natural speech and route conversations instantly without rigid phone menus or IVR trees.

In-app voice agents

Bring real-time voice interaction into web, mobile, desktop, automotive, wearable, and embedded products.

Simple, usage-based pricing. Get started with real-time API for ~$0.12/hour.

Start building

Power up your multilingual AI voice agent

Get production-ready speech-to-text transcription and translation in 60+ languages.

AfrikaansAlbanianArabicAzerbaijaniBasqueBelarusianBengaliBosnianBulgarianCatalanChineseCroatianCzechDanishDutchEnglishEstonianFinnishFrenchGalicianGermanGreekGujaratiHebrewHindiHungarianIndonesianItalianJapaneseKannadaKazakhKoreanLatvianLithuanianMacedonianMalayMalayalamMarathiNorwegianPersianPolishPortuguesePunjabiRomanianRussianSerbianSlovakSlovenianSpanishSwahiliSwedishTagalogTamilTeluguThaiTurkishUkrainianUrduVietnameseWelsh

Privacy and compliance, built right in

Never stored, never saved.

Audio stays in memory, everything is processed in real-time.

Built for privacy-critical use cases.

Adhering to leading global security, privacy, and compliance standards.

Trusted where privacy matters most.

Used in industries where speech is sensitive, from healthcare to enterprise.

Soniox is Soc 2 Type 2 compliant
Soniox is ISO 27001:2022 compliant
Soniox is HIPAA compliant
Soniox is GDPR compliant
SOC 2 Type 2 · ISO/IEC 27001:2022 · HIPAA · GDPR

Frequently asked questions about Soniox for voice agents

What is the Soniox Speech-to-Text API?
Soniox provides a real-time speech-to-text API designed for AI voice agents. It converts live audio into text with low latency, supports streaming use cases, and works across more than 60 languages without switching models or restarting the stream.
Is Soniox suitable for building AI voice agents?
Yes. Soniox is designed for real-time voice agent workflows, including streaming transcription, early token delivery, and endpoint detection for turn-taking, all configurable through the API.
What makes Soniox a low-latency speech-to-text API?
Soniox uses a real-time streaming architecture that emits transcription results incrementally as audio arrives. This allows voice agents to begin processing speech before an utterance is complete, reducing end-to-end response time.
How does Soniox handle partial and final transcripts?
The streaming API provides non-final transcription tokens followed by finalized tokens. This enables early intent detection, real-time UI updates, and stable downstream processing without parsing entire transcripts.
How does Soniox detect when a user finishes speaking?
Soniox includes built-in endpoint detection that identifies speech boundaries. Voice agents can use these events to decide when to respond without relying on client-side silence timers.
Can I customize transcription behavior for my voice agent?
Yes. The Soniox API is configurable, allowing developers to adjust transcription behavior, including custom context for domain-specific vocabulary, eliminating the need to maintain separate fine-tuned models for different tasks.
Does Soniox support multilingual voice agents?
Yes. Soniox supports consistent multilingual transcription and translation across more than 60 languages using a single real-time model. Language identification happens automatically within the same stream.
Can Soniox handle language switching within a conversation?
Yes. Soniox can recognize and transcribe speech when speakers switch languages mid-sentence or mid-conversation, without requiring stream restarts or language-specific routing.
Is Soniox suitable for regulated industries?
Yes. Soniox supports data residency for regulated environments such as medical and legal use cases, allowing speech and transcript data to remain within required geographic regions while using the same real-time API.
Is audio stored when using the Soniox API?
No. Audio is processed in real time and kept in memory only. Soniox is designed for privacy-critical applications where speech data should not be stored by default.
Can I use Soniox for phone and telephony voice agents?
Yes. Soniox can transcribe live telephony audio in real time and integrate with voice infrastructure such as Twilio, LiveKit, and Pipecat. Your application can stream caller audio into Soniox and use the resulting transcript, endpoint events, and structured speech data inside the rest of your voice-agent pipeline.
How do developers get started with Soniox?
Developers can generate an API key on Soniox Console and start streaming audio over websockets to Soniox directly. The API integrates with common voice agent frameworks and real-time media pipelines, making it easy to add speech-to-text to existing systems.

Ready to get started?

Create an account instantly, or contact us to design a custom package for your business.

Build with API

Documentation

Get up and running in minutes and spend your time building, not wrestling with the API.

Explore docs

See what you’ll pay

Pay only for what you use with our flexible pricing. Built to scale with you.

Pricing details