
Soniox
Multilingual speech AI platform offering real-time transcription, translation, and text-to-speech.

About Soniox
Soniox: Multilingual Speech AI Platform
Soniox is a multilingual speech AI platform offering real-time Speech-to-Text, Text-to-Speech, and Speech Translation through one unified API. Its latest model, Soniox STT v6, adds agent-aware recognition for voice agents and higher accuracy across 60+ languages, including speakers who switch languages mid-sentence.
Key Features
Soniox STT v6: Real-time and asynchronous speech recognition across 60+ languages, with improved speaker diarization, language identification, endpoint detection, and context customization.
Agent Text Input and Echo Cancellation: Developers stream the agent's outgoing text into STT v6, giving it context from both sides of the conversation and filtering the agent's own voice out of recognition while still hearing user interruptions.
Text-to-Speech v2: Expressive speech in 60+ languages with audio tags that direct delivery (whisper, laugh, hesitate), precise handling of alphanumerics, and low-latency streaming.
Voice Library and Cloning: More than 200 built-in voices tagged by gender, age, accent, use case, and style, each speaking all 60+ languages, plus instant voice cloning.
Speech Translation: Real-time translation across 3,600 language pairs, with output before sentences finish, delivered in the same API call as transcription.
Data Residency: The same models and API in every region with in-region processing, including a regional deployment in India currently in limited early access.
Privacy and Compliance: Audio is never stored or saved. Compliant with SOC 2 Type 2, ISO/IEC 27001:2022, HIPAA, and GDPR.
Soniox Compare: A public tool to test Soniox side by side with other speech-to-text, text-to-speech, and translation providers on live inputs.
Soniox App: A ready-to-use app for individuals and teams, with Smart Scribe, Translator, and Voice Typing.
Use Cases
Voice Agents: Powering conversational AI with low-latency recognition that understands what the agent just said, plus natural speech output.
Dictation and Voice Typing: Turning speech into accurate text for editors, messaging apps, forms, medical software, and enterprise tools.
AI Meeting Products: Transcribing live or recorded meetings with speaker diarization for notetakers and conversation intelligence.
Wearables and Translation: Delivering streaming recognition, speech generation, and real-time translation on devices that need minimal delay.
Pricing
Speech-to-Text: About $0.10/hour for async (file) and $0.12/hour for real-time (streaming), billed by tokens. Diarization, language identification, and smart formatting are included.
Text-to-Speech: About $0.70/hour of generated speech, billed by tokens.
Getting Started
Website: https://soniox.com/
Documentation: https://soniox.com/docs
Pricing: https://soniox.com/pricing
Soniox gives developers and enterprises one accurate, low-latency speech platform for building multilingual voice products, from voice agents and dictation to meeting tools and real-time translation.




