Introducing Soniox TTS v2, our most powerful text-to-speech model yet. Soniox TTS v2 brings extraordinary voice quality, expressive control through audio tags, exceptional precision, high-fidelity voice cloning, more than 60 languages, natural language mixing, and low-latency streaming together in one model. Built from the ground up with a new model architecture, audio codec, and highly optimized inference engine, Soniox TTS v2 delivers frontier TTS at just $0.70 per generated hour. Available globally today as tts-rt-v2. Try it out: soniox.com/text-to-speech Turn the sound on and hear what’s possible. 🔊

Soniox real-time STT lets you tune endpoint detection to avoid cutting callers off mid-sentence without adding unnecessary latency. For voice agents, the right endpoint settings can help balance natural turn-taking with low latency. Test it on your own calls: soniox.com/speech-to-text

Scaling voice AI in India with @bolna_dev 🤝

Realtime speech analytics at scale.

How much does your voice agent’s greeting cost you? Not much. Until you multiply it by a few million calls and every word compounds. Soniox TTS is built and priced for scale. soniox.com/text-to-speech

Probably more than 200 hrs to be honest

We’re excited to announce that @AgoraIO has selected Soniox as a speech-to-text provider for its real-time Conversational AI platform. For over a decade, Agora has built the global network infrastructure that connects people through real-time voice, video, and messaging. Today, that same carrier-grade infrastructure is powering a new era of interactions: humans talking directly with AI. For voice AI, moving audio in real time is only part of the problem. The system must understand natural, conversational speech instantly—accurately capturing critical details despite accents and noisy environments. Soniox brings: • Native-speaker accuracy across 60+ languages • Ultra-low latency for real-time conversations • Seamless multilingual speech and mid-sentence language switching • High precision on names, numbers, IDs, and other critical details • Reliable performance on real-world audio at production scale Agora provides the orchestration for Conversational AI and the infrastructure for real-time communications. Soniox provides the intelligence for understanding spoken language. Together, we’re making it easier for developers to build voice AI that can communicate naturally with people around the world. Excited to partner with the Agora team and help power the next generation of conversational AI.

When voice becomes an interface to AI, speech recognition is the first layer of intelligence. It has to understand instantly and accurately, across languages, accents, names, numbers and real-world speech.

Soniox TTS v2 now available also on @telnyx

Make your @HeyGen avatars multilingual and more expressive with Soniox TTS v2. • Choose from 200+ voices over 60+ languages. • Instantly clone your voice. • Switch languages naturally mid-sentence • Pronounce names and technical terms correctly. Generate speech with Soniox, lip-sync with HeyGen. soniox.com/text-to-speech 🍪

Interrupt anytime. The next reply is ready. Users interrupt. Voice agents need to change direction instantly, without waiting for the old reply to finish or reconnecting. With Soniox TTS v2, you can run multiple independent streams over one WebSocket. When the user cuts in, start the new response in a fresh stream and cancel the old one. Each stream starts and stops on its own. One connection. Up to 5 concurrent streams. Zero reconnects. Learn more: soniox.com/text-to-speech

Voice AI is not only real-time. Meetings, medical conversations, customer calls, interviews, podcasts, and years of call center recordings all contain valuable information that needs to be turned into structured data for LLMs, search, analytics, and other downstream systems. Soniox Async STT is built to do this at massive scale. It doesn’t just return a transcript. It turns recorded conversations into a structured representation of what was said, who said it, in which language, when it was said, and how confident the model is in each token. Submit recordings up to 5 hours long, provide context about your domain and terminology, and get: • Accurate transcription across 60+ languages • Speaker diarization • Token-level language identification • Token-level confidence scores • Precise timestamps These annotations make recorded conversations much more useful for downstream AI systems, whether you are feeding them into LLMs, building search, extracting insights, or analyzing millions of hours of historical audio. At about $0.10/hour, even very large archives of recorded audio become practical to process. Turn your recorded audio into structured conversational data: soniox.com/speech-to-text

Can you turn Linux kernel source code into a bedtime story? You can if you have an expressive TTS. We took one of the driest things imaginable and gave it pacing, emotion, suspense, and character. The same text can become a narration, a character, a lesson, or a story, simply by changing how it’s delivered. With Soniox TTS v2, you can shape that delivery with audio tags for style, emotion, speed, and vocalization across 60+ languages. Try Soniox TTS for yourself: soniox.com/text-to-speech

Soniox STT pricing is built differently. Instead of simply charging for every second an audio stream is open, Soniox charges based on input and output tokens. That means silence and pauses cost less than dense speech. For real-world conversations, where people naturally pause, listen, and think, the effective cost can be significantly lower than our listed price of $0.12/hour. In many production workloads, it comes out closer to $0.10/hour. More efficient inference should translate into more efficient pricing. Try it on your own audio: soniox.com/compare-stt #VoiceAI #SpeechToText #ConversationalAI #VoiceAgents

We added Grok STT to Soniox Compare. @grok any comments? On the same audio under the same conditions, Soniox STT comes out as more accurate on language identification, speaker diarization, endpoint detection, and transcription continuity, all while costing less. See the difference yourself: soniox.com/compare-stt?provi…

We’re excited to share that @krispHQ has selected Soniox as one of its speech-to-text providers. Krisp has been building real-time voice technology since 2017. Today, its technology is deployed on more than 200 million devices and processes more than 80 billion minutes of voice conversations every month. Few companies have as much experience solving real-world speech and audio problems. That makes Krisp’s decision to choose Soniox especially meaningful. Soniox provides the speech recognition behind Krisp’s products, with native-speaker accuracy across 60+ languages, ultra-low latency, and reliable performance on real-world audio. This includes high precision for names, numbers, IDs, and other critical information. Huge thanks to Karen Movsisyan, Simon Asryan, Davit Baghdasaryan, and the broader Krisp team for the partnership. Excited for what we’ll build together.