Introducing Soniox STT v6, the world’s first speech recognition model that knows what your voice agent is saying and automatically filters out the agent’s own voice. Two connected innovations make this possible: → Agent text input gives recognition context from both sides of the conversation, improving its understanding of... user responses. → Agent voice echo cancellation filters out the agent’s playback while continuing to listen to the user, even when they interrupt. The result: fewer false interruptions, fewer clarification loops, and more reliable conversations. V6 also delivers significantly higher accuracy across 60+ languages for voice typing, meeting transcription, multilingual recognition, and real-time translation, with stronger speaker diarization, language identification, endpoint detection, and context customization. One unified speech model. All 60+ languages. The full feature set, with languages freely mixed. Available now for real-time streaming and asynchronous transcription. Explore what’s new in Soniox STT v6:show more

this is already one of the better STT, this is incredible. Great work!

I see a familiar person 👀 @hollllyli hahaha Soniox STT v6 will replace Gpt Voice for me Was watching Avatar the last Airbendender series while talking to Voice and she took the words of what I was watching instead of mine 😭

Strongest STT so far 🚀

so useful, good work

Huge step for personal assistant

Amazing innovation! A big step toward smarter, more reliable Voice AI.

Let's goooooo

This could make voice agents feel a lot more natural. Being able to handle interruptions without picking up their own voice is a big deal, especially in real conversations.

Strongest stt

great stuff yall... stt access to the agents own texts def opens up some interesting stuff

new SOTA voice model is here 🔥

Congrats on launching STT v6. In your tests, how much does agent text input cut errors on short replies like names and numbers compared with audio alone?

good work on this

Love this!! Bravo..

Using context from both sides can improve recognition accuracy.

Feeding the agent's own text in is the bigger half of this. A mumbled digit string means little on its own, but right after "what's your zip code" it's easy to read, and that's exactly where clarification loops start.

This is a major breakthrough!

Interesting

Let’s connect over dm

I've been following voice AI closely, and this is one problem I've wanted to see solved.

Great work @soniox_ai 💪🏼

How well does it handle interruptions when both the user and AI agent speak at the same time?

This is a big step toward making voice AI feel more natural

Impressive innovation solving echo cancellation elegantly improves conversation reliability

finally... game change product for everyone who use voice AI actually congrats, team!

Ok this is veryyy useful tbh

This is a huge leap smart context plus echo cancellation is exactly what voice agents needed!

i remember my dialogues with siri and damn she was not the smartest... amazing to see incredible progress in speech recognition!!

That's a big step for voice tech! 👀👏

But I don’t have the problem being pitched 🤷‍♂️ am I alone here ?

gonna play a bit with it!