In the open office
Silently dictate at your laptop. Nobody hears a thing.
The interface to your personal intelligence.
Speech that works everywhere. Voice, whisper, silent.
Everywhere you are, however quietly you need to speak.
On the train
Whisper into your headphones. We separate you from the noise.
In a meeting
Take a note without leaving the room.
At the café
Chat with AI without shouting over the music.
Late at night
Text in bed without waking anyone up.
On the walk
Voice is fine here. Same interface, different volume.
Middle of a conversation
Ask AI something without breaking the moment.
On a plane
Reply to a message without disturbing the row.
Silent Speech · Working today
Talk to AI without making a sound.
4% word error rate personalized. Under 10% zero-shot.
Watch the original on XEarly access
Join the waitlist.
We invite people in small waves. You'll hear from us when it's your turn.
On the go
Whisper. On the devices you already own.
Whisper · Next
Speak quietly. Be heard clearly.
We separate your voice from the room and cut the noise. Prototype running on headphones now. The same capability will run across phones, earbuds, glasses, and XR.
A complete speech interface.
Voice, whisper, silent. On the devices you already own.
Reads what you mouth. Not what you say.
Cuts through the room around you.
Works on the devices you already own.
The shift
Voice is the future. But not everywhere, and not always out loud.
Joanna Stern@JoannaSternI want to interview someone who actually wears one of these soundproof beaks to talk to AI in the office.
Charles Maddock@charles_maddockvoice-to-text has gone too far
Sam Altman@samaagreed feels big, i want a new kind of computer
Alex Finn@AlexFinnI don’t think you understand what ChatGPT Voice unlocks... Work can be done ANYWHERE now.
Joanna Stern@JoannaSternVoice is a very big deal.
Aaron Levie@levieVoice AI is absolutely going to be game changing. Typing is fine, but when you just talk to your device and get an instant answer back, it lowers the barrier to asking questions about anything. It’s the closest thing to sci-fi we’ve had in AI. 0 latency is the key though.
Aaron Levie@levieVoice AI will be fairly game changing for interacting with AI Agents to take actions for you on mobile. Texting agents will often be less convenient than simply using a GUI. But talking to agents can be far more efficient than GUIs. This opens up entirely new software use cases.
a16z@a16zVoice is one of the most frequent and information-rich form of human communication — and for the first time, AI is making it programmable at scale.
Hersh Desai@Hersh_DesaiEver since I started working with @natfriedman and @danielgross in 2022, we had a core belief that voice would be a massive unlock for the future of HCI. We were on the lookout for the team that would cross the uncanny valley and build a generational company in voice AI.
cat@catheryn_liExcited to announce that @simpleailab has raised a $14M seed round led by @firstharmonic. For the past year, we’ve been building AI voice agents to transform direct-to-consumer sales. We fundamentally believe that voice AI is the future of all B2C calls.
Atty Eleti@athyuttamrevoice is the future
janvi kalra@janvikalra_go try this out! voice is the future of how we interact with superintelligence, and gpt-live-1 is a step function improvement in voice intelligence, latency, and safety. twas a blast working with the stacked team on this one
signüll@signulllabsolutely fucking love love this. okay sorry openai, you are doing amazingly cool stuff when it comes to interfaces/modalities. i love voice & it is my primary input method for computers now (therefore it deserves a dedicated mechanic to activate).
Jason Averbook@jasonaverbookVoice won’t be the next interface. It will make the interface disappear. AI that understands accents, noise, jargon and fragments removes the need to translate work into fields and forms. Today: transcription. Next: organizational memory.
sourcery@sourceryy“The keyboard and mouse are slowly dying.” Sector Head at Coatue Max Cook: “The keyboard and mouse, you should throw them away. We’re going to natural language interacting with agentic models.”
Tell us what you want to build.
One adaptive speech layer across phones, earbuds, glasses, XR, and cars. API access is early and invite-based. If you are an OEM, tell us the device. If you are a developer, tell us the app.
See where it fits