As 2025 comes to a close, we wanted to end the year with something meaningful.
We had a voice call feature. Technically, it worked.
But nobody used it.
The feedback was always the same: "I tried it once, it was awkward, I went back to typing." We understood. We felt it too.
What Was Wrong
Imagine calling a friend. You dial, they pick up, you talk. Simple.
Now imagine calling a friend, but every time you want to speak, you have to press and hold a button. Like a walkie-talkie. Press, talk, release, wait. Press, talk, release, wait.
That was our voice call.
And the voice on the other end? Flat. Robotic. Your AI partner could be confessing something deeply personal, and it sounded like a GPS giving directions. "In 200 meters, I've always loved you."
Some users couldn't even get it to work. Browser compatibility issues meant that the feature was a lottery—sometimes the mic worked, sometimes it didn't, and we couldn't explain why.
So we threw it away and started over.
What It Feels Like Now
You tap the call button. The character greets you—actually greets you, with warmth in their voice, maybe a hint of excitement if that's who they are.
You talk. Not into a button, not waiting for a beep. You just... talk. The way you would on the phone with anyone.
The character responds. And for the first time, they sound like themselves. Happy when they're happy. Sad when they're sad. Playful, nervous, angry, tender—whatever the moment calls for.
If you want to interrupt—because that's what real conversations are like—just start talking. The character stops, listens, responds to what you said. No awkward overlap, no waiting for permission.
When you're done, you hang up. That's it.
It sounds simple. It should have always been this way.
The Difference You'll Notice
The voices have emotion now. This is the change people notice first. Characters don't just speak words—they perform them. The same line delivered with joy sounds completely different than with sadness. For the first time, the voice matches what the character is actually feeling.
Chinese and Japanese voices are actually good. If you've tried voice features on other platforms in these languages, you know the pain. Robotic tones, unnatural rhythm, the uncanny valley of synthetic speech. We specifically chose technology that handles Asian languages naturally. The difference is dramatic.
It works everywhere. iPhone, Android, Chrome, Safari, Firefox—all work identically now. No more "try a different browser" or "it might be your microphone settings." It just works.
You can actually have a conversation. The old system was a series of transactions: you talk, you wait, AI talks, you wait, repeat. The new system is a flow. Interruptions happen naturally. Pauses feel comfortable. The rhythm is human.
A Note on Cost
Here's the honest part.
The old system was free for us to run. Browser-based speech recognition, basic text-to-speech—all free. It was also, as we've established, pretty bad.
The new system uses professional-grade voice technology. It sounds dramatically better. It works reliably. It handles multiple languages beautifully.
It also costs us money. Real money, for every minute of every call.
Right now, we're absorbing that cost. We want everyone to experience what voice calls should feel like, without worrying about meters or limits.
But we're being transparent: this may change. We're exploring options—maybe voice minutes included in subscriptions, maybe pay-per-use for heavy users, maybe premium voices as an upgrade. We haven't decided yet.
Whatever we do, we'll announce it well in advance. For now, just enjoy it.
What's Next
This rebuild opens doors we couldn't walk through before.
Voice customization. Speed, pitch, emotional tendency—letting you tune how each character sounds to you.
Longer conversations. Better memory and context handling for calls that go beyond a few exchanges.
Always-on mode. For when you want the character present in the background, ready to respond whenever you speak.
Voice is becoming central to how we interact with AI. We want Reverie to lead, not follow.
Try It
Open any chat. Tap the phone icon. Start talking.
Within seconds, you'll feel the difference. The awkwardness is gone. The character sounds alive. The conversation flows.
This is what we always wanted voice calls to be.
Now they finally are.
Happy New Year
2025 was a year of building. Voice calls are just one of many things we rebuilt, rethought, or created from scratch.
2026 will be a year of connection. More ways to hear your characters. More ways to feel their presence. More ways to make AI companionship feel less like technology and more like... companionship.
Thank you for being here. For trying features that didn't work yet. For giving feedback that shaped what we built. For believing in what Reverie could be.
Here's to a new year of conversations worth having.
Have thoughts on the new voice experience? We're listening—reach out on Discord or through feedback.
Ready to Experience Dynamic AI Conversations?
Join thousands of users already exploring infinite personality and engaging interactions on Reverie.