How we picked thirty-one voices

A voice is the first thing you'll know about Sonora. Before the features, before the design, before any of it, you'll hear it say good morning. So we spent a long time making sure it sounded right.

Here's how we went from one voice to thirty-one, and what we learned about why people pick the ones they do.

Why not just one voice

Early on, we assumed we'd ship a single great voice. Pick the best one, polish it, done.

Then we started testing with real people, and the assumption fell apart. The voice one person found warm, another found slow. The one that sounded friendly to us sounded fake to someone else. There was no single best voice. There was only the right voice for you.

So we changed the goal. Instead of one voice everyone tolerates, we wanted enough range that everyone finds one they love.

What makes a voice feel human

The hard part isn't making a voice sound clear. Modern systems do that easily. The hard part is making it sound like it means it.

We focused on the small things that your ear notices but your brain doesn't name:

• Pace. How fast it talks, and where it slows down.

• Pauses. A tiny breath before answering makes it feel like it's thinking, not reciting.

• Warmth. The difference between "your timer is done" said like a fact and said like a person.

• Rises and falls. Flat delivery is the fastest way to sound robotic.

We tuned every voice across all of these, not just the accent or the pitch. A voice can have the right accent and still feel dead. We were chasing the opposite of dead.

Picking the thirty-one

We started with far more than thirty-one and cut hard. A voice made the final set only if it passed three tests:

• It had to sound good saying ordinary things, not just demo lines. "Add milk to the list" is the real job, not poetry.

• It had to hold up over time. A voice that's charming for a day and grating by week two didn't make it.

• It had to feel like a person you'd want in your home, every day, in the background of your life.

The thirty-one that made it cover a wide range on purpose. Different ages, energies, accents, and moods. Some are bright and quick. Some are calm and low. There's no "default" we're nudging you toward. They're all meant to be somebody's favorite.

The one that surprised us

The voice that tested best overall wasn't the most polished or the most upbeat. It was a calm, slightly low voice that took its time. People described it as "steady," "easy to be around," "like someone who's actually listening."

That told us something about what people want from a device they talk to all day. Not the most impressive voice. The most comfortable one.

Picking yours

When you set up Sonora, you'll hear a handful of voices and choose the one that feels like home. You can change it any time. And we'll keep adding more, because the right voice is personal, and personal taste keeps growing.

The voice is where Sonora stops being a gadget and starts being something you talk to. We didn't want to get that almost right. We wanted you to hear it and think, yes, that one.