Say 67 Mic Privacy Explained: What Gets Heard, What Does Not

The first time a game asks to use your microphone, a small alarm goes off in most people's heads, and that is a completely reasonable reaction. So before you shout a single "six seven" into your phone, here is exactly what Say 67 does with your voice, in plain language.

The permission prompt

When you pick mic mode and press go, your browser shows its normal microphone permission dialog. This is the same system-level prompt any site would show for any reason, video calls, voice notes, whatever. Say 67 cannot bypass it and cannot access your mic silently in the background. If you say no, or if you never had a mic to begin with, the game quietly falls back to silent tap mode so you can still play.

What actually happens to the sound

Once access is granted, the game reads a live audio stream and measures its loudness in real time, dozens of times per second. That measurement is a single number: how loud is the room right now. It is not a transcript, not a recording, not a waveform saved anywhere. The moment a measurement is taken and used, it is gone. There is no buffer of your last ten seconds sitting in memory waiting to be uploaded.

This is what "analysed on your device only" actually means in practice. The analysis is a bit of math running inside your browser tab, using a standard web audio feature built for exactly this kind of live signal processing. Nothing about your voice, your words, or your background noise travels anywhere else.

Why calibration comes first

Before your run starts, mic mode asks you to stay quiet for about a second. This is not the game secretly recording a voiceprint of you, it is simply learning your room's baseline noise floor. A quiet bedroom and a noisy kitchen with a dishwasher running produce very different resting loudness levels, and the game needs that baseline so it can tell the difference between "background hum" and "you just shouted six seven."

Skip this step by talking during calibration and the game may misread your normal background noise as a shout later, which is the single most common cause of a confusing first run.

No accounts, no uploads, no transcripts

There is no speech recognition happening here. The game never tries to figure out if you actually said "six seven" versus just making a loud noise at the right moment, it only measures volume and timing. That is a deliberate design choice: it means the whole system works without ever needing to understand or store what you said, only how loud and how fast you reacted.

When your round ends, the only thing that leaves your device is your final score, tied to whatever nickname you picked, the same way any of the other 67 games post a score to the leaderboard. Your voice itself never makes that trip.

If you would rather skip the mic entirely

Some people just do not want an app touching their microphone, full stop, and that is a completely fine position to have. Silent tap mode exists exactly for that case. Same numbers, same decoys, same rounds, same shrinking timer, you just tap the screen the instant 67 appears instead of shouting. You lose the loudness half of the scoring formula, but the reaction half still works exactly the same way, and plenty of players prefer the quieter version anyway.

Curious how the two modes actually compare in practice. Read our breakdown of silent tap mode or head straight into a run. Play Say 67 with whichever mode feels right, and talk it over with other players in the chat.