Not just "snoring detected." Four snore types, timed and replayable.

"You snore" isn't useful information

If your partner has told you that you snore, you already knew that. What you don't know — and what they probably don't either — is what kind of snore. And that's the part that actually tells you something.

Snoring isn't one thing. The sound is generated somewhere in your airway, and where it's generated changes both what it sounds like and what might help. SomniSense classifies your snoring into four acoustic types every night.

The three types we currently classify (and why those three)

I'm going to be specific about what we're doing, because most apps say "we detect snoring" and stop there. We separate snores by acoustic signature into three categories — the three where the source matters most for what you can do about it.

Palatal — the soft-palate rumble

The soft palate (back of the roof of your mouth) vibrates as air passes. It's the snore most people picture. Worse with back-sleeping, alcohol, weight gain. Often improves with side-sleep, an anti-snore pillow, weight loss. Position-responsive — that's what makes this category actionable.

Tongue-base — the lower, irregular kind

The tongue slides back during deep sleep and partially blocks the airway. Lower-pitched than palatal, more uneven. More associated with breathing pauses than palatal snoring is — worth watching. Often improves with side-sleep, head-of-bed elevation, no alcohol within 3 hours of bed. Also position-responsive.

Epiglottic — quieter but irregular

The epiglottis (cartilage flap above the voicebox) flaps during inhale. The most clinically associated of the three with apnea events. If your timeline shows frequent epiglottic events, that's a sleep specialist conversation, not an at-home intervention.

What we don't break out yet — and why I want to be honest about it

There are two other things our model picks up that we currently don't separate into their own categories on the report:

• Nasal-source snores — the whistle or whine. Usually congestion or narrow nasal anatomy. We'd group these into the "other" bucket today. The reason we haven't broken them out yet is that the intervention path is different — saline, nasal strips, treating allergies, sometimes ENT for nasal polyps — and a position experiment won't move them. If your snoring is mostly nasal, the response from our app today is "see your doctor about the nose," not "try side-sleeping for 14 nights." We'd rather be unhelpful than misleading on this one.
• Heavy breathing without obstruction — the loud breath that sounds like a snore but is just air moving through a clear airway. Often happens after exertion, in dry air, or when you're sleeping on your back without partial collapse. It's not a problem worth flagging.

Both of these will get their own classifications in a future model update. Today, our acoustic confidence on them is lower than I want it to be before I label them publicly. If you ever wonder why a particular event in your timeline isn't categorized as one of the three above, that's almost certainly why.

Why classifying it matters

If you're a palatal snorer and you spend $300 on a tongue-stabilizing mouthpiece, you wasted $300. If you're a nasal snorer and you start side-sleep training, you'll see no change for two weeks and conclude SomniSense is wrong.

Knowing which type is yours doesn't tell you exactly what to do — but it cuts the wrong interventions out of your shortlist before you spend time and money on them.

What you get per night

• Total minutes of snoring detected
• Peak loudness in dB
• % of time in each of the 4 types
• Time-stamped events you can play back (See & Hear Playback)

What this is, and isn't

It's an acoustic classifier validated against in-lab PSG (n=70+ paired nights):

Methodology at /accuracy.

It isn't a diagnosis of where exactly your airway is collapsing. Only an ENT exam can confirm that. Think of SomniSense's classification as a hypothesis to test with Lifestyle Lab™, not a final answer.