Your Breathing Coach, Powered by ML
Nostril uses three layers of machine learning to track your breathing while you sleep, measure your breathing fitness, and help you switch to nasal breathing for good. All processing happens on your device.
Available in 9 languages. iOS 26+. Free to download.
You Breathe 20,000 Times a Day
Most of them through your mouth, without knowing it. Mouth breathing causes poor sleep, low energy, dry mouth, snoring, and anxiety. Nostril proves it with data and helps you fix it with science.
Every Screen, Designed to Help
From overnight tracking to daily check-ins, every detail is intentional.
Dashboard
Sleep Tracking
Breathing Reset
Progress
Sleep Analytics
Not Another Meditation App
Real machine learning. Real breathing science. Real results.
Three-Layer ML Fusion
YAMNet deep learning, Apple SoundAnalysis, and custom FFT spectral analysis work together to classify every 5-second audio window while you sleep.
Sleep Breathing Tracker
Place your phone on the nightstand. Wake up to a full report: nasal vs mouth breathing percentage, snoring episodes, intensity, and duration.
100% On-Device Privacy
Zero audio is recorded, stored, or uploaded. All ML inference runs locally on the Neural Engine. Your breathing data never leaves your phone.
Control Pause Test
Measure your COâ‚‚ tolerance with the clinically recognized Buteyko breathing fitness metric. Track your score over weeks and months.
60-Second Reset
Guided nasal breathing with synchronized haptics. Inhale 4 seconds, exhale 6. Log your mood. See the pattern over time.
Progress & Streaks
Charts, heatmaps, and streaks show your breathing journey. Daily check-in reminders build awareness until nasal breathing becomes automatic.
Built With Serious Technology
Three independent ML classifiers fused into one decision per audio window.
How Sleep Tracking Works
Place your phone on the nightstand
Tap Start and go to sleep. The microphone captures audio in 5-second windows throughout the night.
Three classifiers analyze every window
YAMNet (deep learning on 521 audio classes), Apple SoundAnalysis (cross-validation), and custom FFT spectral analysis (snoring band 100-500Hz, breathing band 500-2000Hz) must agree before labeling a window.
Wake up to your breathing report
Nasal vs mouth breathing percentage, snoring episodes with duration and intensity, trends across nights, and personalized insights.
Audio is never saved
All processing runs on the Neural Engine in real time. Only statistics are stored. No audio is recorded, uploaded, or transmitted. Ever.
Deep Sleep Analytics
Dive into your sleep data with detailed charts showing nasal breathing trends, snoring intensity over time, and sleep duration patterns. See if you are improving, declining, or holding steady.
- Nasal breathing percentage trend
- Snoring episode count and intensity
- Sleep duration tracking
- Personalized insights based on your data
FAQ
How does Nostril track breathing during sleep?
Nostril uses three independent ML classifiers (YAMNet, Apple SoundAnalysis, and FFT spectral analysis) running on-device. All three must agree before classifying each 5-second audio window.
Is my audio recorded or uploaded?
No. All processing happens on your device using the Neural Engine. Zero audio is recorded, stored, or uploaded.
What is the Control Pause test?
A clinically recognized Buteyko breathing fitness metric that measures CO2 tolerance. Most beginners score 10-20 seconds; 40+ indicates healthy nasal breathing.
What languages does Nostril support?
9 languages: English, German, French, Spanish (Spain and Mexico), Portuguese (Brazil and Portugal), Italian, and Chinese.
Is Nostril free?
Free to download with core features. Premium unlocks unlimited sleep history, detailed analytics, and all reminders.
Choose Your Plan
Weekly
$4.99 / week
Subscription managed by Apple. Cancel anytime.
Annual
$29.99 / year
Subscription managed by Apple. Cancel anytime.
Start Breathing Better Tonight
Free to download. Available on iPhone. iOS 26 or later.
Available in English, German, French, Spanish, Portuguese, Italian, and Chinese.