Tiny Tell.
Hear what your baby is telling you.
A private cry analyzer for iPhone — tap to record, get an answer in seconds.
Coming soon to the App Store · iPhone only · Free
Private by design
No account. No cloud LLMs. No advertising SDKs. Audio is processed on a single private server and discarded immediately after classification.
Fast & honest
Custom ONNX model trained on real cry data with phone-mic simulation. When the model isn't confident, it says so — instead of guessing.
Built by a parent
One developer, one app, one inbox. Email me at hafs.darwish+tinytell@gmail.com and I'll reply within 48 hours.
How it works
Three taps, no setup, no signup.
-
1
Open the app, tap record.
No onboarding wall, no account screen. Microphone permission is the only thing requested, and it's only used while you're actively recording.
-
2
Capture a few seconds of the cry.
Tap stop when you're done. The audio sample is sent over an encrypted connection to the private inference server — never to a third-party LLM, ad network, or analytics vendor.
-
3
Read the classification.
The result lands in your history within seconds. The audio is discarded immediately after inference. Your history stays on the phone — there is no server-side log.
What it is. What it isn't.
Honest framing matters more than marketing copy.
It is
- A caregiver-convenience tool.
- A pattern recognizer trained on real infant cries.
- Honest about uncertainty — low-confidence predictions are flagged.
- Free, no in-app purchases, no subscription.
It is not
- A medical device.
- A diagnostic, treatment, or clinical decision tool.
- A substitute for a pediatrician.
- An always-listening monitor — only records when you tap.
If your baby is in distress or you have concerns about their health, contact your pediatrician or emergency services. Tiny Tell is intended for caregiver convenience only.
Under the hood
For the curious. Skip if you just want the app to work.
The model is a custom ONNX classifier trained on real infant cry recordings, with synthetic phone-mic noise augmentation so it generalizes to actual iPhone microphones — not just clean studio audio.
Inference happens on a small private server operated by the developer. Audio is sent over an encrypted connection, classified, and the audio file is discarded immediately. The server does not store, log, or archive audio. No third-party AI service ever touches your recording.
Classification history lives only on your iPhone. There is no server-side history, no cross-device sync, no backup unless you have iCloud Drive enabled at the iOS level. You can clear your local history at any time from the app's Settings screen.
Built by one parent who wanted a private, honest tool.
Questions, bugs, feedback → hafs.darwish+tinytell@gmail.com