FAQ

Questions, answered.

Which phones does it run on?

Android 10 and newer. It's tested most on recent Samsung phones. There is no iOS version and none is planned for now.

Do I really need headphones?

Yes, whenever the guide piano or the trainer plays. The phone's microphone has to hear only you, not the app's own audio. Wired or Bluetooth both work; Bluetooth gets a quick timing check so the bars line up with what you hear. Pull the headphones out and the lesson pauses at once.

Does anything leave my phone?

No. There's no account, no cloud and the release build has no internet permission. Pitch and tone analysis, the trained models and the song separation all run on the device, and your results sit in an encrypted database. If you turn on "Save my singing to improve the trainer", the recordings stay on the phone too.

Why did I score 79 when my notes were all in tune?

Strain. When the app hears pressed or strained tone the score is capped at 79 so the level can't be passed by pushing. There's no popup about it; free the voice with a warm-up and sing it again.

It flagged my belt as a fault — I'm a rock singer.

Declare your style before the exercise. Belt, breathiness, twang, a high larynx, straight tone and distortion are only flagged in styles that treat them as faults. In rock and pop they're read as choices.

Why doesn't it tell me when a note is bad?

On purpose. There's no negative feedback for a rough section. Instead, once the same habit is heard twice, a brass card appears with the specific fix, and the trainer plays the difference. Praise is saved for notes that are actually hard for you.

What are the "AI lessons"?

After roughly ten lessons the app has enough evidence to establish a consistent weak area — a particular note, leap, vowel, bridge crossing or habit. It then generates a targeted lesson for it inside your range and keeps it until you've sung it clean five times running.

Can I sing my own songs?

Yes. Import a track from the phone. The vocals are separated on-device, the melody is tracked and the lyrics synced by listening (or from a lyrics file, or by tapping). Then sing it over the backing with the teleprompter.

Who are Ginevra and Rocco?

The two trainer voices. Both are Italian-accented, warm and plain-spoken. The app matches the trainer to your voice by default and you can switch in Settings. The sung examples come from real singers, pitch-shifted into your key.

Is it medical advice?

No. Passaggio is a practice tool. It never diagnoses a voice problem. If singing hurts or hoarseness lasts, see an ENT or a speech pathologist.

Why isn't it on Google Play?

It's still in development, so builds ship as APKs on GitHub Releases. Public availability will be announced here.