Direct line, no ticket queue.

Chord recognition is a probabilistic model, not a transcription oracle. Tracks with heavy distortion, dense drum-only sections, vocal-only passages, or atypical voicings (sus chords on a piano left hand, slash chords, jazz extensions outside the 170-class vocabulary) can confuse it.

Two things to try: (1) flip on Advanced mode — it runs an 18-pass ensemble with pitch shifts, tempo stretches and EQ cuts, then votes on the result, which fixes most edge cases. (2) Look at the per-segment agreement score; anything below ~60% is a soft hint the model isn't sure and you should trust your ear over the label.

There's a dedicated preset for it. Open the Tuner, tap the instrument picker, and under Guitar pick the tuning you want: Standard, Half Step Down (E♭), Drop D, Double Drop D, DADGAD, Open G, Open D, Open E, or Open C. Bass has 4-String Standard, 4-String Drop D, 5-String, 6-String. Ukulele, mandolin, banjo, violin and cello also have their own presets.

The preset retargets the string-by-string readout and the neighboring-string hints. If you want the raw unconstrained readout instead — say you're experimenting with a tuning that isn't in the list — switch to Chromatic, which shows whatever pitch the mic hears with no filtering.

Yes. Everything runs on-device — chord recognition, tuning, chord library, metronome, spectrogram. After install, you can put the phone in airplane mode and nothing changes. There is no server to call and no model to download at runtime. On Android the build pipeline even fails CI if the INTERNET permission is added, so the offline guarantee is enforced.

The current builds target modern mobile audio APIs so the chord model, tuner, metronome, file picker, background audio, and on-device processing can stay stable without a backend, login, or tracking layer.

Supporting older OS versions would mean maintaining separate audio paths or weakening the performance and privacy guarantees. Older-device support may be revisited after the core app stabilizes.

On iOS: anything AVFoundation decodes — mp3, wav, m4a/aac, aiff, caf, flac, alac. On Android: mp3, m4a, wav, ogg, flac, opus.

DRM-protected files (e.g. tracks downloaded from Apple Music or YouTube Music with a subscription) cannot be processed — that's an OS-level restriction, not something the app can work around.

Yes. Background audio is enabled on both platforms. Lock the screen, switch apps, take a call — the click keeps landing on the beat. On iOS the metronome integrates with Control Center and the Dynamic Island so you can pause from the lock screen. On Android it runs as a foreground media session for the same reason.

The picker uses the platform's file provider — Files on iOS, the Storage Access Framework on Android. If a song lives only inside a streaming app (a subscription stream, an iCloud Music Library entry that hasn't been downloaded, a YouTube Music cache), it isn't visible to those system pickers and Bravo Nota can't see it either.

Workaround: download the file to local storage first, or use the platform Share Sheet from the source app and pick "Bravo Nota" as the destination.

Yes. After a transcription finishes, tap the share icon above the timeline. Two formats:

Simple — a clean chord sequence like Gm D Cm D7 Gm F7 A♯ F7 ..., useful for pasting into chat or a song-sheet.

Detailed — a tab-separated table with start time, duration, chord name, and confidence, ready to paste into a spreadsheet or feed to a script.

Both go through the system share sheet, so you can copy to clipboard, email, AirDrop, send to Notes, or save to Files.

Bonus tool for when the tuner isn't enough. It runs a real-time STFT on the mic input and lets you tap anywhere on the time-frequency plane to read note + Hz + cents + dB at that point. Peak-on tracking locks onto the loudest partial; a 3-second rolling peak shows what just went past.

Practical uses: chasing intonation drift in choirs and string sections, checking which harmonic is buzzing on an open string, identifying mystery whistles or feedback, eyeballing whether a sustained note is actually sitting in pitch over time.

The app ships a chord-recognition model trained in-house on a private, commercially-cleared dataset. The architecture is a bi-directional transformer over a Constant-Q transform of the audio — the same family as BTC (ISMIR 2019) — outputting a 170-class chord vocabulary (triads, sixths, all sevenths, half-diminished, minor-major7, suspended, slash chords) per ~0.1-second frame.

Advanced mode runs an 18-pass test-time augmentation ensemble (pitch shifts × tempo stretches × EQ cuts) and majority-votes the result. The goal is practical chord recognition that works on a phone without sending your audio to a server.

Bravo Nota is free for personal use because it was built as a real tool for Yahor's family, not as a funnel into ads or subscriptions. The commitment is simple: personal transcription stays free forever, and the app stays free of registration, advertising, trackers, and analytics.

Commercial and heavier workflows will be handled separately. Bravo Nota Pro is planned for saved recordings, a personal account, uptempo, downtempo, pitch up, pitch down, and expanded workflow features. Bravo Nota API is planned for commercial chord recognition at $0.10 per audio minute.

For API access, partnerships, investor conversations, or collaborations, write directly to hello@bravonota.com.