STOP, AND IT KEEPS RINGING
Playback speed runs continuously from −2× to +2× on one control. Negative is reverse. And at 0× the sound does not stop. わおん goes on synthesising the spectral frame under the playhead, so the chord keeps sounding while you pick it apart — no racing the clock to hear an inner voice again.
Apple’s own audio units cannot hold at rate zero. The WaoN phase vocoder — pv, from the open-source project — can, and reverse, pause, and seeking all follow from it, so they run into one another naturally.
AS LONG AS THE RECORDING IS
The free version works on the first 30 seconds. After Full Unlock there is no preset duration limit at all — わおん processes any finite file it can open, bounded only by the free storage (not the memory) your device can give it. A full hour of audio is an ordinary case here, not an edge case.
The workspace is built for that. The timeline is addressed in source frames rather than in seconds, and わおん reads from the file only the stretch it is working on, computing waveform, spectrum, and analysis over ranges as you move through it.
ONE SESSION AT A TIME
わおん is session-oriented. You work on one recording at a time, and when the session ends, it ends — open something else and the app moves with you. わおん does not take custody of your work. You do. There is no library to manage and no project format to migrate. Keep what matters by exporting it — MIDI, MusicXML, or the audio you recorded here — saved wherever you want it. What carries over between sessions is only reproducible: spectrogram tiles and detected notes in the system cache, which you can throw away at any time and lose nothing but the time to compute them again.
No server, so no account. No account, so no password. Nothing to sign up for, and nothing to forget.
Analysis runs on your device. No recording is ever uploaded, there is no advertising, and there is nothing standing between you and the recording you opened.
IT BEGAN IN 1996
On 3 December 1996 I started a small program called wav2mid. It read a sound file and wrote out MIDI notes. Two days later the first generation was frozen, and over the years that followed it grew into WaoN — a wave-to-notes transcriber, released under the GPL, packaged by Linux distributions, and used by people I never met. The last release went out in 2015.
WaoN is the inverse of a MIDI player. Instead of turning notes into sound, it turns sound into notes.
わおん is that work, rebuilt for the devices people actually carry. What is new is not the idea — it is that you can now see it. The spectrogram and the notes drawn from it, on the same timeline, at the same moment, where you can take hold of a note and move it.
The open-source WaoN is still free and still on GitHub, as it has always been: github.com/kichiki/WaoN.
