Last updated 25 August 2026
Recording yourself is the fastest way to get better and the least pleasant. It is also technically awkward on a phone, for one reason: the microphone cannot tell your voice apart from the music coming out of the speaker two inches away. Here is what actually happens, and what to do about it.
Play a backing track through a phone speaker and sing along, and the recording contains both, and not faintly. In tests on a real phone, roughly a third of what the microphone captured was the backing track coming back in. It is called bleed, and it is why a naive recording sounds like someone singing in a room where a stereo is on, rather than a vocal over music.
It gets worse with distance, and worse fast. Hold the phone at arm's length and the maths turns against you from both ends at once: your voice is further from the microphone, and the speaker is not. Measured on a real take at arm's length, the singer came out roughly 40 dB below the backing. That is not a balance problem you can fix afterwards with a slider. It is a recording with the wrong thing in it.
The clean answer. The backing goes to your ears and never reaches the microphone, so the recording is only you. Nothing to separate, nothing to cancel.
The cost is that singing in headphones is genuinely harder. You cannot hear your own voice in the room properly, so pitch drifts and people push too hard. Wired headphones beat wireless here, because Bluetooth adds a delay between the music and your ears that is small enough to ignore when listening and large enough to be distracting when you are trying to sing in time with it. If you use them, one ear on and one ear off is the old studio trick and it works.
More natural to sing, and it produces a recording with the backing baked into it. The old fix for this was echo cancellation, the same thing that stops you hearing yourself on a phone call. It works, and it takes your voice with it: in testing it cut the vocal by about 6 dB, because a system built to remove one talker's voice from another's line has no notion that the loud thing in front of the microphone is the thing you are trying to keep.
The better answer is to treat the bleed as something to undo afterwards rather than prevent. Homeoke records the microphone raw, then separates that recording the same way it separates a song, into your voice and the room's copy of the music. Your voice is then mixed back over the clean original backing, in the right key, lined up to the millisecond. The bleed stops being a defect and becomes the input: the room's copy of the music is also what tells the app exactly how to align the two.
Practically, that means you can just sing out loud with no headphones, which is what most people want to do anyway.
Recordings stay on the phone. They are not uploaded anywhere, they are listed where you can play, rename, star and delete them, and you can share any of them out as an audio file. Deleting the app deletes them all, so export the ones you want to keep first.
The recording is not the point; hearing it is. Listen once for how it felt and once for what happened: where you ran out of breath, which interval you missed, which word came out flat. Mark those phrases in the lyrics while it is fresh, and the next take starts from what you learned rather than from the beginning.
Free public beta, on any iPhone running iOS 17 or later. No account.
Join the beta on TestFlight