OnKey shows real‑time pitch feedback while a user sings along to a streamed track.
ONKEY MAC APP LETS SINGERS SCORE PITCH AGAINST SPOTIFY TRACKS
*A free, open‑source Mac tool learns a song’s pitch track in minutes and lets users sing along while keeping audio and voice private. Built after a popular karaoke app flooded its library with non‑original tracks, OnKey targets beginners seeking accurate feedback.*
A developer frustrated with a popular singing app that flooded its catalog with non‑original tracks turned to an AI agent named Claude Code. The agent produced a Python script that displayed pitch data in a terminal, but it could not capture the singer’s real‑time pitch. Determined to close the gap, the developer built OnKey, a macOS‑only solution that learns a song’s pitch track after a single listen.
OnKey runs on Apple Silicon Macs (M1 or later) with macOS 14.2 or newer. It taps into Spotify, Apple Music, or YouTube streams, extracts a short Shazam fingerprint, fetches lyrics from LRCLIB, and never saves the source audio or the user’s voice. The result is a free, MIT‑licensed app that offers beginner‑friendly scoring while keeping every note private.
What is OnKey and how does it work?
OnKey is a macOS 14.2+ application for Apple Silicon Macs (M1 or later). The user points the app at a Spotify, Apple Music, or YouTube stream; OnKey captures a single playback, extracts a short Shazam fingerprint, and queries song metadata and lyrics from LRCLIB. It then spends a few minutes learning the song’s pitch track. Once learned, the app plays back the pitch track while the user sings, displaying real‑time scoring in the terminal. The software is free, open source under the MIT license, and does not write the source audio or the user’s voice to disk.
How does OnKey protect user privacy?
OnKey’s network traffic is limited to a short fingerprint used to identify the song and a request for lyrics. The app never uploads the full music file or the user’s vocal recording. All processing of the captured audio occurs locally, and no data is persisted after the session ends. This design ensures that the information a singer expects to stay private—both the backing track and their voice—remains on the device.
What limitations does OnKey have?
The app struggles when a song’s backing is noisy; pitch extraction can miss the correct line. Occasionally it misplaces a note by an octave. Compatibility is restricted to Apple Silicon Macs running macOS 14.2 or later, because newer drivers are unnecessary beyond that point. Scoring is calibrated for beginners, offering a forgiving assessment rather than professional‑grade precision.
What alternatives existed before OnKey?
During development the creator discovered Nightingale, a similar pitch‑learning tool that required a local library of exported songs. In 2026, maintaining such a library is impractical for most users, limiting Nightingale’s appeal to a niche audience. OnKey removes that barrier by learning directly from streamed audio, aiming to help “normies” improve without a pre‑built song collection.
Frequently asked questions
How does OnKey identify songs without storing audio?
It sends a short Shazam fingerprint to look up song details and retrieves lyrics from LRCLIB, but never uploads the full music file.
Which platforms can run OnKey?
OnKey runs on Apple Silicon Macs (M1 or later) with macOS 14.2 or newer and works with Spotify, Apple Music, and YouTube in a browser.
What are the main weaknesses of OnKey?
Noisy backing tracks can confuse pitch extraction, it may misplace notes by an octave, and it is limited to beginner‑level scoring.
OnKey demonstrates that privacy‑first music tools can be built without sacrificing usability. By open‑sourcing the code, the creator invites scrutiny and community contributions, potentially spurring a new wave of secure, low‑cost audio analysis apps. As more singers seek accurate feedback without surrendering their recordings, OnKey may become a reference point for future privacy‑centric audio software.
Sources: Hacker News