Frequently asked questions
Short answers to common questions. For step-by-step instructions, see the guide.
- Is Karaoke Trainer free?
- Yes. It is free, with no sign-up, no account, and no ads.
- Are my song and my voice uploaded?
- No. The song is separated and analyzed in your browser, and your voice is analyzed and recorded in your browser. Nothing about either is sent anywhere. See the privacy policy.
- Do I have to measure my vocal range?
- No. The range is optional: loading a song, singing, and looking back at how you sang all work without it. If you measure it, each song also shows a suggested key.
- Which files can I use?
- Any audio format your browser can play: MP3, WAV, M4A (AAC), FLAC, Ogg, and others, up to 10 minutes. Use files you bought or recorded.
- Can I use songs from YouTube, Spotify, or Apple Music?
- No. The app only opens audio files on your device; it does not take songs from websites or streaming services (music from streaming services is protected and cannot be opened in other apps).
- Why is there a download the first time?
- The AI model that separates the vocals (about 75 MB) runs on your device, so the browser downloads it once and keeps it. On mobile data the app asks first. ⚙ → “Delete the AI model” frees the space.
- How long does loading a song take?
- On a recent computer, a 4-minute song is usually ready in a minute or two; phones take longer. The bar shows about how long is left. A song you loaded before opens at once from “Recent songs”.
- Does it work offline?
- Yes, once the app and the AI model are on your device. The last 8 songs you loaded are kept, so you can open and practice them offline. Loading a new song offline also works if the model was downloaded before.
- I sing an octave lower than the singer. Is that wrong?
- No. Singing an octave lower or higher counts as on pitch. The app finds your octave in the first seconds you sing and draws your line next to the melody. You can also choose the octave in the menu next to “Sing”.
- What do the colours mean?
- Green: within ±50 cents of the melody; yellow: within ±100 cents; red: further off (100 cents = one semitone). Settings → “How strict” changes these ranges.
- Why do I need headphones?
- With a speaker, the microphone hears the song too, including the original singer, and their voice gets mixed into your line. Wired headphones work best. If you must use a speaker, turn the vocal down to 0% and try Settings → “Echo cancellation”.
- My line is drawn later than the melody.
- Every device takes a little time to play sound and to hear the microphone; Bluetooth headphones take more. Use Settings → “Measure the delay” (say “ta” with each beep), or move the slider until your line sits on the melody when you sing in time.
- The grey line is missing or wrong in some places.
- The melody line is made from the separated vocals. Backing vocals, harmonies, rap or spoken parts, and very breathy or distorted singing can give a wrong or missing line there, and some of the accompaniment can stay in the vocals. Use it as a guide.
- Is the percentage a score?
- It is a rough guide, not a score: the share of the time you were within the green range, counted only where the original singer sings and you sang. The list of parts that went off, and listening to your recording, tell you more.
- Is my recording saved?
- Only if you press “Keep this recording” (it is kept on this device with the song), “Save my voice only (WAV)”, or “Save with the song (MP3)” (files on your device). Otherwise it is gone when you sing again or close the page.
- Can I save my singing with the song?
- Yes. After singing, “Save with the song (MP3)” saves your voice with the accompaniment as one MP3 file. It contains the song, so it is for your own practice only: do not upload it to video or social media sites, share it, or distribute it. The rights to the song belong to its rights holders.
- How is the suggested key chosen?
- From the melody, the app takes its low and high notes (leaving out the lowest and highest 5% of the time, so a single stray note does not count), and finds the key change from −6 to +6 semitones, possibly an octave lower or higher, that fits them best inside the range you measured.
- Does it work on a phone?
- Yes, in a recent browser. Loading a new song takes longer than on a computer; keep the screen on until it is ready. On iPhone and iPad, sound plays even with the silent switch on.
- Which AI model does it use?
- Spleeter (2 stems: vocals and accompaniment) by Deezer, released under the MIT License, in the ONNX format converted by the sherpa-onnx project. It is lighter than newer models, so it also runs on phones, but the separation is not perfect.
- How do I delete my data?
- ⚙ → “Delete saved songs” deletes the songs and kept recordings; ⚙ → “Delete the AI model” deletes the model; “My vocal range” → “Delete” deletes your range. Clearing this site's data in the browser settings deletes everything.