Karaoke Night
The vocals step back, the lyrics light up word by word, and every note you sing is scored.
Karaoke and voice practice in your browser
See your voice. Take the karaoke stage.
Pick a song: the vocals step back, the lyrics light up in time, and every note you sing is drawn against the one you aimed for. Or learn your range in about a minute with the Voice Mirror.
The rooms
Whatever you play, there's a room for it.
Singers came first. Pianists, guitarists and drummers now have rooms that listen the same way, and Ear Lab tests your hearing.
Voice Mirror
Sing for a minute. Get your range, your accuracy and a voice twin.
Piano Night
Read it falling, on a staff or on the keys, and loop the bar that won't sit.
Free tools
-
Vocal range test Your lowest and highest note -
Voice type test The voice band your range fits -
Pitch training Sing on pitch, scored in cents -
Vocal remover Split a song into vocals and backing -
Which singer has my range? A famous singer who covers your notes -
Break Glass Hold the note until the glass breaks
Karaoke Night
Drop in a song and the singer steps aside.
The lyrics light up word by word with the note to sing above each one, and every note you sing is drawn against the melody.
Three example songs come ready, with the vocals already separated and the lyrics timed word by word. Try singing Goodbye to Spring or Nothing in the Dark, both by Josh Woodward (CC BY 4.0).
Two of these are free and three are for supporters. See the backgrounds room by room.
The Voice Mirror
Meet your voice twin.
About a minute, in four steps
- Glide up, lowest note to highest.
- Glide back down.
- Hold one note and meet your voice twin.
- Sing five notes back for your voiceprint.
Slide up from your lowest comfortable note to your highest, like a siren, then slide back down. Hold one note steady, and the Mirror names the legendary singer whose range lines up with yours. Then it plays five short notes for you to sing back, any octave counts, and puts your range, accuracy and steadiness on a voiceprint you can share.
Sing it again later in the same browser and your results show how your range changed since your last run.
Sing the Universe: melodies made from stars, pulsars and a black hole.
Illustrative example data
MERCURYPITCH · VOICEPRINT Example
C3 – C5 2 octaves Tenor Example: +2 semitones since last time
- Accuracy
- hit 4 of 5 notes
- Steadiness
- ±9¢ on the hold
- Vibrato
- 5.6 Hz · ±26¢
- Onset
- scoops ~180 ms
Example result
Tap to meet your voice twin
Jam Rooms
One song, up to 12 singers, a lane each.
Open a room and send the link. Pick a singer, drag down their lines, and the lyrics split into parts. Everyone reads the same words, each singer gets a live pitch lane in their own color, and one transport keeps the whole room on the same beat.
Your voice goes to the people in the room: straight to them when your networks allow it, through Cloudflare's relay when they don't. A backing track only ever travels directly, never over the relay.
Piano, guitar and drums
Rooms for the instrument in your hands.
Each room reads the part to you, loops the bar you keep missing, and listens to what you play.
Piano Night
Every note falls toward the key that plays it.
A study is already on the stand when you arrive. Read it as falling notes, a staff or the keys, slow a phrase to half speed, and repeat it up to 100 times until it sits. Play the keys on screen or plug in a MIDI keyboard, and bring your own MIDI file when you're ready.
Guitar Night
Tab that rides a highway to your fret.
Load a recording, a Guitar Pro file or MIDI, and the part becomes a highway of fret numbers. Switch to Grid, Tab, Neck or Sheet when you'd rather read it another way. Tune up, turn on Listening with a room mic, an audio interface or MIDI, and loop the four bars you keep missing.
Drum Night
Find out how late that hit really was.
Play on touch, on your computer keys or on an electronic kit over MIDI. Import MIDI or Guitar Pro percussion and read the part as a pocket guide, from the drummer's seat or as notation. The coach reports how early or late each hit landed and how hard, in milliseconds rather than a grade.
Ear Lab
How small a difference can you hear?
Everything else here listens to you. Ear Lab measures you back: 18 drills across pitch resolution, harmony, melody, timbre and time, each read in cents, milliseconds or decibels, never a percentage. A staircase closes in on the smallest difference you can still tell apart.
The Ear Report turns your mix-ups into a sentence, such as "You answer Fa as Sol on 33% of attempts." The Daily Sprint builds about five minutes from your weakest readings, and the Mercury Index pools your calibrated readings into one figure from 0 to 1000.
Practice and progress
You can't hear yourself improving. You can see it.
Exercises with a guide
Run a scale or sing a melody, and watch your pitch drawn live against each target note. A guide voice plays the note for you. There are nine of them, from Aria's loud piano to Nova's octave below.
Focus Mode clears the screen down to you and the note.
Hear Yourself
Twin Trails lays an older take over a newer one on the same canvas, and neither is stretched to fit the other.
Each take ends on Keep or Discard, and what you keep stays on your device.
Challenges and leaderboards
A weekly challenge to sing, and leaderboards for your league, your friends and everyone.
They need an account, and you choose whether to appear.
Break Glass
The pane rings at a note from your own range. Sing up to it and hold it until the cracks spread. Sixty seconds, free.
Why it's free
A community can't grow behind a paywall.
So singing and practice stay free. Two things are paid: cloud GPU separation runs on credits, and supporters get extra room backgrounds.
Open source, AGPL-3.0 Credits only for Cloud GPU
What leaves your device
- Jam Rooms: your live voice, to the singers in your room: directly, or through Cloudflare's relay when a direct link fails. We don't record it.
- Cloud GPU separation: the song you choose to split, uploaded through our server to RunPod, which makes the stems.
- Your account: your scores, progress and settings, and your song list as titles and sizes, never the audio. Until you sign in, the account is anonymous.
- Device sync: songs go straight from one of your devices to the other, never through a relay.
- Google Drive backup: your songs, sent from your browser straight to your own Drive.
- Voice control: your spoken commands, to your browser's speech service, unless you choose the on-device engine.
- Recorded takes: nothing. They stay on this device.
Questions
Before you join.
Is it really free?
Yes, to sing and practice. Karaoke Night with separation in your browser, the Voice Mirror, the exercises, Jam Rooms and the challenges cost nothing. Two things are paid: cloud GPU separation uses credits, and supporters get extra room backgrounds. See what supporters get.
Do I need an account?
No, you don't need an account. If you create one, you can keep your progress, scores and voiceprint on every device you sign in on, add friends, join the leagues and separate songs on the cloud GPU. Create a free account.
I can't sing yet. Is this for me?
Yes. The Ascent is a seven-week path at about five minutes a day, and 18 short exercises show each note you sing against its target. Progress counts your practice days and streaks, and Hear Yourself keeps your takes on this device, so you can compare your first one with your latest.
What songs can I sing?
Any song you have as an audio file. Karaoke Night looks up its synced lyrics on LRCLIB, or you can add your own. To try it first, sing Goodbye to Spring by Josh Woodward, licensed CC BY 4.0.
Can I remove the vocals from a song?
Yes. On a phone or a computer, Karaoke Night splits a song into vocals and backing in your browser, for free. For the cleanest split, sign in and run it on the cloud GPU for credits. You see the price per song before you choose it. Open the vocal remover.
Can I sing with friends?
Yes. A Jam Room holds up to 12 people in one session, with the lyrics in sync, parts split per singer and a live pitch lane for each voice. Start a Jam Room.
What do supporters get?
Extra backgrounds for the rooms and a supporter badge on your account. Supporting is a one-time donation, not a subscription, and singing and practice stay free either way. See the supporter backgrounds.
Where does my audio go?
Jam sends your live voice to the singers in your room, directly or through a relay, and we don't record it. Cloud separation uploads the song you pick, and voice control uses your browser's speech service unless you choose the on-device engine. Your scores, progress and settings are saved to your account, an anonymous one until you sign in, and your recorded takes stay on this device. Read the privacy notice.
Who are Aria, Echo and the others?
They're the guide voices that play your target note: Aria is a loud piano, Echo a soft one, Harmony adds a third above, Nova an octave below, and Spark plays short plucked notes for timing. There are nine to choose from in the sidebar. Open the app.