Free · No sign-up · 7 tools never upload your audio
Turn a text prompt into a full song with vocals, or split any song into stems. No sign-up, nothing to install, no cost.
1 free song every day · 3 free AI separations a day · nothing to install · see all twelve tools
Two different jobs, both of them one page and one click away. The first writes music from a description; the second reverses a finished mix back into its parts.
Generate
Type a sentence, get a full song — vocals included. 1 free song every day.
Separate
Upload any song, get the instrumental and the isolated vocals back. Files deleted after processing.
Both clips are real output from the tools on this page — a song the generator produced from a text prompt, and a vocal separation run on a complete mix. They are trimmed to 45 seconds so the page stays light; inside the tools you get the full-length files.
Generated from a prompt
One description, one to two minutes, and the generator returns a finished track: sung vocal, full arrangement, mixed and ready to play. Nothing to stitch together from loops, no session to arrange.
Generated track · 45s excerpt
AI Music Generator output
Before and after
Same song, same passage. Play the first clip to hear the original mix, then the second to hear what came back after one pass through the Vocal Remover. The isolated vocal stem is in the tool too, if you would rather keep the singer and lose the band.
Before
Original mix, vocals intact
After
Instrumental (vocals removed)
Twelve tools in total, all free, none of them asking for an account. Seven are analyzers and processors that never send your file anywhere: the maths runs in your browser with the Web Audio API. The AI tools need far bigger models, so they upload once and delete as soon as they are finished.
Vocals, drums, bass and other — separated into four files.
Pull the lead vocal out on its own, clean.
A karaoke-ready backing track with the vocals gone.
The tempo of any song, accurate to ±2 BPM.
Musical key and scale, plus the Camelot code.
Slow down or speed up without changing the pitch.
Transpose in semitones without changing the tempo.
BPM, key, loudness and spectrum in one pass.
Tap along to whatever is playing in the room.
Rotate a song around your head in 8D.
Three steps, whichever tool you pick. Browser-based tools skip the wait in step two entirely — there is no upload, no queue, and no round trip.
No trial that expires, no export limit, no watermark from us, and no paid tier hiding the part you came for. Every tool here is the whole tool.
There is no account to create and no email to hand over. Open a tool, use it, close the tab — nothing here ever needed a login to work.
Seven tools process the audio on your own device — watch the network tab and see for yourself. The AI tools upload once, process once, and delete the file the moment the result is ready. We do not store your audio, share it, or train on it.
We publish what our detectors get right and where every analyzer, ours included, gets it wrong — with a test set you can reproduce on your own tracks. The demo clips on this page are real output from these tools.
Precision claims are easy to make and hard to check, so here are the ones we are willing to be judged on — and the write-up that lets you reproduce the tempo measurement on your own tracks.
The accuracy figure comes with the caveats that matter: octave errors on halftime material, rubato drift on live recordings, and sparse textures with no periodic energy to lock onto are where every detector struggles, ours included. We wrote all of that down, along with a five-track test set you can run in ten minutes — read the accuracy write-up.
Pull the vocal out of any track and keep the band. The instrumental comes back ready to sing over, with the original vocal available as a separate stem if you want a guide track.
Get the tempo and the key before you cue the track, then read the Camelot code to pick a harmonically compatible next song. Both numbers come out of the same pass.
Split a finished mix into vocals, drums, bass and everything else, so you can rebuild it, flip it, or lift one part straight into a new project.
Need a bed for a video, a podcast sting or a game jam build? Describe it, generate it, download it — without hunting for a royalty-free track you are actually allowed to use.
Yes. Every tool on this page works without payment, and there is no paid tier waiting behind the free one. Analyses, conversions, stems and generated songs all come back with no watermark from us and no length cap beyond the limits we state up front: uploads up to 20 MB, one free generated song a day and three free AI separations a day. Those limits exist because neural-network inference costs real money on every run, not because a better version is being held back.
No. There is no sign-up, no email address, no login and no verification step — open a tool and start working. There is no user profile either: your daily count is just a number in a small database, tied to a random identifier our server hands your browser and to your network address, and it holds nothing that identifies you. Clearing cookies gives you a new identifier but not a new allowance, because the same limit applies per network address as well.
Seven tools never upload it at all: the BPM finder, key finder, song analyzer, speed changer, pitch shifter, 8D converter and tap tempo counter all run in your browser with the Web Audio API, so the file never leaves your device — you can confirm that yourself in your browser's network tab. The AI tools (vocal remover, stem splitter, acapella extractor and instrumental maker) do upload it, because the neural networks are far too large to run in a tab. Your file travels over HTTPS, is processed once, and is deleted from our storage the moment the result is ready. We do not store it, share it with anyone other than the inference provider running the model, or train on it.
You keep all rights to the audio you process, and we claim no ownership over your uploads or your results — we don't add a watermark to what comes back. The limit sits on the input side: only process audio you have the right to process. For tracks you generate from a prompt, the free tier licenses them for personal and non-commercial use: listening, practice, study, and sharing in situations where you aren't paid for it. Putting one into a monetised video, stream or music release is commercial use and needs a commercial licence, which comes with a paid plan when it launches. A handful of platforms and broadcasters also ask you to disclose synthetic audio, so check the rules of wherever you plan to publish. The Terms of Service carries the full wording.
MP3, WAV, FLAC, OGG, M4A and AAC, up to 20 MB per file — roughly eight minutes of 320 kbps MP3, or about two minutes of uncompressed WAV. The browser-based tools are limited only by your device's memory, since the file is decoded locally. Tap Tempo needs no file at all: press the pad or the spacebar along with whatever is playing around you. Downloads come back as MP3 from the AI tools and standard WAV from the browser processors.
LoveMusic is a free online music toolbox that runs in your browser. Generate a full song with vocals from a text prompt in the AI music generator, then take any track apart with the vocal remover, stem splitter, acapella extractor and instrumental maker. For everyday listeners and working musicians the analyzers do the quieter jobs: the BPM finder reads tempo in seconds, the key finder returns the musical key and its Camelot code, the song analyzer combines tempo, key, loudness and spectrum in a single pass, and tap tempo answers “what tempo is this?” when all you have is your ears. The processors cover the rest of the workflow — slow a track down without changing its pitch, transpose it into a singable key, or push it into 8D spatial audio. Twelve tools, no account, no watermark from us, and nothing uploaded that can be done locally.