Skip to main content
All toolsYour files stay on your device

Free · No signup

Vocal remover & stem splitter

Make an instrumental track or separate vocals, drums, bass and other instruments.

Checking browser…

Add your music

Start with a song or the soundtrack from a video.

Choose audio or video

Split up to 3 minutes · Trim longer recordings here

MP3, WAV, FLAC, M4A, MP4 and more · Prepare files over 200 MB · Files stay on your device

Add a file to get started.

Frequently asked questions

Is this vocal remover free, and does my file get uploaded?

It is free with no signup. Your audio or video stays on your device; separation runs in your browser. An internet connection is needed to load the tool and download the AI model.

Which AI model does Fuzz Puppy use?

Fuzz Puppy uses SCNet, a music separation model that estimates vocals, drums, bass and other instruments together. It also creates an instrumental by subtracting the estimated vocals from the original audio. The model download is about 13 MB and runs locally in your browser using your GPU when available, with a CPU fallback. All five tracks are created automatically, so you can preview and download the ones you want.

Can I separate drums, bass, guitar or piano?

You automatically get vocals, drums, bass and other instruments as four separate WAV files, plus an instrumental mix combining all instruments without vocals. Guitar, piano, keyboards and similar instruments are grouped in the other instruments track; they cannot be isolated individually. Preview each track and download only the files you want.

Do I need to choose which stems to create?

No. Fuzz Puppy creates all five files automatically. The model runs once for each section of your recording, with no separate model run for each instrument. Creating additional tracks does add audio reconstruction, memory and WAV-export work, but you do not need to repeat separation just to hear another available stem.

How does Fuzz Puppy compare with Moises for stem splitting?

Moises combines stem separation with a broader music workspace and offers dedicated guitar, keys and vocal categories depending on your plan. Fuzz Puppy is a simple alternative for splitting music directly in your browser without an account, media upload or subscription. Fuzz Puppy automatically creates vocals, drums, bass, other instruments and an instrumental mix; preview each track and download the WAV files you want for free. Fuzz Puppy keeps guitar and piano grouped with other instruments and processes up to 3 minutes per run, with inline trimming for longer recordings. Processing happens on your device, so speed depends on your computer.

How is Fuzz Puppy different from LALAL.AI?

LALAL.AI's web splitter offers more instrument-specific separation types, a choice of output formats, and vocal reverb reduction. Fuzz Puppy keeps separation on your device and lets you preview and download every available WAV track for free, without an account or processing credits. Guitar and piano remain grouped with other instruments in Fuzz Puppy. Results vary by recording and model, so compare the same excerpt before deciding which tool suits your music.

How is Fuzz Puppy different from BandLab Splitter?

BandLab Splitter combines stem separation with practice features such as looping and changing key or tempo, and connects to the wider BandLab music workflow. Fuzz Puppy is a focused browser tool for creating downloadable stereo WAVs without signing up or installing an app. Processing stays on your device, and you can trim a longer recording here before splitting a section.

What files can I use, and what do I download?

Choose an audio or video file. Split up to 3 minutes at a time; longer recordings can be trimmed here using the waveform, start and end handles, and selection preview. Files over 200 MB and surround audio can be prepared as a smaller stereo WAV copy. Common formats include WAV, MP3, M4A, FLAC, MP4, MOV and WebM, depending on your browser's decoder. Results are stereo, 44.1 kHz, 32-bit float WAV files. Video input produces audio files, not an edited video.

What affects processing speed?

The tool uses WebGPU when available and otherwise runs on your CPU. Keep the page open during separation; shorter sections use less memory and finish sooner.

Why does the first separation take longer?

The model must download and initialize before separation starts. We cache the verified model file when browser storage is available, so later runs can reuse it. Processing time depends on your computer and clip length. Keep this tab open and visible; you can cancel a run at any time.

Why can I still hear vocals or instruments in a separated track?

Separation is an estimate. Reverb, backing vocals and instruments that overlap the voice can leave traces or cause artifacts. Preview the results before downloading. For difficult songs, try a cleaner source recording and check a short excerpt before splitting the full section.