Master, Repair & Deliver
Your Tracks - Offline.
PumpyDumpy2AudioMaster is an offline desktop app that masters your music to streaming loudness, repairs vocals, separates stems, and packages release-ready files. Drop a track, pick a preset, export. The AI runs on your own PC - your music never leaves your computer.
Windows 10/11 · single installer · free with a watermark on exports
Your Track Sounds Great - Until It's On Spotify
AI music tools and home setups make finished songs in minutes, but the master is often too quiet, the vocals harsh, and the files not ready to publish. PD2AM closes that last gap.
Loudness that travels
Hit streaming-ready loudness (around -14 LUFS) with safe headroom, so your track sits at the right level on Spotify, YouTube, and Apple Music instead of sounding weak.
Vocals that sit right
Tame harsh sibilance, clicks, breaths, and AI artefacts with a gentle repair pass that blends most of the original back in, so the music stays faithful.
Files ready to publish
Export WAV and MP3 masters, embedded cover art, lyrics, stems, and a README in one delivery pack - everything a release needs, packaged in one click.
RACK, FIX, AI - One App, Three Jobs
Switch between mastering, repair, and AI delivery with a single click. The same track flows through all three.
Mastering Rack
An AI assistant analyzes your track - loudness, peaks, dynamics, spectral balance, stereo width, BPM, and key - then suggests a one-click preset like Gentle Polish, Balanced Master, or Smart Master. Build your own chain with Normalize (LUFS), reference matching, EQ, multiband compression, stereo widening, and format conversion. Preview 15 seconds, A/B against the original, then export.
Fix & Repair
Trim and fade with sample accuracy, add silent tail padding, and export just a clip when you need it. The Vocal Repair panel offers de-esser, declick, dropout fill, breath gate, AI vocal denoise, and a tilt-EQ vocal preset - each with its own strength slider. Compare two versions side by side to hear exactly what changed.
AI Delivery
Read ID3 tags, embedded cover, and lyrics straight from your Suno/Udio MP3, align lyrics to a perfectly timed LRC with Whisper, and export a full delivery pack: WAV and MP3 masters for Spotify and YouTube, cover art, Demucs stems, a PD2V bridge file, and a README. Need a longer cut? The Track Extender loops a section seamlessly or generates new audio with AI MusicGen.
Everything You Need Between "Done" and "Published"
Mastering, repair, analysis, and packaging - built for music creators, not audio engineers.
AI Assistant & presets
Analyzes the track and detects its style, then proposes one-click mastering presets tuned to the genre - no guesswork.
Track diagnosis
Read loudness, peak headroom, dynamics, spectral balance, stereo image, vocal position, mud, tempo, and key at a glance.
Mastering pipeline
Normalize to LUFS, match a reference, shape EQ, multiband compress, widen stereo, crossfade, convert format, and tag ID3.
Vocal repair
De-ess, declick, fill dropouts, gate breaths, denoise with AI, and tilt-EQ vocals - each with an adjustable strength.
Stem separation
Split a finished track into vocals, drums, bass, and other stems with Demucs - perfect for remixes and clean re-edits.
Metadata & cover
Read and edit ID3 tags, embedded cover art, and lyrics, then align lyrics to a timed LRC file with Whisper.
Delivery pack
One click exports WAV/MP3 masters for Spotify and YouTube, cover art, stems, a PD2V bridge, and a README summary.
Track extender
Reach a target length by seamlessly looping a section, or generate new audio with AI MusicGen (experimental).
Real AI Models, Running On Your PC
PD2AM bundles open-source AI models so everything runs locally. A capable NVIDIA GPU makes them fast; a CPU fallback keeps them working.
Demucs - stems
Separates a mix into vocals, drums, bass, and other instruments for remixes, clean-ups, and re-edits.
Whisper - lyrics
Transcribes and time-aligns vocals so you get a perfectly synced LRC lyrics file from any song.
DeepFilterNet - denoise
Removes background noise and hiss from vocals and recordings while keeping the voice natural.
MusicGen - generate
Generates new musical audio to extend a track. Experimental, and for personal / non-commercial use only.
Note: most modules (Demucs, Whisper, DeepFilterNet) suit general use. The MusicGen model is licensed for non-commercial use only (CC-BY-NC), so AI-generated audio from MusicGen is meant for personal or experimental use. Loop-based extension and everything else stays free for your own releases.
Free to Use. Pro Removes the Watermark.
PD2AM is a companion to PumpyDumpy2Visual and shares the same Pro license. One subscription unlocks both apps.
Free
Everything you need to master and repair - exports carry a light audio watermark.
- Full mastering rack, presets & track diagnosis
- Vocal repair, trim, and A/B compare
- AI stems, metadata, cover & LRC lyrics
- Loop-based track extension
Pro
Remove the watermark and unlock commercial use in PD2AM and PumpyDumpy2Visual with one subscription.
- No watermark on exported audio
- Commercial use license
- Unlocks PumpyDumpy2Visual Pro too
- $20/month or $200/year, cancel anytime
Already a PumpyDumpy2Visual Pro subscriber? Your existing key activates PD2AM too - just sign in inside the app.
Get PumpyDumpy2AudioMaster
A single Windows installer. Install, drop a track, and start mastering.
PumpyDumpy2AudioMaster v3.1.1
The download is large because the AI models (PyTorch, Demucs, Whisper, MusicGen) are bundled so everything runs fully offline.
The installer is unsigned, so Windows SmartScreen may warn on first run - choose "More info → Run anyway". macOS and Linux are not supported yet.
What You Need to Run It
Windows 10 / 11
64-bit Windows 10 or 11. macOS and Linux builds are not available yet.
NVIDIA GPU recommended
An NVIDIA GPU (CUDA) makes the AI models fast. No GPU? A CPU fallback still works, just slower.
~6 GB disk space
The app and bundled AI models take roughly 6 GB once installed.
8 GB+ RAM
8 GB of RAM is the comfortable minimum; 16 GB helps with longer tracks and stem separation.
Questions About PD2AM
Is PD2AM the same as PumpyDumpy2Visual?
No - they are two apps that work together. PumpyDumpy2Visual turns your music into a video; PumpyDumpy2AudioMaster masters, repairs, and packages the audio itself. One Pro subscription unlocks both.
Does my audio get uploaded anywhere?
No. All processing and AI inference happen on your own PC. Your tracks never leave your computer, which is why the AI models are bundled into the installer.
Why is the download so large?
PD2AM ships full AI models (PyTorch, CUDA, Demucs, Whisper, DeepFilterNet, MusicGen) so it runs completely offline. That makes the installer around 3 GB, but you never depend on a cloud service.
Can I use the results commercially?
Yes - mastered audio, stems, repaired vocals, and delivery packs are yours to release. The only exception is audio generated by the MusicGen model, which is non-commercial (CC-BY-NC). Loop-based extension is unrestricted.
What does Pro unlock?
Pro removes the audio watermark and grants a commercial-use license. The same Pro subscription also unlocks PumpyDumpy2Visual, so you get both apps for one price.
Do I need a powerful computer?
An NVIDIA GPU makes the AI features fast, but PD2AM falls back to your CPU when no GPU is available. Mastering and repair work fine on a normal machine; stem separation is the heaviest task.
Guides for Every PD2AM Job
Want the detail on one task? These guides cover exactly how PD2AM masters, repairs, separates, and packages your audio.
Mastered the Audio? Now Make the Video.
PumpyDumpy2Visual turns your finished track into an audio-reactive music video - 180+ templates, 90+ visual objects, and fast offline export. Together with PD2AM it takes you from raw track to published release.