Review model sources, parameters, and outputs, then choose the right tool to get started.
Active tools
17
Ready to accept new jobs.
Task groups
4
Separation, transcription, image, and voice.
Zero-credit entries
1
Existing MIDI files can use the zero-credit text-encoding converter.
GPU workflows
18
Complex separation / transcription typically needs GPU.
Task group
Split a song into vocals, accompaniment, or instrument stems for practice, mixing, sampling, and further analysis.
Two-stem tools return vocals and accompaniment; multi-stem tools separate more instruments.
Positioning: five-stem harmony split with harmony vocals, vocals with harmony, harmony accompaniment, clean accompaniment, and lead-vocal cleanup for covers, practice, and detailed mixing
Positioning: open-weight, locally deployable two-stem vocal/accompaniment separation for practice, covers, mixing, and sampling prep.
Standard four-stem references from a finished mix for remixing, practice, sampling, and arrangement analysis
Fixed vocals, drums, bass, guitar, piano, and other stems from one mix
Drum remixing, replacement, transient editing, practice tracks, and drum-focused analysis.
Exploring and identifying more instruments than four-stem or six-stem separation exposes
Trying the MSR Challenge 2025 winner on full songs or degraded audio where restoration matters, not only separation
Task group
Convert full audio or single-instrument clips to MIDI; ideal for transcription, teaching, arranging, and quick drafts.
Use multi-instrument transcription for full songs, a piano-specialist model for solo piano, and the zero-credit converter for existing MIDI files.
Editable MIDI drafts from full mixes, plus cleaner solo/common-instrument inputs such as piano/keyboard, guitar, bass, drums, strings/winds, and vocal drafts
Challenge-winning MIROS route for classical/chamber multi-instrument MIDI drafts and aggregate solo-instrument comparison, without public per-instrument solo rankings
Quick piano MIDI drafts for checking melody, rhythm, chord shape, and timing
Side-by-side comparison with default V2 to judge whether the augmented checkpoint fits the recording better
Quality-first piano MIDI drafts for detailed score cleanup, arrangement, and review
Pedal timing, velocity, and note lengths preserved for score cleanup and DAW playback
Garbled MIDI lyrics, track names, instrument names, markers, cue points, and copyright text cleanup
Task group
Generate design-forward images from text prompts, including layouts where typography and composition matter.
Describe the scene, visible text, and aspect ratio. The recommended quality preset fits most jobs.
Fast text-to-image exploration, moodboards, product concepts, editorial visuals, and creative drafts
One-step semantic image edits while preserving the original structure as much as possible
Text-sensitive posters, covers, logos, packaging, product visuals, and structured design concepts
Task group
Handle vocals, timbre conversion, and cover workflows; well suited for demos, test takes, and character-voice experiments.
Clean, single-speaker or single-singer recordings give the most stable results.
Voice-color conversion, cover demos, and character voice experiments
Short-reference narration, voice tests, and multilingual speech samples
How it works
Run real models with free credits. Tasks run on the server; download outputs when done.
STEP 01
New accounts get free credits; no card required. Pick a tool and run a representative sample first.
STEP 02
Submitted tasks join a queue. Free credits are deducted according to GPU time; failed tasks automatically return those credits.
STEP 03
When complete, download stems, MIDI, images, or converted output directly. There is no subscription and no payment.