DAW for Windows

Record your vocals. Then mix and master the song.

llmuse is a DAW for recording vocals over a backing track and editing them. You can also write parts for a virtual singer, split songs into stems, use your own plug-ins and let an AI assistant work in the project with you.

Windows 10 and 11, 64-bit

llmuse in Pro mode with an eight-track song: section markers, vocal takes, MIDI parts and an automation lane
153MCP tools for AI assistants
19Vocal presets
12Mastering templates
14Interface languages

Three modes

Use as much of the DAW as you need

Switch modes at the top right of the window. Easy records a song in four steps, Pro is the full editor, and Karaoke is for singing along.

Easy mode

Load a backing track, record, mix and export in four steps across the top. llmuse detects the tempo and key and keeps every take. A genre preset, Auto-Tune and one-click mastering finish the mix.

Easy mode with a backing track, four recorded takes, mix presets and export in one window
The vocal editor after Auto Correct, with the pitch line and the properties of the selected note

Vocals

Record takes, then fix the pitch

Every recording becomes a take. llmuse checks each take, sets its level and tells you what to change next time. Fix notes one at a time in the vocal editor, or use Auto-Tune in real time.

  • 19 vocal presets, each with a de-esser and sibilance control
  • Drag notes by semitones or cents in the vocal editor
  • Denoise, De-Reverb and Clean Up Take for roomy or noisy takes
  • Record harmonies on up to eight vocal tracks
The piano roll with the controller lane menu and a sustain pedal lane under the chords

MIDI and instruments

Write parts in the piano roll or on a staff

The piano roll has lanes for velocity, sustain, sostenuto and soft pedals, modulation, expression and pitch bend. Score View shows the same clip on a grand staff.

  • Step sequencer and a drum machine with six kits
  • Sampler with SFZ import, a synthesizer and the GeneralUser GS SoundFont
  • Quantize with swing, humanize, strum and chord tools
  • Pedals, wheels and pitch bend from your MIDI keyboard are recorded too
  • Arpeggiator before any instrument, synced to the tempo, with latch and chord modes
  • Score pages with title and composer, exported as PDF or printed
The mixer with two track groups, effect chains, sends and a reverb bus

Mixing

A mixer with buses, sends and groups

Every strip has a channel EQ, inserts and sends. Tracks fold into groups, and the compressor and gate can take a sidechain from any track. The main effects are set by dragging handles on a large graph.

  • Compressor, limiter, gate, EQ, saturation, delay and reverb editors with A/B compare
  • Convolution reverb with 15 built-in spaces
  • Pedal effects: wah, rotary speaker, amp simulator and octaver
  • Mix Assistant and reference matching
  • Curved automation between points: ease in, ease out or S-curve, with adjustable tension
  • Automation recording from knobs and faders in Touch, Latch, Write and Trim modes
One-click mastering from the master strip, showing the Streaming template result in LUFS and dBTP

Mastering

One-click mastering with loudness targets

Pick a template for a platform or genre. llmuse renders and measures the song, sets up the master chain and shows the result in LUFS and true peak. Switch between Before and After to compare.

  • 12 templates, from Apple Music at -16 LUFS to EDM at -7 LUFS
  • Staged limiting and clipping, so loud masters sound less squashed
  • Auto picks a template from your genre or from an analysis of the song
The Separate Stems dialog with the built-in HTDemucs model and its four stems

Stem separation

Split a song into stems on your PC

HTDemucs, which comes with the installer, separates vocals, drums, bass and other parts. Separation runs on your computer, so nothing is uploaded.

  • Two stems (vocals, instrumental) or four (vocals, drums, bass, other)
  • HTDemucs comes with the installer; the UVR vocal model is downloaded during setup or in the app
  • Erase Vocals in Easy mode keeps only the backing
  • About two minutes for a three-minute song, depending on the CPU
The Lyrics Library with the best match for the loaded backing track at the top

Karaoke and lyrics

Sing along and keep your lyrics

Open an LRC file or paste lyrics, then tap along to time them. Lyrics you apply are saved to the Lyrics Library, and the best match loads when you open the same backing track again.

  • Lyric subtitles in Easy and Pro mode
  • Search the library and apply lyrics with a double-click
  • A score for pitch, timing and volume in Karaoke mode

More features

Also in llmuse

  • Edit cursor and ranges

    Click to place the edit cursor, drag to select a range across tracks, then cut, copy, ripple delete or split.

  • Tempo and time signatures

    Change the tempo or the time signature at any bar, and ramp the tempo smoothly. The grid, metronome, recording and export follow.

  • Takes and comping

    Loop recording puts each pass on its own take lane. Drag across a lane to use that take there, then flatten when you're done.

  • Clip speed and reverse

    Change speed like tape or keep the pitch, and reverse audio and MIDI clips. Audio is always rebuilt from the original.

  • Track groups

    Fold tracks into groups, nest them, and mute, solo or level them together like a bus.

  • Loop station

    Record loops in time with the song, stack overdubs, then move them to the arrangement.

  • VST3, CLAP and VST2

    Use your 64-bit plug-ins with their own presets, A/B and extra outputs. ARA 2 editing of audio clips is experimental.

  • Free instruments

    Download Surge XT, Dexed, a grand piano and more from their makers. llmuse checks each file before it installs it.

  • Export what you need

    The whole song, selected tracks or clips, combined or one file each, as WAV, FLAC, MP3 or a project.

  • Low latency

    WASAPI shared and exclusive modes and a built-in ASIO host. Auto finds the smallest buffer that doesn't drop out.

  • 14 languages

    Switch the interface language at any time, with full Korean input support.

  • Custom shortcuts

    Start from the llmuse, FL Studio style or Ableton style key map and change any key.

  • Warp markers and audio quantize

    Drag a point in an audio clip to fix its timing, or move every attack to the grid. The pitch stays the same.

  • Clip gain envelopes

    Draw a volume curve inside an audio clip. It stays in place when you trim or split the clip.

  • MIDI learn

    Right-click a knob, fader or transport button and move a control on your MIDI controller to map it.

  • Project and track templates

    Start a song from a template, or save tracks with their instruments and effects to use again.

AI and MCP

Make music with an AI assistant

llmuse comes with an MCP server, llmuse-mcp. Claude and other MCP clients can use it to create tracks, write parts, tune the virtual singer, mix, master and export.

  • 153 tools

    They cover tracks, notes, mixing, mastering, stems, lyrics and export.

  • Live on screen

    With llmuse open, edits show up as they happen, and Ctrl+Z undoes them one by one.

  • Local only

    Connections stay on 127.0.0.1. Before any download, the assistant has to show you the terms and ask.

  • Claude Skill included

    Composition notes, 15 genre guides and a vocal tuning guide teach Claude to work in llmuse.

Example request AI connected

Start a lo-fi song at 80 BPM with soft piano chords, a lazy drum loop and a bass line, then master it for streaming.

  1. new_project80 BPM
  2. add_trackKeys, Drums, Bass
  3. add_chordsFmaj7 Em7 Dm7 Cmaj7
  4. add_drum_patternDrums
  5. add_melodyBass
  6. masteringStreaming
Register with Claude Code
claude mcp add llmuse -- "C:\Users\<user>\AppData\Local\Programs\llmuse\llmuse-mcp.exe"

Setup for Claude Desktop and other clients is in USER_GUIDE.en.md in the install folder.

Virtual singer tuning with a drawn pitch scoop, deep vibrato in Note Properties and a dynamics lane
Score View of the virtual singer melody, with the lyrics under each note

Virtual singer

A singer that sings your lyrics

Put lyrics on notes and a UTAU voicebank sings them in Japanese or Korean. The installer can download free voices from their official sites, so it can sing right after you install.

  • Write lyrics in Hangul, kana or romaji
  • Tune each note: pitch curves, portamento, vibrato, dynamics, breath, gender and brightness
  • Get the free voices Namine Ritsu, Kasane Teto and Song-SoRi during setup or later in the app
  • Import UTAU voicebank ZIPs and folders, including OpenUtau singers

Open Virtual Singer Example on the start screen to hear it.

System requirements

What you need

llmuse runs on most Windows PCs. Use headphones when you record, and an audio interface for the lowest latency.

System
Windows 10 or 11, 64-bit
Installation
For the current user, no administrator rights needed
Disk space
About 270 MB, installer about 160 MB. The free voices and models add about 260 MB
Memory
Stem separation uses about 1 GB for a three-minute song
Audio
Any Windows audio device. ASIO needs the manufacturer's driver.
Plug-ins
64-bit VST3, CLAP and VST2. 32-bit plug-ins aren't supported.
Internet
Only for optional downloads, such as voices, the UVR vocal model and free instruments, and the update check
MP3 export
Windows N editions need the Media Feature Pack

FAQ

Questions and answers

Why does Windows SmartScreen warn me?

The installer isn't code-signed yet, so SmartScreen may warn the first time you run it. Click More info, then Run anyway. To check that the download is complete and unchanged, compare its SHA-256 with the value under Check the file below the download button.

Does it run on macOS or Linux?

No. llmuse runs only on 64-bit Windows 10 and 11.

Can I use my plug-ins?

Yes, 64-bit VST3, CLAP and VST2 plug-ins. llmuse looks in the standard folders and the VST folders FL Studio installs, and you can add your own. FL Studio's own Fruity plug-ins only run inside FL Studio.

Does stem separation upload my music?

No. Stem separation runs on your PC and doesn't use the internet. HTDemucs comes with the installer, and the UVR vocal model is downloaded once, during setup or in the app.

Can I release songs made with the virtual singer?

That depends on the voicebank's terms. The free Namine Ritsu voicebank allows commercial use without credit. Every voice has its own terms, and llmuse shows a summary and a link before you download.

How do I connect Claude?

Install llmuse, then register llmuse-mcp.exe from the install folder as an MCP server in Claude Code or Claude Desktop. The exact command and settings are in USER_GUIDE.en.md in the same folder.

Does llmuse update itself?

No. Once a day it checks a small version file on this website and tells you when a new version is out. It never installs anything by itself, and you can turn the check off in Settings.

Download llmuse

For Windows 10 and 11. No account needed.