Notes from audio
Basic Pitch is built in. Convert a voice or instrument into editable notes without an extra model download.
Get started
Your first idea
starts with a question.
DawIt is free and in very early access, with macOS as its first platform. Expect rough edges while you explore.
Availability
Build from source
for now.
DawIt’s source code is available on GitHub. Installers are not available yet. Download or clone the repository, then follow the steps below to build and run the app.
You’ll need the DawIt source folder, Git, CMake 3.22 or later, Ninja and a C++20 compiler. On macOS, install Xcode Command Line Tools or Xcode.
Windows and Linux build instructions are included below. The project is cross-platform, with macOS first.
FROM THE DAWIT SOURCE FOLDER
./build.sh --test
./run.sh -- --demo
Build the Release app and run its checks, then open the example song.
Windows and Linux build instructions
Windows
Install Git, CMake and Visual Studio 2022 with Desktop development with C++. From the source folder:
cmake -S . -B build -G "Visual Studio 17 2022"
cmake --build build --config Release
The executable is build/DawIt_artefacts/Release/DawIt.exe.
Linux
Install a C++ compiler, Git, CMake, Ninja and JUCE’s system dependencies. Debian and Ubuntu package names:
sudo apt install build-essential cmake ninja-build libasound2-dev libjack-jackd2-dev libfreetype-dev libfontconfig1-dev libx11-dev libxcomposite-dev libxcursor-dev libxext-dev libxinerama-dev libxrandr-dev libxrender-dev libgl1-mesa-dev libcurl4-openssl-dev
./build.sh --test
./run.sh -- --demo
No account, no subscription.
Music-making and guides run on your computer. Some optional tools need an extra model or program: see what downloads, and why, below.
A first session
Make a small sketch.
Learn one thing.
You don’t have to try every feature. Follow one path, listen to the change, then save something you want to return to.
-
01
Start with something to hear.
Open the example song, or choose File → New Project from Style… for a song in a style, key and tempo you choose.
-
02
Ask why the chords work.
Open View → Theory Guides… (Cmd+G), choose Chord progressions, and look at your chord track. Click an example to hear it.
-
03
Change one thing.
Try a different chord. Draw a melody in the piano roll. Open DawIt EQ and point at a control to see its explanation.
-
04
Keep the idea.
Save the project. Drag a MIDI clip into your full DAW, export a MIDI file, or bounce the mix to WAV when you want to take it further.
Optional tools
Your computer
does the work.
There is no cloud account. Optional model downloads need an internet connection; the processing runs locally.
Stems & karaoke
The Demucs model downloads once when you first use it, after DawIt asks. It separates drums, bass, vocals and the rest.
Words from vocals
Choose a Whisper model that downloads once: Small, 488 MB, or Medium, 1.5 GB. English vocals only.
A local co-writer
Install Ollama separately and add a chat model. DawIt uses it for musical ideas and writing lyrics. The rest of DawIt works without it.
A computer singing voice
Add an English DiffSinger voice yourself. None comes with DawIt. Each voice and its vocoder have terms from their makers.
Plugins for another DAW
Build the thirteen effects and synths with ./build.sh --plugins. Audio Unit is macOS only; VST3 is cross-platform. DawIt Tune stays in the app.
A few useful answers.
Do I need to read music?
No. You can start with chord names, piano-roll notes, pictures and examples you can hear. The guides explain the vocabulary as you use it.
Can I record my own voice or instrument?
Yes. Arm an audio track, choose an input on your audio interface and record. MIDI keyboards can also record editable notes. The recording guide explains levels, monitoring, latency and takes.
Is DawIt a replacement for my full DAW?
DawIt focuses on learning and trying ideas. It has real recording and mixing tools, but a full DAW is the next step for a complete production workflow. You can use the two together.
Does the AI upload my music?
No. The models process on your computer. Ollama runs locally as a separate program. Models may need downloading before their first use.
What languages do the vocal tools support?
Detect lyrics and Sing this support English. A singing voice is a separate download with its own terms.