cjbayron / autochord Goto Github PK

Hello there, and thanks again for the work!

I was wondering if it would be possible to run some training on the model in order to possibly increase the accuracy.

Would it be possible to know what data set you used and how?

Thanks!

Hi!

I know your project is only meant to be supported on Linux, but I got it running on Mac OSX (with m2 architecture!) and I'd like to share how I did it.

Feel free to make of this information what you wish.

1. Install dependencies
   • brew install boost vamp-plugin-sdk
2. Download nnls-chroma plugin
   • git clone https://github.com/c4dm/nnls-chroma; cd nnls-chroma
3. Make the dependencies available in the source tree (edit version accordingly)
   • ln -s /opt/homebrew/Cellar/vamp-plugin-sdk/2.10.0/ vamp
   • ln -s /opt/homebrew/Cellar/boost/1.81.0_1 boost
4. Horribly copy files around to be where the Makefile expects them
   • cp vamp/lib/*a vamp/include/
5. Patch Makefile.osx as follows

diff --git a/Makefile.osx b/Makefile.osx
index d6fa701..46839a3 100644
--- a/Makefile.osx
+++ b/Makefile.osx
@@ -1,9 +1,9 @@

-VAMP_SDK_DIR = ../vamp-plugin-sdk
+VAMP_SDK_DIR = vamp/include

-BOOST_ROOT = ../boost_1_48_0
+BOOST_ROOT = boost/include

-ARCHFLAGS ?= -mmacosx-version-min=10.7 -arch x86_64
+ARCHFLAGS ?= -mmacosx-version-min=10.7
 OPTFLAGS  ?= -O3 -ffast-math
 PLUGIN_EXT = .dylib

6. Build and install
   • make -f Makefile.osx
   • mkdir -p ~/Library/Audio/Plug-Ins/Vamp
   • cp nnls-chroma.dylib ~/Library/Audio/Plug-Ins/Vamp

Hope this may save a bit of time and headache to the next person trying to run this :)

Thanks for your work!

My question here only applies to songs where both the tempo and the time signature are known, but that should be most of the songs out there.

Imagining a song in 4/4 with 120 qpm that changes chord every 4 bars, you would have a change every 8 seconds (quarter is 60s/120=0.5s, bar is 0.5s * 4 = 2s, chord length is 2s * 4 = 8s). So the ideal output would be for example:

0.0   8.0   E:min
8.0   16.0  A:maj
16.0  24.0  E:min
[.....]

In reality time offsets in the predictions are a bit wonky, that is probably because in real sound there is not really an exact time when a chord starts. I have also tested this on a .wav render of a midi.

If the tempo is low and bars are long then durations can be sort of quantised "with a wrench hit" by approximating to the closest bar, but when the tempo is high enough (100+?) the timing error becomes too big, making it impossible to pin exactly when in the score the chord is changed.

I don't know much about how your NN works, but perhaps this is because the wave is analysed "continuously"? could it be made to analyse segments that are aligned with bars instead? In the previous song, for example, could the prediction function be made to guess what chord is there from 0.0 to 2.0, then from 2.0 to 4.0, etc?

Thanks!

Anyone knows how could i resolve this error when i'm installing the package, aparently, the error occurs when it's going to install the vamp package as a dependency, in the setup.py files, it doesn't recognize the numpy package even if it's already installed.

It's amazing, I'm making an app to help myself learn guitar and the recognition is pretty accurate.

Keep the good work.

When I import autochord, I get an error that says, "Vamp plugin has not been installed correctly" Says to put C:\Users[user].conda\envs\ldm\lib\site-packages\autochord\res\nnls-chroma.so to vamp plugin folder, which doesn't exist, so I put it in the Vamp library folder in my conda env packages but that didn't work either. Any help?

Hi there, thanks for the codes! It seems like it can only recognize simplified chords though, is it also capable of identifying more complicated chord qualities like aug6, min7b5 etc?

Thanks!