Update README & requirements

README.md

@@ -28,6 +28,7 @@ An introduction to neural networks and deep learning:
 ### Machine Learning & Deep Learning methods
 
 [Using CNNs on spectrogram images](https://www.ofai.at/~jan.schlueter/pubs/2014_icassp.pdf) (Schlüter, Böck, 2014) :
+ - [MADMOM implementation](https://madmom.readthedocs.io/en/v0.16/modules/features/onsets.html)
  - Python implementation for Taiko rythm games : https://github.com/seiichiinoue/odcnn
 
 ### Other methods
@@ -45,7 +46,9 @@ If we ever want to use an AI to identify syllables without a reference lyrics fi
 
 - MKVToolnix (at least the CLI utils)
 - Python >= 3.8
-- PyTorch : follow the instructions [here](https://pytorch.org/get-started/locally/)
+
+Optional :
+- PyTorch for custom model training : follow the instructions [here](https://pytorch.org/get-started/locally/)
 
 All other python modules can be installed directly through pip, see further.
 
@@ -62,35 +65,12 @@ $ pip install -r requirements.txt
 $ deactivate              
 ```
 
-Having a CUDA-capable GPU is optional, but can greatly reduce processing time.
+Having a CUDA-capable GPU is optional, but can greatly reduce processing time in some situations.
 
 
 # Use
 
-
-## Training
-
-To extract vocals and ASS from MKV video files:
-```bash
-$ ./process_train_data video_folder train_folder
-```
-
-To prepare the training data for the model :
-```bash
-$ python cnn_prepare_data.py train train_folder 
-```
-
-Prepared data will be stored in `./data/pickles/train_data.pickle`
-
-To train the model on the prepared data :
-```bash
-$ python cnn_train.py
-```
-
-The model will be written to `./models/model.pth`
-
-
-## Infer
+## Inference
 
 To execute AutoKara on a MKV video file :
 ```bash

requirements.txt

@@ -5,4 +5,8 @@ soundfile
 sklearn
 matplotlib
 numpy
-tqdm
\ No newline at end of file
+tqdm
+scipy
+cython
+mido
+git+https://github.com/CPJKU/madmom.git
\ No newline at end of file