search

wayonbvc / Violin-Fingering-Generation-Through-Audio-Symbolic-Fusion

2 stars 0 forks source link

Error Message #2

Open YtkMAEDA opened 1 month ago

YtkMAEDA commented 1 month ago

Following Dr. Vincent Cheung's suggestion, I have tried your VFG-ASF. However, I couldn't go through the codes as attached messages. Would you kindly teach me the required inputs and the relation between train.py and generateaudio feature.py? Furthermore, I can get only fluidsynth ver. 0.2 in Anaconda Navigator. Version 2.2.5 is essential? ytkmaeda@gmail.com VFG-ASF_ Error Message

wayonbvc commented 1 month ago

Following Dr. Vincent Cheung's suggestion, I have tried your VFG-ASF. However, I couldn't go through the codes as attached messages. Would you kindly teach me the required inputs and the relation between train.py and generateaudio feature.py? Furthermore, I can get only fluidsynth ver. 0.2 in Anaconda Navigator. Version 2.2.5 is essential? ytkmaeda@gmail.com VFG-ASF_ Error Message

generateaudio feature.py generates audio features by aligning annotated fingering data and audio recordings. (Such as the file vio1_bach1.npy and vio1_bach1_recording.wav in the EXAMPLE folder). The output audio features are going to be the input of train.py.

The version of fluidsynth would be fine as long as the code "fs.midi_to_audio" can be executed successfully (It takes me some effort to install fluidsynth)

To run train.py successfully, you have to download the audio features first. (The link is in the README.md in the Code folder). You don't have to run generateaudio feature.py to train unless you want to use your own data.