MidiMus-1K / README.md
Kotodobr's picture
Update README.md
feb4363 verified
|
Raw History Blame Contribute Delete
1.64 kB
metadata
license: apache-2.0
task_categories:
  - text-to-audio
language:
  - en
tags:
  - text
  - audio
pretty_name: MidiMUS 1K
size_categories:
  - 1K<n<10K

This dataset provided for Midi generation models using. Dataset have 1000 .txt files with descriptions and 1000 .mid files with MIDI. ##Instructions for install: ###1. Install : Install the dataset to folder - in guide this be "music_dataset" In terminal if you on GNU Linux or on CMD if you on Windows write this:

git clone https://huggingface.co/datasets/RedCatTeam/MidiMus-1K

###2.Preparing : Rename the "MidiMus-1K" to "music_dataset" (or yourself choosed name) and delete .git file and gitgnore file - after it - dataset downloading done! ###.3 Using : Example loader with Keras's Tokenizer training:

from mido import MidiFile
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
import numpy as np

tokenizer=Tokenizer()

def loadFile(fl):
    mid=MidiFile('./music_dataset/'+fl)
    notes=[]
    for track in mid.tracks[1]:
        if track.dict().get('note')!=None:
            notes.append(track.dict().get('note')/88)
    return notes

def loadDataset():
    x=[]
    texts=[]
    y=[]

    for i in range(1000):
        mid='piece_'+str(i)+'.mid'
        txt='piece_'+str(i)+'.txt'
        
        notes=loadFile(mid)
        with open('./music_dataset/'+txt,'r') as f:
            desc=f.read()
        texts.append(desc)
        y.append(notes)
    tokenizer.fit_on_texts(texts)
    x=pad_sequences(tokenizer.texts_to_sequences(texts),maxlen=6)
    return np.array(x),np.array(y)

# ...