Prelim. Version 0.7. Encoder is done. Working on Decoder...

DOREMI.ipynb

@@ -45,27 +45,19 @@
  "cellView": "form"
  },
  "source": [
- "#@title Install dependencies, import modules, check GPU\n",
+ "#@title Install dependencies, import modules, set GPU, create MIDI Dataset dir\n",
  "!pip install miditoolkit\n",
  "import miditoolkit\n",
  "import numpy as np\n",
  "import os\n",
+ "from os import listdir\n",
+ "from os.path import isfile, join\n",
  "os.environ['CUDA_VISIBLE_DEVICES'] = '0'\n",
- "import copy"
- ],
- "execution_count": null,
- "outputs": []
- },
- {
- "cell_type": "code",
- "metadata": {
- "id": "m054X5rd2ARx",
- "cellView": "form"
- },
- "source": [
- "#@title (Best Choice/Works best stand-alone) Alex Piano Only Drafts Original 1500 MIDIs \n",
- "!wget 'https://github.com/asigalov61/AlexMIDIDataSet/raw/master/AlexMIDIDataSet-CC-BY-NC-SA-All-Drafts-Piano-Only.zip'\n",
- "!unzip -j 'AlexMIDIDataSet-CC-BY-NC-SA-All-Drafts-Piano-Only.zip'"
+ "import copy\n",
+ "import glob\n",
+ "import tqdm.auto\n",
+ "if not os.path.exists('/content/midis'):\n",
+ " os.makedirs('/content/midis')"
  ],
  "execution_count": null,
  "outputs": []
@@ -651,15 +643,11 @@
  },
  "source": [
  "#@title Calculate Number of Indexable MIDI files :)\n",
- "import os\n",
- "from os import listdir\n",
- "from os.path import isfile, join\n",
  "\n",
  "# Modify these paths to your own MIDI Dataset\n",
  "data_dirs = ['/content/midis']\n",
  "\n",
- "if not os.path.exists('/content/midis'):\n",
- " os.makedirs('/content/midis')\n",
+ "\n",
  "\n",
  "data_files = []\n",
  "for data_dir in data_dirs:\n",
@@ -680,28 +668,45 @@
  },
  "source": [
  "#@title Convert MIDIs to REMI and then TXT\n",
- "from tqdm import tqdm_notebook as tqdm\n",
- "import os\n",
- "import numpy as np\n",
+ "\n",
  "\n",
  "dataset_dir = 'dataset'\n",
- "#os.chdir('remi')\n",
+ "\n",
  "if not os.path.exists(dataset_dir):\n",
  " os.makedirs(dataset_dir)\n",
  "\n",
- "for i in tqdm(range(len(data_files))):\n",
+ "for i in tqdm.auto.tqdm(range(len(data_files))):\n",
  " print(data_files[i])\n",
  " eventlist = get_events_from_midis(data_files[i])\n",
  " #eventlist = events\n",
  " #print(eventlist)\n",
  " \n",
- " save_file = dataset_dir + '/' + str(i)\n",
+ " save_file = dataset_dir + '/' + str(i) + '.txt'\n",
  " data = {'eventlist': eventlist}\n",
  " #np.savez(save_file, **data, allow_pickle=False)\n",
- " #np.savetxt(save_file, eventlist, delimiter=',')\n",
- "with open('output.txt', 'w') as midi_file:\n",
- " midi_file.write(str(eventlist))\n",
- " midi_file.close()\n",
+ " np.savetxt(save_file, eventlist, fmt='%s', delimiter=',')\n",
+ "\n",
+ "# Creating a list of TXT filenames \n",
+ "\n",
+ "os.chdir('dataset')\n",
+ "\n",
+ "# Open file3 in write mode \n",
+ "if os.path.exists(\"Dataset.txt\"):\n",
+ " os.remove(\"Dataset.txt\")\n",
+ " print('Removing old Dataset...')\n",
+ "else:\n",
+ " print(\"Creating new Dataset file...\")\n",
+ "read_files = glob.glob(\"*.txt\")\n",
+ "with open('Dataset.txt', 'w') as outfile: \n",
+ " \n",
+ " # Iterate through list \n",
+ " for names in read_files: \n",
+ " \n",
+ " # Open each file in read mode \n",
+ " with open(names) as infile: \n",
+ " outfile.write(infile.read()) \n",
+ "\n",
+ "os.chdir('/content/')\n",
  "print('Done!')"
  ],
  "execution_count": null,