악보제작_및_코드데이터베이스_구축.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "악보제작 및 코드데이터베이스 구축.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyPDmd3it6MzTOn6dd8LVGBn",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jikerbug/jibaek_project_generator/blob/master/%EC%95%85%EB%B3%B4%EC%A0%9C%EC%9E%91_%EB%B0%8F_%EC%BD%94%EB%93%9C%EB%8D%B0%EC%9D%B4%ED%84%B0%EB%B2%A0%EC%9D%B4%EC%8A%A4_%EA%B5%AC%EC%B6%95.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QZNqrfiou7Qt",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# class & 악보추출\n",
        "\n",
        "import librosa\n",
        "# pip install librosa\n",
        "\n",
        "import vamp\n",
        "\"\"\"\n",
        "1. pip install vamp\n",
        "2. https://code.soundsoftware.ac.uk/projects/vamp-plugin-pack 에서 운영체제에 맞는 exe파일을 받고 Chordino and NNLS chroma 플러그인만 설치\n",
        "\"\"\"\n",
        "\n",
        "import glob\n",
        "import os.path\n",
        "import youtube_dl\n",
        "\n",
        "from PIL import Image, ImageDraw, ImageFont\n",
        "#pip install pillow\n",
        "\n",
        "\n",
        "class _Chord_Classification_Service:\n",
        "\n",
        "    _instance = None\n",
        "    \n",
        "    # 유튜브 url로부터 시간별 코드리스트, 비트리스트, 유튜브 추출 파일 이름을 반환하는 함수\n",
        "    def load_audio_data_from_url(self, url):\n",
        "        ydl_opts = {\n",
        "            'format': 'bestaudio',\n",
        "        }\n",
        "        with youtube_dl.YoutubeDL(ydl_opts) as ydl:\n",
        "            ydl.download([url])\n",
        "\n",
        "        # 확장자 변경\n",
        "        files = glob.glob(\"*.webm\")\n",
        "        audio_file_name = ''\n",
        "        for x in files:\n",
        "            if not os.path.isdir(x):\n",
        "                filename = os.path.splitext(x)\n",
        "                try:\n",
        "                    os.rename(x, filename[0] + '.wav')\n",
        "                    audio_file_name = filename[0]\n",
        "                except:\n",
        "                    pass\n",
        "\n",
        "        # load audio file\n",
        "        signal, sr = librosa.load(filename[0] + '.wav')\n",
        "        os.remove(audio_file_name + '.wav')\n",
        "\n",
        "        predicted_beat_list = vamp.collect(signal, sr, \"beatroot-vamp:beatroot\")\n",
        "\n",
        "        predicted_chord_list = vamp.collect(signal, sr, \"nnls-chroma:chordino\")\n",
        "\n",
        "        chord_list = []\n",
        "        for i in predicted_chord_list['list']:\n",
        "            i['timestamp'] = float(i['timestamp'])\n",
        "            chord_list.append(i)\n",
        "\n",
        "        beat_list = []\n",
        "        for i in predicted_beat_list['list']:\n",
        "            beat_list.append(float(i['timestamp']))\n",
        "\n",
        "        return chord_list, beat_list, audio_file_name.split('-')[0]\n",
        "    \n",
        "    # load_audio_data_from_url로부터 정보를 받아와 악보를 저장하는 함수\n",
        "    def make_music_sheet(self, url, title):\n",
        "        chord_list, beat_list, audio_file_name = self.load_audio_data_from_url(url)\n",
        "        print(chord_list)\n",
        "        print(beat_list)\n",
        "        print(len(beat_list))\n",
        "\n",
        "        # for문을 돌다가 마지막 element에 도달하기 전에 끊기지 않게 하기 위함\n",
        "        while((len(beat_list) + 6) % 16 != 0):\n",
        "            beat_list.append(beat_list[-1])\n",
        "        print(len(beat_list))\n",
        "        beat_list.append(beat_list[-1])\n",
        "\n",
        "        text = ''\n",
        "        if title == '':\n",
        "            title = audio_file_name\n",
        "\n",
        "\n",
        "        text += title\n",
        "        text += '\\n\\n\\n'\n",
        "\n",
        "\n",
        "        total_chord_cnt = 0\n",
        "        beat_flag = 0\n",
        "        for chord in chord_list:\n",
        "            if chord['timestamp'] >= 0 and chord['timestamp'] < beat_list[0]:\n",
        "                if chord['label'] == 'N':\n",
        "                    continue\n",
        "                text += chord['label'] + '  '\n",
        "                total_chord_cnt += 1\n",
        "\n",
        "        init_flag = True\n",
        "        for beat_id in range(len(beat_list)):\n",
        "            if init_flag == False:\n",
        "                if beat_flag != 16:\n",
        "                    beat_flag += 1\n",
        "                    continue\n",
        "            else:\n",
        "                if beat_flag != 10:\n",
        "                    beat_flag += 1\n",
        "                    continue\n",
        "                init_flag = False\n",
        "\n",
        "            for chord in chord_list:\n",
        "                if chord['timestamp'] >= beat_list[beat_id - beat_flag] and chord['timestamp'] < beat_list[beat_id]:\n",
        "                    text += chord['label'] + '  '\n",
        "                    total_chord_cnt += 1\n",
        "            text += '\\n\\n'\n",
        "            beat_flag = 0\n",
        "\n",
        "        for chord in chord_list:\n",
        "            if chord['timestamp'] >= 0 and chord['timestamp'] < beat_list[0]:\n",
        "                if chord['label'] == 'N':\n",
        "                    continue\n",
        "                text += chord['label'] + '  '\n",
        "                total_chord_cnt += 1\n",
        "\n",
        "        sheet_height = len(text.split('\\n\\n') )\n",
        "        sheet_width = int(250 * total_chord_cnt/sheet_height)\n",
        "        #./ batang.ttc\n",
        "        print(sheet_height * 90 / sheet_width)\n",
        "        if  sheet_height * 90 / sheet_width > 8:\n",
        "            sheet_width += 250\n",
        "        font = ImageFont.truetype('C:/Windows/Fonts/batang.ttc', 45)\n",
        "        img = Image.new(mode='RGB', size=(sheet_width, sheet_height * 90), color='#FFF')\n",
        "        draw = ImageDraw.Draw(img)\n",
        "        draw.text((50, 50), text, font=font, fill='#000')\n",
        "        img.save(title + '.png', format='PNG')\n",
        "\n",
        "    def make_chord_database(self, filename):\n",
        "\n",
        "        # 확장자 변경\n",
        "        files = glob.glob(filename + '.webm')\n",
        "        audio_file_name = ''\n",
        "        for x in files:\n",
        "            if not os.path.isdir(x):\n",
        "                file_name = os.path.splitext(x)\n",
        "                try:\n",
        "                    os.rename(x, file_name[0] + '.wav')\n",
        "\n",
        "                except:\n",
        "                    pass\n",
        "\n",
        "        print(filename)\n",
        "\n",
        "        signal, sr = librosa.load(filename + '.wav')\n",
        "        os.remove(filename + '.wav')\n",
        "\n",
        "        predicted_beat_list = vamp.collect(signal, sr, \"beatroot-vamp:beatroot\")\n",
        "\n",
        "        predicted_chord_list = vamp.collect(signal, sr, \"nnls-chroma:chordino\")\n",
        "\n",
        "        chord_list = []\n",
        "        for i in predicted_chord_list['list']:\n",
        "            i['timestamp'] = float(i['timestamp'])\n",
        "            chord_list.append(i)\n",
        "\n",
        "        beat_list = []\n",
        "        for i in predicted_beat_list['list']:\n",
        "            beat_list.append(float(i['timestamp']))\n",
        "\n",
        "        print(chord_list)\n",
        "        print(beat_list)\n",
        "\n",
        "        return chord_list, beat_list\n",
        "\n",
        "\n",
        "\n",
        "def  Chord_Classification_Service():\n",
        "    \n",
        "    #singleton 구조인데 그냥 유튭강의 따라한 흔적. 속도 아주 조금더 빠르지 않을까 생각. 없애도 무방\n",
        "\n",
        "    #ensure that we only have 1 instance of CCS\n",
        "    if  _Chord_Classification_Service._instance is None:\n",
        "        _Chord_Classification_Service._instance = _Chord_Classification_Service()\n",
        "    return  _Chord_Classification_Service._instance\n",
        "\n",
        "if __name__ == \"__main__\":\n",
        "\n",
        "    ccs = Chord_Classification_Service()\n",
        "\n",
        "    let_it_be_url = 'https://www.youtube.com/watch?v=QDYfEBY9NM4'\n",
        "    The_visitor_url = 'https://www.youtube.com/watch?v=y5YmTj8KDWM'\n",
        "    brown_url = 'https://www.youtube.com/watch?v=-5duYTCCtqI'\n",
        "\n",
        "    title = ''\n",
        "    ccs.make_music_sheet(brown_url, title)\n",
        "\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sRSmy0-MvDQc",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# main문 & 코드데이터베이스 구축\n",
        "\n",
        "import youtube_dl\n",
        "from chord_classification_service import Chord_Classification_Service\n",
        "\n",
        "\n",
        "if __name__ == \"__main__\":\n",
        "\n",
        "    css = Chord_Classification_Service()\n",
        "\n",
        "    playlist_url = 'https://www.youtube.com/playlist?list=OLAK5uy_kfRVORKFdFjpkjalascOpqWbNawtinMmI'\n",
        "\n",
        "    ydl = youtube_dl.YoutubeDL({\n",
        "        'format': 'bestaudio',\n",
        "        'dump_single_json': True,\n",
        "        'extract_flat' : True\n",
        "    })\n",
        "\n",
        "    with ydl:\n",
        "        info = ydl.extract_info(playlist_url, False)\n",
        "    print(214)\n",
        "    print(info)\n",
        "\n",
        "    ydl = youtube_dl.YoutubeDL({\n",
        "        'format': 'bestaudio'\n",
        "    })\n",
        "\n",
        "    with ydl:\n",
        "\n",
        "        result = ydl.download([\n",
        "            'https://www.youtube.com/playlist?list=OLAK5uy_kfRVORKFdFjpkjalascOpqWbNawtinMmI'])\n",
        "\n",
        "\n",
        "    for music in info['entries']:\n",
        "        filename = music['title'] + '-' + music['id']\n",
        "        css.make_chord_database(filename)\n",
        "\n",
        "\n"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}