monophonic_melodies.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-12-05T12:27:34.209906Z",
     "start_time": "2020-12-05T12:27:28.280263Z"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import math\n",
    "import time\n",
    "import magenta.music as mm\n",
    "from magenta.models.melody_rnn import melody_rnn_sequence_generator\n",
    "from magenta.models.shared import sequence_generator_bundle\n",
    "from magenta.music import DEFAULT_QUARTERS_PER_MINUTE\n",
    "from magenta.music.protobuf.generator_pb2 import GeneratorOptions\n",
    "from magenta.music.protobuf.music_pb2 import NoteSequence\n",
    "from visual_midi import Plotter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-12-05T12:27:34.224867Z",
     "start_time": "2020-12-05T12:27:34.211900Z"
    }
   },
   "outputs": [],
   "source": [
    "def generate(bundle_name, seq_generator, gen_id, primer, qpm = DEFAULT_QUARTERS_PER_MINUTE, \n",
    "             total_steps = 64, temp = 1.0, beam_size = 1, branch_factor = 1, steps_per_iteration = 1):\n",
    "    \n",
    "    mm.notebook_utils.download_bundle(bundle_name, \"bundles\")\n",
    "    bundle = sequence_generator_bundle.read_bundle_file(os.path.join(\"bundles\", bundle_name))\n",
    "    \n",
    "    generator_map = seq_generator.get_generator_map()\n",
    "    generator = generator_map[gen_id](checkpoint = None, bundle = bundle)\n",
    "    generator.initialize()\n",
    "    \n",
    "    if primer:\n",
    "        primer_seq = mm.midi_io.midi_file_to_note_sequence(primer)\n",
    "    else:\n",
    "        primer_seq = NoteSequence()\n",
    "        \n",
    "    if primer_seq.tempos:\n",
    "        if len(primer_seq.tempos) > 1:\n",
    "            raise Exception('Monophonic Melody Generation. No support for multiple tempos')\n",
    "        qpm = primer_seq.tempos[0].qpm\n",
    "        \n",
    "    secs_per_step = 60.0 / (qpm * getattr(generator, 'steps_per_quarter', 4))\n",
    "    \n",
    "    primer_seq_steps = math.ceil(primer_seq.total_time / secs_per_step)\n",
    "    primer_seq_time = primer_seq_steps * secs_per_step\n",
    "    primer_end_ad = (0.00001 if primer_seq_time > 0 else 0)\n",
    "    primer_start_time = 0\n",
    "    primer_end_time = primer_start_time + primer_seq_time - primer_end_ad\n",
    "    \n",
    "    gen_steps = total_steps - primer_seq_steps\n",
    "    \n",
    "    if gen_steps <= 0:\n",
    "        raise Exception('Total steps too small : {}. Need at least 1 bar bigger than primer : {}'.format(\n",
    "        total_steps, primer_seq_steps))\n",
    "    \n",
    "    gen_time = gen_steps * secs_per_step\n",
    "    gen_start_time = primer_end_time\n",
    "    gen_end_time = gen_start_time + gen_time + primer_end_ad\n",
    "    \n",
    "    print('Primer start : {} \\t Primer end : {}'.format(primer_start_time, primer_end_time))\n",
    "    print('Generation start : {} \\t Generation end : {}'.format(gen_start_time, gen_end_time))\n",
    "    \n",
    "    generator_options = GeneratorOptions()\n",
    "    generator_options.args['temperature'].float_value = temp\n",
    "    generator_options.args['beam_size'].int_value = beam_size\n",
    "    generator_options.args['branch_factor'].int_value = branch_factor\n",
    "    generator_options.args['steps_per_iteration'].int_value = steps_per_iteration\n",
    "    generator_options.generate_sections.add(start_time = gen_start_time, end_time = gen_end_time)\n",
    "    \n",
    "    seq = generator.generate(primer_seq, generator_options)\n",
    "    \n",
    "    # Write midi file\n",
    "    date_time = time.strftime('%Y-%m-%d_%H%M%S')\n",
    "    gen_name = str(generator.__class__).split('.')[2]\n",
    "    midi = '%s_%s_%s.mid' % (gen_name, gen_id, date_time)\n",
    "    midi_path = os.path.join('output', midi)\n",
    "    mm.midi_io.note_sequence_to_midi_file(seq, midi_path)\n",
    "    print('Generated midi : {}'.format(os.path.abspath(midi_path)))\n",
    "    \n",
    "    # Generate plot\n",
    "    date_time = time.strftime('%Y-%m-%d_%H%M%S')\n",
    "    gen_name = str(generator.__class__).split('.')[2]\n",
    "    plot = '%s_%s_%s.html' % (gen_name, gen_id, date_time)\n",
    "    plot_path = os.path.join('output', plot)\n",
    "    \n",
    "    pretty_midi = mm.midi_io.note_sequence_to_pretty_midi(seq)\n",
    "    plotter = Plotter()\n",
    "    plotter.show(pretty_midi, plot_path)\n",
    "    print('Generated plot : {}'.format(os.path.abspath(plot_path)))\n",
    "    \n",
    "    return seq"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-12-05T10:45:55.422536Z",
     "start_time": "2020-12-05T10:45:54.843149Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:The saved meta_graph is possibly from an older release:\n",
      "'model_variables' collection should be of type 'byte_list', but instead is of type 'node_list'.\n",
      "INFO:tensorflow:Restoring parameters from C:\\Users\\siddh\\AppData\\Local\\Temp\\tmpml5dilpn\\model.ckpt\n",
      "Primer start : 0 \t Primer end : 1.6463307499999995\n",
      "Generation start : 1.6463307499999995 \t Generation end : 11.707311999999995\n",
      "INFO:tensorflow:Beam search yields sequence with log-likelihood: -114.740669 \n",
      "Generated midi : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_basic_rnn_2020-12-05_054555.mid\n",
      "Generated plot : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_basic_rnn_2020-12-05_054555.html\n"
     ]
    }
   ],
   "source": [
    "sequence = generate('basic_rnn.mag', melody_rnn_sequence_generator, \n",
    "                    'basic_rnn', primer = 'primers/Fur_Elisa_Beethoveen_Monophonic.mid',\n",
    "                    total_steps = 64, temp = 0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-12-05T10:48:19.095536Z",
     "start_time": "2020-12-05T10:48:17.213256Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:The saved meta_graph is possibly from an older release:\n",
      "'model_variables' collection should be of type 'byte_list', but instead is of type 'node_list'.\n",
      "INFO:tensorflow:Restoring parameters from C:\\Users\\siddh\\AppData\\Local\\Temp\\tmpjcn82waa\\model.ckpt\n",
      "Primer start : 0 \t Primer end : 1.6463307499999995\n",
      "Generation start : 1.6463307499999995 \t Generation end : 11.707311999999995\n",
      "INFO:tensorflow:Beam search yields sequence with log-likelihood: -89.669540 \n",
      "Generated midi : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_lookback_rnn_2020-12-05_054818.mid\n",
      "Generated plot : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_lookback_rnn_2020-12-05_054818.html\n"
     ]
    }
   ],
   "source": [
    "sequence = generate('lookback_rnn.mag', melody_rnn_sequence_generator,\n",
    "                    'lookback_rnn', primer = 'primers/Fur_Elisa_Beethoveen_Monophonic.mid',\n",
    "                    total_steps = 64, temp = 1.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2020-12-05T10:52:45.143769Z",
     "start_time": "2020-12-05T10:52:44.340022Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:The saved meta_graph is possibly from an older release:\n",
      "'model_variables' collection should be of type 'byte_list', but instead is of type 'node_list'.\n",
      "INFO:tensorflow:Restoring parameters from C:\\Users\\siddh\\AppData\\Local\\Temp\\tmp78v1yg4m\\model.ckpt\n",
      "Primer start : 0 \t Primer end : 1.6463307499999995\n",
      "Generation start : 1.6463307499999995 \t Generation end : 23.414623999999993\n",
      "INFO:tensorflow:Beam search yields sequence with log-likelihood: -224.228256 \n",
      "Generated midi : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_attention_rnn_2020-12-05_055245.mid\n",
      "Generated plot : C:\\Users\\siddh\\Documents\\GitHub\\magenta_project\\output\\melody_rnn_attention_rnn_2020-12-05_055245.html\n"
     ]
    }
   ],
   "source": [
    "sequence = generate('attention_rnn.mag', melody_rnn_sequence_generator,\n",
    "                    'attention_rnn', primer = 'primers/Fur_Elisa_Beethoveen_Monophonic.mid',\n",
    "                    total_steps = 128, temp = 1.1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}