Tutorial 2 - Extract changing notes.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "78e342ae",
   "metadata": {},
   "source": [
    "# Extract diffrent and changing notes from audio\n",
    ">In the last tutorial we learned hot recognize the wave in an audio.<br>\n",
    "But what to do when the notes are changing?<br>\n",
    "<br>\n",
    "**This tutorial is about that topic.**"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6dce73e5-7ec8-4feb-baad-048366b1ae46",
   "metadata": {},
   "outputs": [],
   "source": [
    "from AudioParamersExtraction import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "652585b2-32c7-42b5-bebc-5f7f4851dcec",
   "metadata": {},
   "outputs": [],
   "source": [
    "sound_array, sr = librosa.load('sounds/John Small.mp3')\n",
    "sound_array, _ = librosa.effects.trim(sound_array, top_db=10)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "92633058-d05e-46e4-bf76-ecf4f006aa7a",
   "metadata": {},
   "source": [
    ">#### When next cell executed, your speaker will start to play the audio.<br>In order to stop - execute the next cell after it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d7255289-135b-446e-a546-558368981cbf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(sound_array)\n",
    "\n",
    "# Play the original audio\n",
    "sd.play(sound_array, sr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "44ffc0e3-2e91-49bc-87af-0af07a5c7e40",
   "metadata": {},
   "outputs": [],
   "source": [
    "sd.stop()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "726c9fba",
   "metadata": {},
   "source": [
    "## The strategy - moving window\n",
    ">1. We will split the audio into short segments (window), and analyze each one separately.\n",
    ">2. Then, we'll extend all the predicted short segments into one audio array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e7e1134b-26a1-4ac6-a3a5-ebf9a369f212",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Length of a sample\n",
    "N = 1_000\n",
    "# Starting point of this example\n",
    "start = 0\n",
    "# The duration of this sample in seconds\n",
    "DURATION = N/sr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "ca414ea7-7b5a-41c5-b921-e32682aeeac6",
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\Users\\97250\\anaconda3\\lib\\site-packages\\scipy\\optimize\\minpack.py:828: OptimizeWarning: Covariance of the parameters could not be estimated\n",
      "  warnings.warn('Covariance of the parameters could not be estimated',\n"
     ]
    }
   ],
   "source": [
    "sounds = []\n",
    "freqs = []\n",
    "mags = []\n",
    "for i in range(0, sound_array.shape[0]-N, N):\n",
    "    window = sound_array[i:i+N]\n",
    "    sound_array_pred, predicted_freqs, predicted_mag = predict_wave_freqs_magnitudes(window, sr, DURATION, N)\n",
    "    \n",
    "    sounds.append(sound_array_pred)\n",
    "    freqs.append(predicted_freqs)\n",
    "    mags.append(predicted_mag)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "41022042-0fb6-4353-90c2-71d7b3334972",
   "metadata": {},
   "outputs": [],
   "source": [
    "sounds_append = np.array([])\n",
    "for s in sounds:\n",
    "    sounds_append = np.append(sounds_append, s)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "9d6e1403",
   "metadata": {},
   "source": [
    "Let's hear the result:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "aab362ac-af35-40e2-beee-3be9613d43e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "sd.play(sounds_append, sr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "abeb0dac-c8c6-48b3-a71d-cd73487293f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "sd.stop()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "51780955",
   "metadata": {},
   "source": [
    ">Well, it sounds ok... but needs improvement.<br>\n",
    "Let's check out the predicted notes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "37a4fc48-baa5-4793-b1eb-37b2ece39d1a",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "C8\tG5\tD7\tG5\tB6\tE6\tB6\tE5\tC8\tE6\tC8\tB6\tE6\tF5\tF6\tF5\tE6\tB4\tE6\tB4\tE5\tC8\tA6\tD6\tD5\tD6\tA6\tD5\tC5\tD5\tC8\tD5\tB4\tD5\tC5\tC8\tC7\tC5\tB4\tC8\tD5\tD6\tD5\tA4\tC8\tE6\tB6\tE5\tB6\tE6\tC8\tF5\tF6\tF5\tE5\tB6\tE5\tA4\tG5\tD7\tG5\tB7\tG5\tG7\tG5\tC8\tG5\tD#5\tG5\tB4\tD#5\tC8\tG5\tB7\tD7\tG5\tC8\tE6\tB7\tB6\tE5\tC8\tE6\tB6\tE6\tE5\tE6\tC8\tF5\tF6\tF5\tA7\tE5\tC8\tD5\tD6\tD5\tA#6\tD5\tB4\tC5\tC7\tB4\tC5\tB4\tG6\tF7\tC8\tE6\tB6\tE5\tG5\tE5\tC8\tG5\tB7\tD7\tG5\tC5\tB4\tC5\tG6\tC6\tA4\tD#5\tG4\tA4\tD#5\tG4\tD#5\tG4\tA3\tA4\tD#5\tD5\tC8\tD5\tA6\tD5\tD#5\tC7\tD5\tD6\tD5\tB4\tC8\tD5\tC5\tD5\tC8\tD5\tA6\tB4\tE7\tC8\tE6\tB6\tE5\tC8\tF5\tE5\tC5\tC8\tE6\tB6\tE5\tC8\tB6\tB7\tE5\tB6\tC8\tE6\tB6\tE5\tB6\tC8\tE6\tB6\tB7\tB6\tE5\tB6\tC8\tE6\tB6\tE5\tC8\tF5\tF6\tF5\tE5\tE6\tC8\tG5\tC8\tB7\tC8\tD7\tB7\tG5\tD#5\tG5\tD#5\tC8\tG5\tC8\tD7\tG5\tD7\tC8\tG5\tC8\tE6\tB6\tE5\tB6\tC8\tB6\tE6\tC8\tF5\tC7\tE5\tE6\tE5\tE6\tE5\tC5\tD5\tD#5\tD5\tD#5\tD5\tA4\tD#5\tA4\tC5\tC7\tB4\tC5\tB4\tC8\tE6\tC8\tB6\tE5\tC8\tG5\tE5\tC8\tG5\tB7\tG5\tC5\t"
     ]
    }
   ],
   "source": [
    "last_note = ''\n",
    "for f, m in zip(freqs, mags):\n",
    "    mx = m.max()\n",
    "    max_mag_index = np.where(m == mx)[0][0]\n",
    "    current_pred_note = get_note_of_freq(f[max_mag_index])\n",
    "    if last_note != current_pred_note:\n",
    "        print(current_pred_note, end='\\t')\n",
    "        last_note = current_pred_note"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "cadc8829",
   "metadata": {},
   "source": [
    ">It seems that the frequency with the largest amplitude is not the right way to choose the current note.<br>\n",
    "I'll go back to it later.\n",
    "<!---->\n",
    "#### The next improvment:<br>\n",
    "1. predict the frequencies of the segment from the center of the last window to the center of the next one.\n",
    "2. This will be like \"glue\".<br>\n",
    "<!---->\n",
    "This will help in solving the problem of breaking the wave continuity of different segments."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d7df71fc",
   "metadata": {},
   "outputs": [],
   "source": [
    "sounds = []\n",
    "for i in range(int(N/2), sound_array.shape[0]-int(N*1.5), N):\n",
    "    window = sound_array[i:i+N]\n",
    "    sound_array_pred, predicted_freqs, predicted_mag = predict_wave_freqs_magnitudes(window, sr, DURATION, N)\n",
    "    \n",
    "    sounds.append(sound_array_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e19ab8bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "sounds_append2 = np.array([])\n",
    "for s in sounds:\n",
    "    sounds_append2 = np.append(sounds_append2, s)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "3e56c927",
   "metadata": {},
   "source": [
    "##### Since `sounds_append2` is shorter then `sounds_append` in `N`,<br>we'll add zeros before and after it to their shapes will match"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "58d097ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "half_N_zeros = np.zeros((int(N/2),))\n",
    "sounds_append2 = np.append(half_N_zeros, sounds_append2)\n",
    "sounds_append2 = np.append(sounds_append2, half_N_zeros)\n",
    "\n",
    "sounds_append = sounds_append + sounds_append2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7f17bfa",
   "metadata": {},
   "source": [
    "Let's hear the result:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "12ddfa43",
   "metadata": {},
   "outputs": [],
   "source": [
    "sd.play(sounds_append, sr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "6b199d52",
   "metadata": {},
   "outputs": [],
   "source": [
    "sd.stop()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "9855d16b",
   "metadata": {},
   "source": [
    ">#### This sounds match better!<br>I will use this in order to improve the notes prediction."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "1321ef74",
   "metadata": {},
   "source": [
    "># NOTE: **This project is still under construction**!"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}