fastai2_audio_conv1d.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai2.vision.all import *\n",
    "#import librosa\n",
    "#import librosa.display\n",
    "import torchaudio"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "path = untar_data('https://storage.googleapis.com/ml-animal-sounds-datasets/macaques_24414Hz.zip')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class TensorAudio(TensorBase):\n",
    "          \n",
    "    @classmethod\n",
    "    def create(cls, o):\n",
    "        o, sr = torchaudio.load(o, normalization=True)\n",
    "        #o = cls(tensor(o).float())\n",
    "        o = cls(o)\n",
    "        o.sr = sr\n",
    "        o.mode = 'raw'\n",
    "        #print(str(sr))\n",
    "        return o\n",
    "    '''\n",
    "    def show(self, ctx=None):\n",
    "        if self.mode == 'raw':\n",
    "            print(self.shape)\n",
    "            librosa.display.waveplot(np.asarray(self.squeeze()), sr=self.sr)\n",
    "            #print(img.shape)\n",
    "    '''   \n",
    "\n",
    "class AudioFixLength(Transform):\n",
    "\n",
    "    def __init__(self, length=0.0):\n",
    "        self.length = length\n",
    "\n",
    "    def encodes(self, o: TensorAudio):\n",
    "        if self.length > 0.0:\n",
    "            n_samples = int(o.sr * self.length)\n",
    "            if n_samples < len(o.squeeze()):\n",
    "                o = torch.split(o, n_samples, dim=1)[0]\n",
    "            else:\n",
    "                n_pad = int(o.sr * self.length - len(o.squeeze()))\n",
    "                n_pre = (torch.rand(1) * n_pad).int()\n",
    "                n_post = n_pad - n_pre\n",
    "                o = F.pad(input=o, pad=(n_pre,n_post), mode='constant', value=0)\n",
    "        return o\n",
    "\n",
    "class AudioAddNoise(RandTransform):\n",
    "    \"Randomly add noise with probability `p`\"\n",
    "    def __init__(self, p=0.5): super().__init__(p=p)\n",
    "    def encodes(self, o: TensorAudio): \n",
    "        noise_amp = 0.001*torch.rand(1).cuda()*torch.max(o).cuda()\n",
    "        #print(o.is_cuda, noise_amp.is_cuda)\n",
    "        o = o + noise_amp * torch.empty(o.shape).normal_().cuda()\n",
    "        return o\n",
    "\n",
    "class AudioToTensor(Transform):\n",
    "\n",
    "    def encodes(self, o: TensorAudio):\n",
    "        o = tensor(o).float()\n",
    "        #print(type(o))\n",
    "        return o\n",
    "\n",
    "def AudioBlock(length=0.0):\n",
    "    return TransformBlock(type_tfms=TensorAudio.create, batch_tfms=[AudioAddNoise])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "dblocks = DataBlock(blocks = (AudioBlock,CategoryBlock),\n",
    "                 get_items=get_files, \n",
    "                 splitter=RandomSplitter(seed=42),\n",
    "                 get_y=parent_label,\n",
    "                 item_tfms=[AudioFixLength(0.75)],\n",
    "                 batch_tfms=[AudioToTensor]\n",
    "                 )\n",
    "\n",
    "dls=dblocks.dataloaders(path, bs=64)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# https://www.kaggle.com/readilen/resnet-for-mnist-with-pytorch\n",
    "\n",
    "def conv1x3(in_channels, out_channels, kernel_size=3, stride=1):\n",
    "    padding = kernel_size // 2\n",
    "    return nn.Conv1d(in_channels, out_channels, kernel_size=kernel_size,\n",
    "                    stride=stride, padding=padding, bias=False)\n",
    "\n",
    "# Residual block\n",
    "class ResidualBlock(nn.Module):\n",
    "    def __init__(self, in_channels, out_channels, kernel_size=3, stride=1, downsample=None):\n",
    "        super(ResidualBlock, self).__init__()\n",
    "        self.conv1 = conv1x3(in_channels, out_channels, kernel_size, stride)\n",
    "        self.bn1 = nn.BatchNorm1d(out_channels)\n",
    "        self.relu = nn.ReLU(inplace=True)\n",
    "        self.conv2 = conv1x3(out_channels, out_channels, kernel_size)\n",
    "        self.bn2 = nn.BatchNorm1d(out_channels)\n",
    "        self.downsample = downsample\n",
    "\n",
    "    def forward(self, x):\n",
    "        #print(\"in\", x.shape)\n",
    "        residual = x\n",
    "        out = self.conv1(x)\n",
    "        #print(\"conv1\", out.shape)\n",
    "        out = self.bn1(out)\n",
    "        #print(\"bn1\", out.shape)\n",
    "        out = self.relu(out)\n",
    "        out = self.conv2(out)\n",
    "        #print(\"conv2\", out.shape)\n",
    "        out = self.bn2(out)\n",
    "        #print(\"bn2\", out.shape)\n",
    "        if self.downsample:\n",
    "            #print(\"downsample\")\n",
    "            residual = self.downsample(residual)\n",
    "        #print(\"+\", out.shape, residual.shape)\n",
    "        out += residual\n",
    "        out = self.relu(out)\n",
    "        return out\n",
    "    \n",
    "# ResNet\n",
    "class ResNet(nn.Module):\n",
    "    def __init__(self, block, layers, num_classes=10, kernel_size=3, stride=2):\n",
    "        super(ResNet, self).__init__()\n",
    "        self.in_channels = 16\n",
    "        self.conv = conv1x3(1, 16)\n",
    "        self.bn = nn.BatchNorm1d(16)\n",
    "        self.relu = nn.ReLU(inplace=True)\n",
    "        self.layer1 = self.make_layer(block, 16, layers[0], kernel_size)\n",
    "        self.layer2 = self.make_layer(block, 32, layers[0], kernel_size, stride)\n",
    "        self.layer3 = self.make_layer(block, 64, layers[1], kernel_size, stride)\n",
    "        self.layer4 = self.make_layer(block, 128, layers[2], kernel_size, stride)\n",
    "        self.layer5 = self.make_layer(block, 256, layers[3], kernel_size, stride)\n",
    "        self.layer6 = self.make_layer(block, 512, layers[4], kernel_size, stride)\n",
    "        self.avg_pool = nn.AvgPool1d(8)\n",
    "        self.fc1 = nn.Linear(1024, 96)\n",
    "        self.fc2 = nn.Linear(96, num_classes)\n",
    "\n",
    "    def make_layer(self, block, out_channels, blocks, kernel_size=3, stride=1):\n",
    "        downsample = None\n",
    "        if (stride != 1) or (self.in_channels != out_channels):\n",
    "            downsample = nn.Sequential(\n",
    "                conv1x3(self.in_channels, out_channels, kernel_size=kernel_size, stride=stride),\n",
    "                nn.BatchNorm1d(out_channels))\n",
    "        layers = []\n",
    "        layers.append(block(self.in_channels, out_channels, kernel_size, stride, downsample))\n",
    "        self.in_channels = out_channels\n",
    "        for i in range(1, blocks):\n",
    "            layers.append(block(out_channels, out_channels))\n",
    "        return nn.Sequential(*layers)\n",
    "\n",
    "    def forward(self, x):\n",
    "        out = self.conv(x)\n",
    "        out = self.bn(out)\n",
    "        out = self.layer1(out)\n",
    "        out = self.layer2(out)\n",
    "        out = self.layer3(out)\n",
    "        out = self.layer4(out)\n",
    "        out = self.layer5(out)\n",
    "        out = self.layer6(out)\n",
    "        out = self.avg_pool(out)\n",
    "        out = out.view(out.size(0), -1)\n",
    "        out = self.fc1(out)\n",
    "        out = self.fc2(out)\n",
    "        return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "kernel_size = 15\n",
    "stride = 4\n",
    "\n",
    "net_args = {\n",
    "    \"block\": ResidualBlock,\n",
    "    \"layers\": [2, 2, 2, 2, 2, 2],\n",
    "    \"kernel_size\": kernel_size,\n",
    "    \"stride\": stride,\n",
    "    \"num_classes\": 8\n",
    "}\n",
    "model = ResNet(**net_args)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = Learner(dls, model, loss_func=F.cross_entropy, metrics=accuracy, cbs=ShowGraphCallback())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "SuggestedLRs(lr_min=0.0003019951749593019, lr_steep=0.0004786300996784121)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: left;\">\n",
       "      <th>epoch</th>\n",
       "      <th>train_loss</th>\n",
       "      <th>valid_loss</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>time</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <td>0</td>\n",
       "      <td>1.018006</td>\n",
       "      <td>0.778250</td>\n",
       "      <td>0.766644</td>\n",
       "      <td>00:15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>1</td>\n",
       "      <td>0.494289</td>\n",
       "      <td>0.340718</td>\n",
       "      <td>0.890185</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>2</td>\n",
       "      <td>0.322716</td>\n",
       "      <td>0.317243</td>\n",
       "      <td>0.880577</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>3</td>\n",
       "      <td>0.307700</td>\n",
       "      <td>0.402817</td>\n",
       "      <td>0.886754</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>4</td>\n",
       "      <td>0.286236</td>\n",
       "      <td>0.380419</td>\n",
       "      <td>0.885381</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>5</td>\n",
       "      <td>0.240478</td>\n",
       "      <td>0.388511</td>\n",
       "      <td>0.886067</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>6</td>\n",
       "      <td>0.218757</td>\n",
       "      <td>0.281911</td>\n",
       "      <td>0.906658</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>7</td>\n",
       "      <td>0.223835</td>\n",
       "      <td>0.549075</td>\n",
       "      <td>0.827728</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>8</td>\n",
       "      <td>0.194920</td>\n",
       "      <td>0.752088</td>\n",
       "      <td>0.771448</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>9</td>\n",
       "      <td>0.191769</td>\n",
       "      <td>0.226953</td>\n",
       "      <td>0.925189</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>10</td>\n",
       "      <td>0.160903</td>\n",
       "      <td>0.325445</td>\n",
       "      <td>0.892931</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>11</td>\n",
       "      <td>0.186478</td>\n",
       "      <td>0.260539</td>\n",
       "      <td>0.920384</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>12</td>\n",
       "      <td>0.148671</td>\n",
       "      <td>0.153295</td>\n",
       "      <td>0.948524</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>13</td>\n",
       "      <td>0.155075</td>\n",
       "      <td>0.333540</td>\n",
       "      <td>0.903912</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>14</td>\n",
       "      <td>0.144725</td>\n",
       "      <td>0.127229</td>\n",
       "      <td>0.962251</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>15</td>\n",
       "      <td>0.125516</td>\n",
       "      <td>0.297908</td>\n",
       "      <td>0.906658</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>16</td>\n",
       "      <td>0.106858</td>\n",
       "      <td>0.204216</td>\n",
       "      <td>0.938229</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>17</td>\n",
       "      <td>0.099309</td>\n",
       "      <td>0.416561</td>\n",
       "      <td>0.861359</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>18</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>0.084585</td>\n",
       "      <td>0.966369</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>19</td>\n",
       "      <td>0.089397</td>\n",
       "      <td>0.138157</td>\n",
       "      <td>0.953329</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>20</td>\n",
       "      <td>0.086616</td>\n",
       "      <td>0.085644</td>\n",
       "      <td>0.970487</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>21</td>\n",
       "      <td>0.080185</td>\n",
       "      <td>0.140378</td>\n",
       "      <td>0.947152</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>22</td>\n",
       "      <td>0.066159</td>\n",
       "      <td>0.081550</td>\n",
       "      <td>0.973233</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>23</td>\n",
       "      <td>0.064594</td>\n",
       "      <td>0.914539</td>\n",
       "      <td>0.814001</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>24</td>\n",
       "      <td>0.065992</td>\n",
       "      <td>0.085793</td>\n",
       "      <td>0.969115</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>25</td>\n",
       "      <td>0.051474</td>\n",
       "      <td>0.062975</td>\n",
       "      <td>0.976664</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>26</td>\n",
       "      <td>0.050991</td>\n",
       "      <td>0.069950</td>\n",
       "      <td>0.975978</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>27</td>\n",
       "      <td>0.041390</td>\n",
       "      <td>0.058331</td>\n",
       "      <td>0.981469</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>28</td>\n",
       "      <td>0.041667</td>\n",
       "      <td>0.071247</td>\n",
       "      <td>0.980096</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>29</td>\n",
       "      <td>0.033631</td>\n",
       "      <td>0.053096</td>\n",
       "      <td>0.982155</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>30</td>\n",
       "      <td>0.028000</td>\n",
       "      <td>0.061179</td>\n",
       "      <td>0.980096</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>31</td>\n",
       "      <td>0.026678</td>\n",
       "      <td>0.040041</td>\n",
       "      <td>0.986960</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>32</td>\n",
       "      <td>0.022722</td>\n",
       "      <td>0.042418</td>\n",
       "      <td>0.987646</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>33</td>\n",
       "      <td>0.019113</td>\n",
       "      <td>0.055198</td>\n",
       "      <td>0.984214</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>34</td>\n",
       "      <td>0.015735</td>\n",
       "      <td>0.040784</td>\n",
       "      <td>0.988332</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>35</td>\n",
       "      <td>0.011064</td>\n",
       "      <td>0.026037</td>\n",
       "      <td>0.991764</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>36</td>\n",
       "      <td>0.010054</td>\n",
       "      <td>0.040610</td>\n",
       "      <td>0.989019</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>37</td>\n",
       "      <td>0.010363</td>\n",
       "      <td>0.039125</td>\n",
       "      <td>0.989019</td>\n",
       "      <td>00:14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>38</td>\n",
       "      <td>0.008521</td>\n",
       "      <td>0.039742</td>\n",
       "      <td>0.989019</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <td>39</td>\n",
       "      <td>0.007808</td>\n",
       "      <td>0.042196</td>\n",
       "      <td>0.989705</td>\n",
       "      <td>00:13</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit_one_cycle(40, 4e-4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "fastai2",
   "language": "python",
   "name": "fastai2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}