notebooks/test_model.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "62c5865f",
   "metadata": {},
   "source": [
    "<a href=\"https://colab.research.google.com/github/teticio/audio-diffusion/blob/master/notebooks/test_model.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6c7800a6",
   "metadata": {},
   "outputs": [],
   "source": [
    "try:\n",
    "    # are we running on Google Colab?\n",
    "    import google.colab\n",
    "    !git clone -q https://github.com/teticio/audio-diffusion.git\n",
    "    %cd audio-diffusion\n",
    "    !pip install -q -r requirements.txt\n",
    "except:\n",
    "    pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b447e2c4",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "sys.path.insert(0, os.path.dirname(os.path.abspath(\"\")))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c2fc0e7a",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import random\n",
    "import numpy as np\n",
    "from datasets import load_dataset\n",
    "from IPython.display import Audio\n",
    "from audiodiffusion.mel import Mel\n",
    "from audiodiffusion import AudioDiffusion"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7fd945bb",
   "metadata": {},
   "source": [
    "### Select model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "97f24046",
   "metadata": {},
   "outputs": [],
   "source": [
    "#@markdown teticio/audio-diffusion-256                     - trained on my Spotify \"liked\" playlist\n",
    "\n",
    "#@markdown teticio/audio-diffusion-breaks-256              - trained on samples used in music\n",
    "\n",
    "#@markdown teticio/audio-diffusion-instrumental-hiphop-256 - trained on instrumental hiphop\n",
    "\n",
    "model_id = \"teticio/audio-diffusion-256\"  #@param [\"teticio/audio-diffusion-256\", \"teticio/audio-diffusion-breaks-256\", \"audio-diffusion-instrumenal-hiphop-256\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "011fb5a1",
   "metadata": {},
   "source": [
    "### Run model inference to generate mel spectrogram, audios and loops"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a3d45c36",
   "metadata": {},
   "outputs": [],
   "source": [
    "audio_diffusion = AudioDiffusion(model_id=model_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b809fed5",
   "metadata": {},
   "outputs": [],
   "source": [
    "generator = torch.Generator()\n",
    "for _ in range(10):\n",
    "    print(f'Seed = {generator.seed()}')\n",
    "    image, (sample_rate, audio) = audio_diffusion.generate_spectrogram_and_audio(generator)\n",
    "    display(image)\n",
    "    display(Audio(audio, rate=sample_rate))\n",
    "    loop = AudioDiffusion.loop_it(audio, sample_rate)\n",
    "    if loop is not None:\n",
    "        display(Audio(loop, rate=sample_rate))\n",
    "    else:\n",
    "        print(\"Unable to determine loop points\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0bb03e33",
   "metadata": {},
   "source": [
    "### Generate variations of audios"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "80e5b5fa",
   "metadata": {},
   "source": [
    "Try playing around with `start_steps`. Values closer to zero will produce new samples, while values closer to `steps` will produce samples more faithful to the original. You can also try generatring variations of a `slice` of an `audio_file` instead of passing in a `raw_audio`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a7e637e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "seed = 16183389798189209330  #@param {type:\"long\"}\n",
    "image, (sample_rate,\n",
    "        audio) = audio_diffusion.generate_spectrogram_and_audio_from_audio(\n",
    "            generator=torch.Generator().manual_seed(seed))\n",
    "display(image)\n",
    "display(Audio(audio, rate=sample_rate))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a0fefe28",
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "start_steps = 500  #@param {type:\"slider\", min:0, max:1000, step:10}\n",
    "loop = AudioDiffusion.loop_it(audio, sample_rate, loops=1)\n",
    "for variation in range(12):\n",
    "    image2, (\n",
    "        sample_rate, audio2\n",
    "    ) = audio_diffusion.generate_spectrogram_and_audio_from_audio(\n",
    "        raw_audio=audio,\n",
    "        slice=0,\n",
    "        start_step=start_steps,\n",
    "        steps=1000)\n",
    "    display(image2)\n",
    "    display(Audio(audio2, rate=sample_rate))\n",
    "    loop = np.concatenate([loop, AudioDiffusion.loop_it(audio2, sample_rate, loops=1)])\n",
    "display(Audio(loop, rate=sample_rate))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ef54cef3",
   "metadata": {},
   "source": [
    "### Compare results with random sample from training set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f028a3c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "mel = Mel(x_res=256, y_res=256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "269ee816",
   "metadata": {},
   "outputs": [],
   "source": [
    "ds = load_dataset(model_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b9023846",
   "metadata": {},
   "outputs": [],
   "source": [
    "image = random.choice(ds['train'])['image']\n",
    "image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "492e2334",
   "metadata": {},
   "outputs": [],
   "source": [
    "audio = mel.image_to_audio(image)\n",
    "Audio(data=audio, rate=mel.get_sample_rate())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c59bcc0f",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "provenance": []
  },
  "gpuClass": "standard",
  "kernelspec": {
   "display_name": "huggingface",
   "language": "python",
   "name": "huggingface"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.4"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}