audiogen_demo.ipynb

audiogen_demo.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "private_outputs": true,
      "provenance": [],
      "gpuType": "T4",
      "authorship_tag": "ABX9TyNWlmoOQ5qz7A+eZM7LgufJ",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/aidiary/40cb12a1345159b363aee3ddc21e31ef/audiogen_demo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# AudioGen\n",
        "\n",
        "https://github.com/facebookresearch/audiocraft/blob/main/demos/audiogen_demo.ipynb"
      ],
      "metadata": {
        "id": "5WFJ-8U7KSIn"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "%pip install audiocraft"
      ],
      "metadata": {
        "id": "1S471mOQKRJZ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 訓練済みモデルをロード"
      ],
      "metadata": {
        "id": "4hW_grOKLQHV"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from audiocraft.models import AudioGen\n",
        "\n",
        "# https://huggingface.co/facebook/audiogen-medium\n",
        "model = AudioGen.get_pretrained(\"facebook/audiogen-medium\")"
      ],
      "metadata": {
        "id": "-91JOQxtKVgX"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# 生成パラメータの設定\n",
        "# use_sampling（bool、オプション）：Trueの場合はサンプリングを使用し、それ以外の場合はargmaxデコーディングを行います。デフォルトはTrueです。\n",
        "# top_k（int、オプション）：サンプリングに使用されるtop_k。デフォルトは250です。\n",
        "# top_p（float、オプション）：サンプリングに使用されるtop_p、0に設定されている場合はtop_kが使用されます。デフォルトは0.0です。\n",
        "# temperature（float、オプション）：softmax温度パラメーター。デフォルトは1.0です。\n",
        "# duration（float、オプション）：生成された波形の持続時間。デフォルトは10.0です。\n",
        "# cfg_coef（float、オプション）：クラシファイアフリーガイダンスに使用される係数。デフォルトは3.0です。\n",
        "model.set_generation_params(\n",
        "    use_sampling=True,\n",
        "    top_k=250,\n",
        "    duration=5\n",
        ")"
      ],
      "metadata": {
        "id": "1cXM-qRdK34b"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 与えた音声の続きを生成する"
      ],
      "metadata": {
        "id": "p_R2y_TdLo7Z"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import math\n",
        "import torchaudio\n",
        "import torch\n",
        "from audiocraft.utils.notebook import display_audio\n",
        "\n",
        "\n",
        "def get_bip_bip(bip_duration=0.125, frequency=440, duration=0.5, sample_rate=16000, device=\"cuda\"):\n",
        "    \"\"\"指定された周波数でビープ音の系列を生成します。\"\"\"\n",
        "    t = torch.arange(int(duration * sample_rate), device=\"cuda\", dtype=torch.float) / sample_rate\n",
        "    wav = torch.cos(2 * math.pi * frequency * t)[None]\n",
        "    tp = (t % (2 * bip_duration)) / (2 * bip_duration)\n",
        "    envelope = (tp >= 0.5).float()\n",
        "    return wav * envelope"
      ],
      "metadata": {
        "id": "wMZsYGVRLNEA"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "audio_samples = get_bip_bip(0.125)\n",
        "audio_samples.shape"
      ],
      "metadata": {
        "id": "xi2npWVjMmJw"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(audio_samples.squeeze().cpu())\n",
        "display_audio(audio_samples, 16000)"
      ],
      "metadata": {
        "id": "mD0Y9SizMsyQ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# コピーされるので2つの同じサンプル与えたことになる\n",
        "audio_samples.expand(2, -1, -1).shape"
      ],
      "metadata": {
        "id": "z43SB4obNLzy"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# プープーの続きの音を2つ生成する\n",
        "# プロンプトは2つ与える\n",
        "# プロンプトは与えても与えなくてもよい（Unconditional）\n",
        "res = model.generate_continuation(\n",
        "    audio_samples.expand(2, -1, -1),\n",
        "    16000,\n",
        "    # [\"Whistling with wind blowing\", \"Typing on a typewriter\"],\n",
        "    progress=True)"
      ],
      "metadata": {
        "id": "z8XTQ6FtMx2M"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "res.shape"
      ],
      "metadata": {
        "id": "7u36fehRNyTZ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "display_audio(res, 16000)"
      ],
      "metadata": {
        "id": "gvuOgKnJNGhP"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## テキストで条件付けた生成"
      ],
      "metadata": {
        "id": "aWc_4AvVPDmm"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from audiocraft.utils.notebook import display_audio\n",
        "\n",
        "output = model.generate(\n",
        "    descriptions=[\n",
        "        \"Subway train blowing its horn\",\n",
        "        \"A cat meowing\",\n",
        "    ],\n",
        "    progress=True\n",
        ")\n",
        "\n",
        "display_audio(output, sample_rate=16000)"
      ],
      "metadata": {
        "id": "EeIemxfoNvev"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}