volume_analysis.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "authorship_tag": "ABX9TyMdpPkPBq6Y9Evw/pixjQz5",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/kth0522/AI_news/blob/main/volume_analysis.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "cCI8vdt4WLQc"
      },
      "outputs": [],
      "source": [
        "import os\n",
        "import torchaudio\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import torch\n",
        "import re\n",
        "from tqdm import tqdm"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def calculate_dB_torchaudio(audio_tensor):\n",
        "    \"\"\"Convert audio tensor amplitude to decibels in torchaudio.\"\"\"\n",
        "    magnitude = audio_tensor.abs()\n",
        "    magnitude[magnitude == 0] = 1e-6  # Avoid log(0)\n",
        "    return 20 * torch.log10(magnitude)\n",
        "\n",
        "\n",
        "def extract_number_from_filename(filename):\n",
        "    \"\"\"Extract the number from the filename.\"\"\"\n",
        "    match = re.search(r'\\d+', filename)\n",
        "    return int(match.group()) if match else None\n",
        "\n",
        "\n",
        "def process_wav_files_torchaudio(folder_path, output_csv):\n",
        "    results_data = []\n",
        "\n",
        "    # 파일 목록을 미리 정렬\n",
        "    filenames = sorted(os.listdir(folder_path), key=lambda x: extract_number_from_filename(x))\n",
        "\n",
        "    for filename in tqdm(filenames, desc=\"Processing files\"):\n",
        "        if filename.lower().endswith(\".wav\"):\n",
        "            file_path = os.path.join(folder_path, filename)\n",
        "\n",
        "            try:\n",
        "                audio_tensor, sample_rate = torchaudio.load(file_path)\n",
        "                audio_tensor = audio_tensor[0]  # Assuming mono audio\n",
        "                audio_dB = calculate_dB_torchaudio(audio_tensor)\n",
        "                mean_dB = torch.mean(audio_dB).item()\n",
        "                variance_dB = torch.var(audio_dB).item()\n",
        "                median_dB = torch.median(audio_dB).item()\n",
        "                max_dB = torch.max(audio_dB).item()\n",
        "\n",
        "                file_number = extract_number_from_filename(filename)\n",
        "\n",
        "                results_data.append({\n",
        "                    'Filename': file_number,\n",
        "                    'Mean_dB': mean_dB,\n",
        "                    'Variance_dB': variance_dB,\n",
        "                    'Median_dB': median_dB,\n",
        "                    'Max_dB': max_dB\n",
        "                })\n",
        "            except Exception as e:\n",
        "                print(f\"Error processing {filename}: {e}\")\n",
        "\n",
        "    results = pd.DataFrame(results_data)\n",
        "\n",
        "    # 결과를 'Filename'을 기준으로 오름차순으로 정렬\n",
        "    results.sort_values(by='Filename', inplace=True)\n",
        "\n",
        "    results.to_csv(output_csv, index=False)"
      ],
      "metadata": {
        "id": "jTD-0AmuWTZQ"
      },
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "BASE_DIR = '/datasets/skku_anchor/'\n",
        "DATASET = 'mbn_morningdailyeco'\n",
        "\n",
        "# folder_path에 vocal 파일 들어있는 폴더 경로 넣고 output_csv는 결과 경로 넣어주세요\n",
        "\n",
        "# Example usage with actual paths\n",
        "folder_path = BASE_DIR+DATASET+'/vocal_samples' # Replace with actual folder path\n",
        "output_csv = BASE_DIR+DATASET+'/volume_analysis.csv'  # Replace with actual output CSV file path"
      ],
      "metadata": {
        "id": "9tQyREljWZaU"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "process_wav_files_torchaudio(folder_path, output_csv)"
      ],
      "metadata": {
        "id": "ez5f7SoTWiaj"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}