Auto Transcription

A script made in 2022 to automate the transcription of audio files using Whisper AI.

Table of Contents

• Description
• Features
• Prerequisites
• Installation
• Input File Organization
  • Example Directory Structure
• Usage
• Directory Structure

Description

Auto Transcription is a bash script designed to automate the transcription of audio files using Whisper AI. The script processes audio files, converts them to a standard format, and generates transcriptions that is added to the id3v2 tag and also saved as txt in a structured manner.

Features

• Automated transcription of audio files using Whisper AI.
• Conversion of audio files to a standard format (320kbps MP3).
• Organization of output files in a structured directory.
• Logging of processing details.
• Removal of special characters and normalization of file names.

Prerequisites

• Bash
• Python 3
• ffmpeg
• id3v2
• Whisper AI model

Installation

1. Clone the repository:

   git clone https://github.com/your_username/auto_transcription.git
   cd auto_transcription

2. Install the required dependencies:

   sudo apt-get install ffmpeg id3v2
   pip install openai-whisper

3. Ensure the config.xml file is correctly set up with the input and output paths:

   <general_params>
       input_path=/path/to/your/input
       output_path=/path/to/your/output
   </general_params>

4. Ensure the right model is selected in tools/whisper_fun.py:

   model = whisper.load_model("large-v3")

Input File Organization

To ensure the run.sh script processes your audio files correctly, you need to organize your input files in a specific directory structure. Follow these guidelines:

1. Input Directory Structure:

   • The input directory should be specified in the config.xml file under the <input_path> tag.
   • Inside the input directory, create subdirectories for each person. Each person's directory should contain subdirectories for different events.

2. Event Directory Naming:

   • Each event directory should be named in the format yyyy_mm_dd_eventName, where yyyy is the year, mm is the month, and dd is the day of the event.
   • Example: 2024_05_24_BirthdayParty.

3. Audio Files:

   • Place the audio files (in .mp3 format) inside the respective event directories.
   • Ensure that the audio files are named appropriately, as the script will process and rename them.

Example Directory Structure

/media/pedro/Arquivos/acervos/input
├── Person1
│   ├── 2024_05_24_BirthdayParty
│   │   ├── track1.mp3
│   │   ├── track2.mp3
│   └── 2024_06_15_Conference
│       ├── track1.mp3
│       ├── track2.mp3
├── Person2
│   ├── 2024_07_10_Wedding
│   │   ├── track1.mp3
│   │   ├── track2.mp3

Usage

1. Place your audio files in the input directory specified in config.xml.
2. Run the script:

   ./run.sh

The script will process the audio files, generate transcriptions, and save the output in the specified output directory.

Directory Structure

bash/auto_transcription
├── config
│   └── config.xml
├── logs
├── run.sh
└── tools
    ├── aux_functions.sh
    ├── gather_txt.sh
    └── whisper_fun.py