Audio Transcriber

A command-line tool that transcribes audio files to text using OpenAI's Whisper model. It supports local files and URLs, shows progress during transcription, and can handle various audio formats.

Features

• Transcribe audio files from local paths or URLs
• Support for multiple audio formats: MP3, WAV, M4A, OGG
• Progress indication during transcription
• Automatic clipboard copying of transcription
• Optional saving to a text file
• Memory-efficient chunked processing for large files

Prerequisites

• Python 3.7 or higher
• ffmpeg (required by pydub for audio processing)

Installing ffmpeg

• Ubuntu/Debian:

  sudo apt update
  sudo apt install ffmpeg

• macOS:

  brew install ffmpeg

• Windows: Download from ffmpeg.org and add to system PATH

Installation

1. Clone the repository:

   git clone [your-repo-url]
   cd audio-transcriber

2. Create and activate a virtual environment (recommended):

   • Linux/macOS:

     python -m venv venv
     source venv/bin/activate

   • Windows:

     python -m venv venv
     venv\Scripts\activate

3. Install required Python packages:

   pip install -r requirements.txt

4. Make the script executable and create a system-wide command:

   • Linux/macOS:

     chmod +x transcribe.py
     sudo ln -s "$(pwd)/transcribe.py" /usr/local/bin/transcribe

   • Windows:

     1. Add the script directory to your system PATH
     2. Create a batch file named transcribe.bat in the same directory:

        @echo off
        python "%~dp0transcribe.py" %*

Usage

• Basic usage:

  transcribe path/to/audio.mp3

• With output file:

  transcribe path/to/audio.mp3 -o /output/directory

• From URL:

  transcribe https://example.com/audio.mp3

Command-line Options

• audio_source: Path or URL to the audio file (required)
• -o, --output: Directory to save the transcription file (optional)

Supported Audio Formats

• MP3 (.mp3)
• WAV (.wav)
• M4A (.m4a)
• OGG (.ogg)

How It Works

1. Validates the input (local file or URL)
2. Downloads the audio file if it's a URL
3. Loads the Whisper model (tiny.en)
4. Splits the audio into 30-second chunks for efficient processing
5. Transcribes each chunk with progress indication
6. Combines the transcriptions
7. Copies the result to the clipboard and optionally saves it to a file
8. Cleans up temporary files and frees memory

Dependencies

The tool relies on the following Python packages:

• openai-whisper: Core transcription functionality (includes torch as a dependency)
• tqdm: Progress bar display
• requests: URL handling and downloads
• pyperclip: Clipboard operations
• pydub: Audio file processing and chunking

System Dependencies

• ffmpeg: Required by pydub for audio processing

Troubleshooting

Common Issues

1. ffmpeg not found:

   • Ensure ffmpeg is installed and in your system PATH
   • Try running ffmpeg -version to verify installation

2. Clipboard access error:

   • On Linux, install xclip: sudo apt-get install xclip
   • On Windows, ensure you have access to the clipboard

3. Memory issues with large files:

   • The tool automatically processes audio in chunks
   • If still experiencing issues, try closing other applications

Error Messages

• "File does not exist": Check the file path and permissions
• "Not a supported audio format": Convert your audio to a supported format
• "Could not copy to clipboard": Check clipboard access permissions