Media Transcriber

Local files: ./target/release/media-transcriber --source /path/to/your/audio.mp3

A streamlined tool to transcribe any audio or video media from various sources using OpenAI's Whisper API. This tool can process podcast RSS feeds, YouTube videos, YouTube channels, and more.

Overview

This tool provides high-performance transcription services using a Rust implementation that:

• Processes podcast RSS feeds - downloading and transcribing all episodes
• Processes YouTube videos - transcribing individual videos
• Processes YouTube channels/playlists - transcribing all videos from a channel
• Processes local MP3 files - transcribing audio files from your local filesystem
• Processes multiple sources at once from a sources file

All transcriptions are performed using OpenAI's Whisper API, which provides high-quality multilingual transcription.

Installation

# Clone the repository
git clone https://github.com/deepakjois/podscript.git
cd podscript

# Navigate to the rust-transcriber directory
cd rust-transcriber

# Run the installation script (installs Rust if needed)
./install_rust.sh

# Alternatively, if you already have Rust installed:
cargo build --release

Getting Started

# Navigate to the rust-transcriber directory
cd rust-transcriber

# Process a podcast RSS feed
./target/release/media-transcriber --source https://example.com/podcast.rss

# Process a YouTube video
./target/release/media-transcriber --source https://www.youtube.com/watch?v=VIDEO_ID

# Process a local MP3 file
./target/release/media-transcriber --source /path/to/your/audio.mp3

# Process with language and prompt specification
./target/release/media-transcriber --source URL --language en --prompt "Tech podcast"

# Process multiple sources from a file
./target/release/media-transcriber --file sources.txt

Rust Implementation Features

The Rust implementation offers excellent performance characteristics:

# Navigate to the rust-transcriber directory
cd rust-transcriber

# Run the transcriber with various options
./target/release/media-transcriber --source <url> [options]

Key benefits:

• Higher Performance: Faster processing and better multithreading
• Better Memory Management: More efficient handling of large files
• Improved Error Handling: Robust recovery from transient failures
• Consistent Behavior: Reliable operation across different platforms

Requirements

• OpenAI API key for Whisper transcription
• Rust 1.56.0 or later (installed via the provided script)
• Dependencies:
  • ffmpeg (for audio processing)
  • yt-dlp (for YouTube sources)

API Key Configuration

The implementation supports multiple API key sources (in order of precedence):

1. Command-line option: --api-key YOUR_API_KEY
2. Environment variable: OPENAI_API_KEY=YOUR_API_KEY
3. .env file: In the current directory, parent directory, or podscript subdirectory

Large File Handling

The implementation handles audio files exceeding the 25MB Whisper API limit by:

1. Splitting files into smaller chunks
2. Transcribing each chunk separately
3. Combining the results into a single transcript

This offers efficient processing of large files with excellent parallelization and memory management.

Output Structure

Transcripts are organized in a consistent directory structure:

transcripts/
├── Podcast_Name/
│   ├── podcast_info.txt
│   ├── Episode_Title_1/
│   │   └── transcript.txt
│   └── Episode_Title_2/
│       └── transcript.txt
├── YouTube_Channel/
│   ├── channel_info.txt
│   ├── Video_Title_1/
│   │   ├── video_info.txt
│   │   └── transcript.txt
│   └── Video_Title_2/
│       ├── video_info.txt
│       └── transcript.txt
└── local_files/
    └── Audio_File_Name/
        ├── file_info.txt
        └── transcript.txt

Troubleshooting

Common Issues

1. API Key Not Found: Ensure your OpenAI API key is correctly set using one of the methods described above.

2. Dependencies Missing: Verify that all required dependencies are installed:

   # Check ffmpeg installation
   ffmpeg -version
   
   # Check yt-dlp installation
   yt-dlp --version

3. File Format Issues: Ensure your audio files are in a format supported by ffmpeg (MP3, WAV, M4A, etc.).

4. Compilation Errors: If you encounter compilation errors, ensure your Rust toolchain is up-to-date:

   rustup update

5. Running from Wrong Directory: Make sure to run the Rust binary from within the rust-transcriber directory or specify the full path.

License

MIT