Audio Transcription Scripts

This repository contains Python scripts for transcribing audio files to text using various methods.

Features

• Support for multiple audio formats (MP3, WAV, M4A, FLAC, etc.)
• Multiple transcription engines (OpenAI Whisper, Google Speech Recognition)
• Various output formats (TXT, SRT, VTT)
• Command-line interface
• Simple and advanced versions

Installation

1. Clone this repository:

git clone <repository-url>
cd transcribe-audio-to-text

2. Install Python dependencies:

pip install -r requirements.txt

3. Install FFmpeg (required for audio processing):
   • Windows: Download from https://ffmpeg.org/download.html
   • macOS: brew install ffmpeg
   • Linux: sudo apt install ffmpeg

Usage

Simple Transcription (simple_transcribe.py)

Basic usage for quick transcriptions:

python simple_transcribe.py audio_file.mp3

This will create a text file with the transcription.

Advanced Transcription (transcribe_audio.py)

Full-featured script with more options:

# Basic usage
python transcribe_audio.py audio_file.mp3

# Specify output file
python transcribe_audio.py audio_file.mp3 -o transcript.txt

# Use different Whisper model (tiny, base, small, medium, large)
python transcribe_audio.py audio_file.mp3 -m large

# Output as subtitle file (SRT format)
python transcribe_audio.py audio_file.mp3 -f srt

# Use Google Speech Recognition (requires internet)
python transcribe_audio.py audio_file.mp3 --method google

Command Line Options

• -o, --output: Specify output file path
• -m, --model: Choose Whisper model size (tiny, base, small, medium, large)
• -f, --format: Output format (txt, srt, vtt)
• --method: Transcription method (whisper, google)

Supported Audio Formats

• MP3
• WAV
• M4A
• FLAC
• AAC
• OGG
• WMA

Transcription Methods

OpenAI Whisper (Recommended)

• Works offline
• High accuracy
• Multiple model sizes available
• Supports 99 languages

Google Speech Recognition

• Requires internet connection
• Good accuracy for English
• Free tier available

Model Size Comparison

Model	Size	Speed	Accuracy	Use Case
tiny	39MB	Fastest	Good	Quick drafts
base	142MB	Fast	Better	General use
small	244MB	Medium	Good	Balanced
medium	769MB	Slow	Better	High quality
large	1550MB	Slowest	Best	Maximum accuracy

Examples

Transcribe a podcast episode:

python transcribe_audio.py podcast_episode.mp3 -m medium -o podcast_transcript.txt

Create subtitles for a video:

python transcribe_audio.py video_audio.wav -f srt -o subtitles.srt

Quick transcription of a voice memo:

python simple_transcribe.py voice_memo.m4a

Tips for Best Results

1. Audio Quality: Use clear, high-quality audio for better transcription
2. Model Selection: Use larger models for better accuracy, smaller for speed
3. File Size: Large files may take significant time to process
4. Language: Whisper auto-detects language but works best with English
5. Background Noise: Minimize background noise for better results

Troubleshooting

Common Issues:

1. FFmpeg not found: Install FFmpeg and ensure it's in your system PATH
2. Memory errors: Use smaller Whisper models or process shorter audio segments
3. Slow processing: Use smaller models or ensure you have adequate hardware

Error Messages:

• File not found: Check the audio file path
• Unsupported format: Convert to a supported format first
• Model loading failed: Check internet connection for first-time model download

License

This project is open source and available under the MIT License.