DEV Community

Cover image for How to create video transcription with ffmpeg and whisper
Mark Kop
Mark Kop

Posted on Edited on

How to create video transcription with ffmpeg and whisper

#tutorial

Requirements

  • ffmpeg
  • whisper
  • Python 3.10+ (for Whisper)

Installation

macOS 

# Install Homebrew if you don't have it
/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

# Install ffmpeg
brew install ffmpeg

# Install Python (if needed)
brew install python

# Install Whisper
pip3 install --upgrade pip
pip3 install git+https://github.com/openai/whisper.git
Enter fullscreen mode Exit fullscreen mode

Windows 

# Install Chocolatey if you don't have it
# Run in PowerShell as administrator:
Set-ExecutionPolicy Bypass -Scope Process -Force
[System.Net.ServicePointManager]::SecurityProtocol = [System.Net.ServicePointManager]::SecurityProtocol -bor 3072
iex ((New-Object System.Net.WebClient).DownloadString('https://community.chocolatey.org/install.ps1'))

# Install ffmpeg
choco install ffmpeg

# Install Python (from python.org)
# Make sure to check "Add Python to PATH" during installation

# Install Whisper
pip install -U openai-whisper
Enter fullscreen mode Exit fullscreen mode

Linux 

# Install ffmpeg
sudo apt update && sudo apt install ffmpeg

# Install Python and pip
sudo apt install python3 python3-pip

# Install Whisper
pip3 install git+https://github.com/openai/whisper.git
Enter fullscreen mode Exit fullscreen mode

Transcription Steps

  1. Extract audio from video using ffmpeg 
   ffmpeg -i input_video.mp4 -vn -acodec mp3 output.mp3
Enter fullscreen mode Exit fullscreen mode
  1. Transcribe audio with Whisper 
   whisper output.mp3 --language English --model small --output_format txt
Enter fullscreen mode Exit fullscreen mode

Model Options

  • tiny: Fastest, lowest accuracy (~1GB RAM)
  • base: Fast, decent accuracy (~1GB RAM)
  • small: Balanced speed/accuracy (~2GB RAM)
  • medium: Good accuracy (~5GB RAM)
  • large: Best accuracy (~10GB RAM)

Output Formats

  • txt: Plain text transcript
  • srt: Standard subtitle format
  • vtt: Web Video Text Tracks format
  • json: Detailed JSON with timestamps

Additional Options

  • --task translate: Translates non-English audio to English
  • --language en: Specifies the source language (faster and more accurate)
  • --model: Selects the model size (tiny/base/small/medium/large)

Source: macos.gadgethacks.com
Source: dev.to

Top comments (1)

Collapse
 
testerslol123 profile image
Danny S •

When using that Whisper, no need to input API Key?