DeovidLang: Automate Video Creation from Audio with AI Subtitles

Ever wished you could turn a simple audio recording into a polished video—with subtitles in multiple languages—all automatically? Whether you’re a content creator, educator, or localizer, you probably know how time-consuming it is to manually transcribe, translate, and sync subtitles.

What if there was a tool that could handle all of this for you in one go?

DeovidLang does exactly that. It takes your audio files, transcribes them using OpenAI’s Whisper, translates the text, generates bilingual subtitles, adds image overlays at specific timestamps, and produces a ready-to-share MP4 video.

What You’ll Learn

• How DeovidLang works under the hood
• How to set up and run the tool
• How to use different Whisper models
• How to customize language and output settings
• Real-world use cases for content creators

Key Features

How It Works

DeovidLang follows a straightforward pipeline:

Audio File → Whisper Transcription → Translation → SRT Subtitles → Image Overlay → Final MP4 Video

Step-by-Step Process

1. Input: You provide an audio file (M4A, MP3, WAV, etc.)
2. Transcription: Whisper converts speech to text
3. Translation: Translated text is generated in your target language
4. Subtitle Generation: Both original and translated text become SRT files
5. Image Overlay: Specified images appear at timestamps you define
6. Video Assembly: FFmpeg combines everything into the final MP4

Installation

Step 1: Install System Dependencies

FFmpeg is required for video processing:

sudo apt install ffmpeg

brew install ffmpeg

# Download from https://ffmpeg.org/download.html
# Or use winget:
winget install ffmpeg

Step 2: Install Python Dependencies

pip install pydub pillow

Step 3: Install Whisper

pip install git+https://github.com/openai/whisper.git

Step 4: Clone the Repository

git clone https://gitlab.com/krafi/deovidlang.git
cd deovidlang

Usage

Basic Command

python deovidlang.py <directory> <audio_file> --model <model> --language <language> --task <task>

Example

python deovidlang.py myproject myproject/speech.m4a --model medium --language en --task translate

Parameters

Real-World Use Cases

1. Educational Content Localization

Create bilingual videos for language learners. Transcribe your English lecture, translate to Spanish, and generate subtitles showing both languages simultaneously.

2. Podcast to Video

Turn podcast episodes into YouTube videos. Add cover art or branding images at the beginning, and let DeovidLang handle the rest.

3. Tutorial Video Creation

Record your voiceover, then automatically generate subtitles and create a professional video with slides matching your narration.

4. Accessibility

Make your content accessible to deaf or hard-of-hearing viewers with accurate auto-generated subtitles in multiple languages.

Example Output

After running DeovidLang, you’ll get:

output/
├── 2024-01-15/
│   ├── original.srt      # Original language subtitles
│   ├── translated.srt    # Translated subtitles
│   └── final.mp4         # Final video with everything

The final MP4 includes:

• Your original audio
• Image overlays at specified timestamps
• Burned-in bilingual subtitles

Troubleshooting

Why This Project is Useful

DeovidLang solves several pain points for content creators:

• Saves Hours: Manual transcription takes 4-6x the audio length. DeovidLang does it in minutes.
• No Expensive Tools: No need for Adobe Premiere, Final Cut, or subscription services.
• Multilingual Ready: Reach global audiences with translated subtitles.
• Automated Workflow: One command handles the entire pipeline.

Conclusion

DeovidLang is a powerful yet simple tool that automates video creation from audio. Whether you’re a educator, podcaster, or content creator, it handles the heavy lifting—transcription, translation, subtitles, and video assembly—so you can focus on creating content.

Give it a try and transform your audio files into shareable videos in minutes!

Source Code

View and contribute to the project: DeovidLang on GitLab

Happy automating!