FEATURE: Generate subtitle feature for multilingual videos #77
Description
🎯 Problem Statement
ScriptAI currently lacks an automated subtitle generation feature, which is essential for accessibility and reaching a global audience on YouTube.
💡 Proposed Solution
- Implement a feature that takes a video as input and generates subtitles
- Enable users to upload videos, Store videos securely in Supabase Storage and Record video metadata (user ID, video ID, language) in Supabase database.
- Extract high-quality audio from videos using FFmpeg or similar tools server-side. Perform noise reduction and normalize audio volume to improve clarity.
- Send preprocessed audio or audio chunks to Gemini 2.0 Flash API for text transcription. Specify the language spoken in the API call for better model accuracy. Receive transcripts with timestamps to enable subtitle timing.
- Show video and subtitle preview and edit subtitles at the same time for best user experience
- limit upload video limit to 5-10 minutes only
- Store subtitle files and transcripts in Supabase Storage.
- Provide UI for users to download subtitle files in .SRT, .VTT and .TXT file in each language.
🔄 Alternative Solutions
- Manual subtitle creation by users
📊 Impact Assessment
- User Impact: High; improves accessibility and reach
- Development Effort: High
- Priority: High
🧪 Acceptance Criteria
- Video input supported
- Subtitles generated in at least three languages
- Subtitles exportable in standard formats
🔗 Related Features
Script generation, export tools
📚 Additional Context
Consider using open source models like OpenAI Whisper or Google APIs for speech recognition.
📋 Checklist
- I have searched the existing issues for duplicates
- I have provided a clear problem statement
- I have described the proposed solution
- I have considered alternative approaches
- I have provided acceptance criteria