Skip to content

Content Creator Guide

This guide helps content creators understand how to use the SRT Translator and interpret its outputs.

What is ai_config.json?

The ai_config.json file is a snapshot of your translation run that contains:

Essential Information

  • Version: The SRT Translator version used
  • Timestamp: When the translation was performed
  • Target Languages: Which languages you translated to
  • Source Files: Which files were translated

Translation Settings

  • DNT Terms: "Do Not Translate" terms that were preserved
  • Termbases: Custom translations for specific terms in each language
  • Model: Which AI model was used for translation
  • Mode: Whether you used CLI or GUI

Why This Matters

This file ensures that: - Quality evaluation uses the exact same settings as your translation - Reproducibility - you can re-run evaluation later with identical results - Audit trail - complete record of what was translated and how

Understanding Coverage Reports

When you run evaluation, you'll see coverage information that tells you how complete your translation setup was:

DNT Coverage

  • dnt_coverage: "present" - You provided "Do Not Translate" terms
  • dnt_coverage: "none" - No DNT terms were specified

What this means: DNT terms help preserve important terminology (like product names, technical terms) that shouldn't be translated.

Termbase Coverage

  • termbase_coverage: "full" - All target languages have custom term translations
  • termbase_coverage: "partial" - Some languages have custom terms, others don't
  • termbase_coverage: "none" - No custom term translations were provided

What this means: Termbases ensure consistent translation of important terms across all languages.

Entry Counts

The report shows how many custom terms you have for each language:

"termbase_entry_counts": {
  "es": 15,
  "fr": 12,
  "de": 8
}

Reading Quality Reports

What to Look For

  1. Decision Banner: Shows pass (✅), review (⚠️), or fail (❌)
  2. Coverage: Make sure DNT and termbase coverage match your expectations
  3. Issues: Review any flagged problems in the punch list
  4. Language-specific: Check file status for each target language

Common Issues and Solutions

"Term Issues" - High Count

  • Cause: Important terms being translated when they shouldn't be
  • Solution: Add missing terms to your DNT list or termbase

"Empty Subtitles"

  • Cause: Usually segmentation artifacts, not translation problems
  • Solution: This is normal and doesn't affect quality

"CPS Violations"

  • Cause: Text too long for subtitle timing
  • Solution: Consider breaking long sentences or adjusting timing

If Evaluation Fails

Common Failure Reasons

  1. Missing ai_config.json - Translation didn't complete properly
  2. Missing source files - Original SRT files not found
  3. Missing translated files - Translation output incomplete
  4. Corrupted files - Files damaged during processing

What to Do

  1. Re-run translation - Most failures are due to incomplete translation
  2. Check file permissions - Ensure the app can read/write to the output directory
  3. Verify disk space - Translation needs space for output files
  4. Contact support - If the problem persists, include the error message

Recovery Steps

  1. Delete the failed batch folder (starts with translation-batch-)
  2. Check your input files - Ensure they're valid SRT files
  3. Verify settings - Make sure your configuration is correct
  4. Try again - Most issues resolve with a fresh run

Best Practices

Before Translation

  1. Prepare your DNT list - Identify terms that shouldn't change
  2. Build your termbase - Create custom translations for important terms
  3. Test with a small file - Verify settings work as expected

During Translation

  1. Monitor progress - Watch for any error messages
  2. Don't interrupt - Let the process complete
  3. Check output - Verify files are created in the expected locations

After Translation

  1. Run evaluation - Get quality metrics and identify issues
  2. Review reports - Understand what went well and what didn't
  3. Save results - Keep the batch folder for future reference

Getting Help

Documentation

  • README.md - Project overview and quick start
  • INSTALLATION.md - Setup instructions
  • GUI_USER_MANUAL.md - Detailed GUI usage

Support

  • GitHub Issues - Report bugs or ask questions
  • Discussions - Community help and tips
  • Wiki - Additional guides and examples

Before Asking for Help

  1. Check the logs - Look in translation_issues_*.log for error details
  2. Verify your setup - Ensure you have the latest version
  3. Include details - Version, error messages, and steps to reproduce

Quick Reference

File Locations

  • Input: original_captions/ folder
  • Output: translated_srt_files/translation-batch-YYYYMMDD_HHMMSS_/
  • Logs: translation_issues_YYYYMMDD_HHMMSS_.log
  • Reports: artifacts/eval_report.json

Key Commands

  • GUI: Run srtx from command line
  • CLI: Run srtx-cli with appropriate flags
  • Evaluation: Automatically runs after translation

Success Indicators

  • ✅ Translation completes without errors
  • ✅ Output files created for each target language
  • ✅ Evaluation runs and produces reports
  • ✅ Coverage matches your expectations

Remember: The SRT Translator is designed to be reliable and user-friendly. Most issues can be resolved by re-running the translation or checking your configuration. When in doubt, start with a small test file to verify everything is working correctly.