jevbooks

← All projects

jev-subtitle-translator

GeekLinkDev/jev-subtitle-translator · Homepage

Translate SRT subtitles with LLMs and QC every line with Jev.

Verification & Guardrails61%Runner-up: Evaluation & BenchmarkingFunnel

What it is

An open-source SRT translator and subtitle QC tool for LLM-powered workflows. It translates SRT files with OpenRouter or local models, preserves every cue, and flags suspicious lines for human review before export.

How it uses Jev

After deterministic checks, Jev reviews every non-empty source–translation pair and answers one question: does this translation need human review? Jev flags likely defects such as missing meaning, changed names or numbers, or reversed negation. Flags are suggestions for inspection, not proof of error; Jev does not rewrite translations.

Primitives:noul

Technique worth stealing

Parse SRT locally, translate ID-tagged batches with JSON Schema output, validate returned IDs, and retry only missing or empty IDs so one failure cannot shift later cues.

Try it

bash run_web.sh, or use the CLI: jev-subtitle-translator translate input.srt --source-language en --target-language de --model openai/gpt-4o-mini --output translated.srt

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.81
  • Shows a System One patternyes0.61
  • Handles uncertaintyno0.08
  • Measuredno0.06
  • Runnableyes0.85
  • Worth recommendingunclear0.50
  • Model replicano0.05
  • Problem scopescore on a 0–2 scale1.06
  • About Jevyes0.93

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 23 Sept 2026.