Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

GeekLink Jev Subtitle Translator

SRT subtitle translator that translates with any OpenRouter LLM and asks Jev a review Noul per line to flag defects before export.

ProjectbetaGPL-3.0-or-later
View on GitHub ↗

Overview

GeekLink AI Subtitle Translator is an open-source SRT translator and subtitle QC tool. It translates subtitles with an LLM of your choice through OpenRouter (GPT, Claude, Gemini, DeepSeek, Grok and others), keeping original cue numbers, order and timestamps unchanged. Missing or empty translations are retried by subtitle ID and flagged if they remain unresolved.

After deterministic checks, Jev reviews every completed source-translation pair. According to a list summary, it asks a Noul review question per line to flag omissions, changed meaning, names, numbers, negation or other defects for human review before export. You can compare source, translation and review status in a local web interface or use a CLI, then download the translated SRT and a QC report; QC can also run independently without calling a translation model. Documentation is available in four languages. A hosted product handles transcription and video workflows separately.

What the decision model decides

TypeQuestionAnswers
Yes / noDoes this translated line have a defect worth human review (omission, changed meaning, names, numbers, negation)?yes · no

How it works

  1. Translate SRT cues with an OpenRouter LLM, preserving timing
  2. Retry and flag missing lines
  3. Jev reviews each source-translation pair
  4. Human reviews flagged lines; export SRT and QC report

Strengths

  • Cue order and timestamps preserved
  • Every line checked, not sampled
  • QC usable on existing translations

Caveats

  • GPL-3.0 licence
  • Requires an OpenRouter key

Getting started

With Python 3.10+ and an OpenRouter key, run bash run_web.sh and use Translate + QC in the local web UI.

At a glance

Type
Project
Status
beta
Language
Python
Runtime
local web app + CLI
Has code
yes
License
GPL-3.0-or-later

Use cases and integrations

How curated lists describe it

  • Subtitle translation: asks Jev a `Noul` review question for each translated subtitle line to flag omissions, changed meaning, names, numbers, negation, or other defects for human review before export.

    yibie/awesome-jev

  • Translate SRT subtitles with structured LLM output and check every translation with Jev.

    hellogumbo/awesome-jev

Frequently asked questions

What does Jev check?

Each source-translation pair for defects worth human review.

Can I QC existing translations?

Yes, independent QC does not call a translation model.

Are timestamps changed?

No, cue numbers, order and timestamps are preserved.

Related projects

  • fake-real-jev

    Published Jev client and claim check from the Fake/Real evidence-review site, where Jev filters sources and judges each claim.

    ProjectproductionTypeScript
  • jev_project_context

    Evidence-first experiment-memory skill for coding agents, with an optional Jev audit layer that checks provenance and claim support.

    Agent skillbetaPython
  • jev-cli

    jevctl: Node CLI for typed Jev judgments (verify claims, screen agent hijacks, classify) with exit codes for CI gating.

    CLIbetaTypeScript
  • jev-decisions

    Hermes plugin that asks Jev to review risky plans, check claims against sources, and judge whether a task is really finished.

    Pluginbeta
  • jev-enforce

    Claude Code plugin that checks every reply and file edit against your CLAUDE.md rules with one Jev yes/no per rule.

    PluginbetaTypeScript
  • jev-fidelity

    Editorial QA tool where Jev judges per fact whether an edit preserved the original, with a confidence gate and human review fallback.

    ProjectbetaJava