Neko Translator

Neko Translator - A professional AI-powered tool for automatically translating manga/comic images — from extracting dialogue, to summarizing each page's context with a multimodal model, to fine-grained translation by a large language model. Run the whole pipeline automatically in batches, or refine it box by box manually, then export or read with a single tap.

Automated translation pipeline: Text-box detection → text recognition (OCR) → scene-context parsing → translation. All four steps are independently controllable, and each can be skipped or re-run as needed.
Text-box detection / OCR: Automatically outlines dialogue and text regions in the image, with support for manually adding boxes, running OCR individually, reordering, and more.
Fine-grained LLM translation: Translates with scene context taken into account, and supports a proofreading workflow for easy human refinement.
Bring your own AI provider: Compatible with the OpenAI API, so you can connect a local Ollama, OpenRouter, and more. Model, concurrency, rate limits, and retries are all customizable, and your API key is stored only in your device's secure local storage.
Glossary support: Keep character names and proper nouns translated consistently, with "do not translate" and "banned terms" options to ensure consistent wording throughout the book.
Scene-context analysis: Parses the image content and uses the source text and summaries of adjacent pages as context, so the AI keeps dialogue coherent from page to page.
Batch processing: Select multiple pages and chapters at once for automatic queued processing, with real-time progress and automatic retries on failure.
Built-in reader: Horizontal / vertical layouts, paged / continuous scrolling — the reading direction is fully customizable.
One-tap export: Export to LabelPlus TXT, or package everything together with the original images as a ZIP, making the follow-up typesetting work easy.