What it is

TranslateBooksWithLLMs is a Python desktop application under AGPL-3.0 for translating full-length books, documents, and subtitle files with large language models. It lives in the Content and Publishing ecosystem, specifically around ebook translation, document localization, and publishing-related workflows. The project ships as a local app that opens a browser interface at http://localhost:5000 after the bundled executable is run.

The concrete problem it addresses is translating long files while preserving document structure and continuing interrupted work. The README states that EPUB, SRT, DOCX, and TXT files can be processed with no file size limit, and that an intelligent chunking system preserves context between segments. It also states that formatting is preserved: EPUB styles and structure remain intact, SRT timecodes stay synchronized, and tags and timestamps are kept.

Key capabilities

• The application translates EPUB, SRT, DOCX, and TXT files through a desktop interface that accepts a file and a target language.
• It processes documents of any length by chunking content while preserving context between segments.
• It preserves EPUB formatting, styles, and structure, and it keeps SRT timecodes synchronized.
• It saves automatic checkpoints, so an interrupted translation can resume where it stopped.
• It supports reusable writing style presets extracted from sample books or written by hand.
• It applies a consistent register, rhythm, and imagery across chunks when a style preset is used.
• When Auto is selected, it derives a glossary and style from the document with one extra LLM call each before the job starts, and it saves nothing.

Who uses it and how

• Individual readers or translators convert EPUB books into another language while keeping the original ebook structure.
• Subtitle editors translate SRT files and rely on the tool to preserve timecodes.
• Document users translate DOCX or TXT files with no file size limit.
• Privacy-focused users run Ollama locally so translation requests do not leave their machine.
• Users connect DeepSeek, Gemini, Mistral, NVIDIA NIM, OpenAI, OpenRouter, or Poe with an API key, or they connect Ollama or an OpenAI-compatible server locally.

Getting started

Users download a Windows or macOS release archive, extract it, run TranslateBook.exe or ./TranslateBook, and open http://localhost:5000 in a browser; the README also mentions installing from source. On first launch they choose a provider, such as Ollama locally, a cloud provider by API key, or an OpenAI-compatible server by endpoint.

When to use it — and when not to

TranslateBooksWithLLMs is useful when the main need is long-file translation with format preservation and resume support, especially for EPUB and SRT workflows. It is less suitable when users need a hosted web service, because the README describes a desktop app that runs locally and exposes a localhost interface. It also depends on the selected model or provider, so local users must operate Ollama or another OpenAI-compatible server, while cloud users must supply API keys.

git clone https://github.com/hydropix/TranslateBooksWithLLMs.git
cd TranslateBookWithLLM
ollama pull qwen3:14b    # Download a model

# Windows
start.bat

# Mac/Linux
chmod +x start.sh && ./start.sh

# Basic (auto-generates "book (Chinese).epub")
python translate.py -i book.epub -sl English -tl Chinese

# With OpenRouter
python translate.py -i book.txt --provider openrouter \
    --openrouter_api_key YOUR_KEY -m anthropic/claude-sonnet-4 -tl French

# With OpenAI
python translate.py -i book.txt --provider openai \
    --openai_api_key YOUR_KEY -m gpt-4o -tl French

# With Gemini
python translate.py -i book.txt --provider gemini \
    --gemini_api_key YOUR_KEY -m gemini-2.0-flash -tl French

# With Mistral
python translate.py -i book.txt --provider mistral \
    --mistral_api_key YOUR_KEY -m mistral-large-latest -tl French

# With DeepSeek
python translate.py -i book.txt --provider deepseek \
    --deepseek_api_key YOUR_KEY -m deepseek-v4-pro -tl French

# With Poe
python translate.py -i book.txt --provider poe \
    --poe_api_key YOUR_KEY -m Claude-Sonnet-4 -tl French

# With NVIDIA NIM
python translate.py -i book.txt --provider nim \
    --nim_api_key YOUR_KEY -m meta/llama-3.1-8b-instruct -tl French

# With local OpenAI-compatible server (llama.cpp, LM Studio, vLLM, etc.)
python translate.py -i book.txt --provider openai \
    --api_endpoint http://localhost:8080/v1/chat/completions -m your-model -tl French

# Provider
LLM_PROVIDER=ollama

# Ollama
API_ENDPOINT=http://localhost:11434/api/generate
DEFAULT_MODEL=qwen3:14b

# API Keys (if using cloud providers)
OPENROUTER_API_KEY=sk-or-v1-...
OPENAI_API_KEY=sk-...
GEMINI_API_KEY=...
MISTRAL_API_KEY=...
DEEPSEEK_API_KEY=...
POE_API_KEY=...
NIM_API_KEY=...

# Performance
REQUEST_TIMEOUT=900
MAX_TOKENS_PER_CHUNK=450     # Token-based chunking (default: 450 tokens)
PARALLEL_TRANSLATIONS=1      # Concurrent chunks (cloud only; Ollama stays at 1)