Feature Description#

Designed for multilingual localization scenarios, this tool integrates batch processing, high-performance caching, smart context handling, and multi-language export to help you build an efficient and stable translation workflow.

Core Features#

Multi-Language Translation: Translate Once, Output Many#

You can translate the same file into multiple languages simultaneously, perfect for multilingual subtitles or internationalization projects:

• Example: Translate English subtitles into Chinese, Japanese, German, French, etc.
• Support: 70+ languages currently supported, with more on the way.

High-Performance Translation Cache#

The tool is built with an enterprise-grade instant caching system powered by IndexedDB:

• Unlimited Capacity: Bypasses browser LocalStorage limits to store millions of translation records without consuming memory.
• Persistent Storage: Translation results are saved permanently and persist across browser refreshes or restarts.
• Smart Hitting: Generates a unique key based on Source Text_Target Language_Source Language_Translation API_Model Settings. Cache is called directly when parameters match, achieving "translate once, instant access forever".
• On-Demand Management: Supports manual cache disabling for specific scenarios (like debugging Prompts) or one-click clearing in settings.

Smart Concurrency & Long Text Optimization#

For long documents and batch tasks, the tool utilizes a refactored task scheduling engine:

• Concurrency Control: Built-in request queue mechanism supports custom Translate Rate. It allows you to maximize paid API usage or limit rates to protect free APIs from being banned.
• Large File Support: Uses streaming and chunking strategies to easily handle ultra-large subtitles or Markdown documents, supporting concurrent rendering of large texts in the editor.
• Context Awareness: Intelligently slices context windows for subtitle or document translation, ensuring LLMs understand plot continuity and improving translation accuracy.

---

Translation Modes & Workflow#

The tool automatically detects user actions and switches modes:

• Batch Mode (Default): Drag and drop multiple files to upload. They enter a processing queue automatically, and results are packed and downloaded upon completion.
• Single-File Mode: Upload a single file or paste text to enter single-file mode, supporting real-time preview and line-by-line proofreading/editing.

Note: If the current tool supports it, you can also manually toggle Single-File Mode in Advanced Settings to force a preview/proofreading-oriented workflow.

Advanced Settings#

Settings Import/Export#

Supports one-click backup of all current configurations (including API Keys and Prompt settings). The generated configuration file can be imported across devices, facilitating team collaboration or environment migration.

General Translation Options#

• Use Cache: Enabled by default. When parameters match, results are read from IndexedDB cache to speed up batch jobs; you can disable it temporarily when debugging Prompts/model settings.
• Retry Count: Maximum number of retries when a request fails (increase for unstable networks or rate-limited/free endpoints).
• Retry Timeout: Per-request timeout threshold (seconds). Increase when APIs respond slowly or when translating long content.
• Remove Characters After Translation: Automatically removes specified characters/fragments from the translated result (e.g., ♪ in subtitles, or noisy HTML fragments like <i>).
• Custom Export Filename: Unifies file naming when exporting/downloading. Supports placeholders: {name} (original filename without extension), {lang} (target language), {ext} (extension), {date} (date, YYYY-MM-DD), {time} (time, HHMMSS). Example: {name}_{lang}_{date}_{time}.{ext}.

API Parameter Tuning#

To balance translation quality and speed, it is recommended to adjust the following parameters based on the selected API type:

Chunk Size#

For non-LLM interfaces (like Google/Azure API), the tool automatically splits long text into small chunks. Chunk Size refers to the maximum character count per chunk. Common limits are:

Translation Service	Max Characters per Request
DeepL API	128,000 characters
DeepLX Free	1,000 characters
Azure Translate	10,000 characters
Google Translate (Web)	5,000 characters
Google Cloud API	30,000 characters

⚠️ Note: The Google Translate web version may break line breaks, so chunked translation is not used.

Delay (ms)#

The cooldown time between chunk requests, helpful to avoid frequent failures:

• Recommended to set a higher delay for poor networks or free APIs.
• Example: When using the Azure Translate free tier, a setting of 5000ms or higher is recommended.

Concurrent Lines#

• Default: 20
• Description: The maximum number of lines translated simultaneously.
• Note: Setting this too high may lead to API Rate Limiting or empty responses.

Context Lines#

• Default: 50
• Description: The number of lines included per batch in Context-Aware Translation mode.
• Note: Higher values provide richer context but may exceed the model's Token context limit.