Automatic Translation

Traditional internationalization solutions all require manual translation. VoerkaI18n supports calling online translation services for automatic translation.

The voerkai18n translate command can call online translation services to automatically translate extracted text.

In VoerkaI18n 3.0, two translation engines are supported:

• Baidu Translation
• AI Translation

Currently supports accessing the Baidu Online API for automatic translation. Baidu provides a free online API, although it only supports QPS=1 (one call per second). However, the voerkai18n translate command merges the text to be translated before making the API call, which is sufficient for most cases.

Basic Usage

Simply calling voerkai18n translate in your project directory will automatically translate all JSON files in the current project.

Step 1: Apply for Online Translation Service API

• Baidu Online Translation Service

The Baidu online translation service can be found here

• AI Translation Service

Any online large language model service that provides OpenAI API compatible interfaces, such as deepseek or Qwen.

We recommend using Qwen-max-latest, which can be applied for here

Note

The above online translation services generally have some free quota, which is usually sufficient for small-scale translation.

Step 2: Configure API

After obtaining the API for the above translation services, configure the API access in languages/api.json as follows:

{
    "deepseek":{
        "key":"sk-a27804xxxxxxxxxxxxxxxxxxxxxxxxxxxx",
        "url":"https://api.deepseek.com/chat/completions",
        "model":"deepseek-chat"
    },
    "qwen":{
        "key":"sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
        "url":"https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions",
        "model":"qwen-max-latest"
    },
    "baidu":{
        "key":"xxxxxxxxxxxxxxx",     // appkey
        "id":"xxxxxxxxxxxxxxx"       // appid
    }
}

Step 3: Execute Translation

• AI Translation

> voerkai18n translate --api <name declared in api.json>
> voerkai18n translate --api qwen

• Baidu Translation

> voerkai18n translate --api baidu --provider baidu

AI Translation

VoerkaI18n requires that the large language model used for AI Translation supports OpenAI API compatible interfaces.

Generally, the translation quality of AI Translation is better than Baidu Translation. However, the quality of AI Translation depends on the model being used, and different models may have varying translation quality. Additionally, the prompt used when submitting translations can also affect the translation results, especially when translating large blocks of text.

VoerkaI18n allows users to fine-tune the translation prompts.

• myapp
  • src
    • languages
      • prompts/Prompts
        • tranlate-messages.md Translation prompts
        • translate-paragraphs.md
      • api.json
      • ...

• tranlate-messages.md and translate-paragraphs.md are prompts for translating short sentences and paragraphs respectively, which users can adjust according to their needs.

Note

Since the quality of large model translation depends on the model and prompts being used, hallucinations may sometimes result in inaccurate translations. Therefore, when using AI Translation, you may need to adjust the prompts as needed.

Baidu Translation

• Baidu Translation also provides very good translation quality, but in some special cases, there might be inaccurate translations. For example, Baidu Translation might have issues handling interpolation variables, which would require manual adjustment.

• Baidu Translation API has a certain free quota and also has restrictions on call frequency. Generally, you need to configure QPS=1, meaning one call per second.

• VoerkaI18n merges the text to be translated before making the API call, which is sufficient for most cases.