pyVideoTrans

Appearance

Starting from version 1.92, you can use ByteDance Volcano Engine's models for translation in the software. Specific activation and usage methods are as follows:

500 million tokens of free usage are available between May 15th and August 30th on ByteDance Volcano Engine.

Step 1: Login & Registration ​

Open the website: https://console.volcengine.com/auth/login/ Register if you don't have an account, otherwise log in directly.

Step 2: Real-name Authentication ​

This is a mandatory requirement by Chinese law for domestic services. You must complete it to use the service. After logging in, open the account center address and click on real-name authentication.

https://console.volcengine.com/user/authentication/detail/

Facial recognition or bank card authentication is optional.

Step 3: Activate and Claim Benefits ​

After successful real-name authentication, open the address https://console.volcengine.com/ark/region:ark+cn-beijing/openManagement until you see "Activation Management".

Then click any "Activate Service". In the opened page, select all models and confirm activation.

Click Submit.

After clicking OK and refreshing the activation management page, you will see that all services have been activated.

Now that it's activated, how do you use it in the code? Keep reading.

Create API Key ​

Below the "Activation Management", you can see the "API key" menu, click it.

There's one last step: create an "Inference Endpoint". Use this endpoint name instead of the model name where it's required.

Create Model Inference Endpoint ​

In the same location, look above to find the "Model Inference" menu and click it.

After clicking "Create Inference Endpoint", the following interface will pop up. Fill in the endpoint name as you wish and click "Add Model". A large number of available models will appear.

As shown below, in the model selection, it is recommended to only select the models that exist in "Activation Management", such as those starting with Doubao-lite, Doubao-pro, Moon. After selecting, several model versions will appear on the right. It is recommended to only select the numeric version names, such as 240515.

Here's my selection:

After creation, return to the "Model Inference" list page. Pay attention to the text starting with ep, which is the "Endpoint Name" to be used in the code.

This name and the API key are the two required parameters in code usage.

Using in pyVideoTrans ​

https://github.com/jianchang512/pyvideotrans

  1. Upgrade pyVideoTrans to version 1.92+ or higher. The address is above.
  2. Open the videotrans/set.ini file and search for zijiehuoshan_model=. Fill in your endpoint name after the "=" sign. Note that this name must be the automatically generated one.

.

The effect after filling is similar to:

zijiehuoshan_model=ep-20240615073625-zrlbj,ep-41521-auyed

If there are multiple endpoints, separate them with English commas.

  1. Open the software, go to the Settings menu, find ByteDance Volcano Engine, fill in your API key, and select the endpoint to use.
  2. If the test is successful, you can use it on the main interface.

Usage Notes ​

For well-known reasons, when using this model, be careful not to send or ask any sensitive questions or content containing sensitive words, and unhealthy content. Minor issues may result in a "I cannot answer this question" response, while serious issues may lead to account suspension.