Feature Request: Add Traditional Chinese Output Option for Speech-to-Text

Suggestion
1

Dear Shotcut Development Team,

Hello!

I am an enthusiastic user of Shotcut, and I’m very grateful for the excellent and powerful open-source video editing software you provide. I frequently use various Shotcut features in my video production.

Recently, I started experimenting with Shotcut’s built-in “Speech to Text” feature and found it to be very practical for converting spoken content into text. However, I’ve encountered an issue and would like to suggest a functional improvement for future versions.

Problem Description: When I set the “Language” option to “Chinese” in the “Speech to Text” settings, the generated subtitles are defaulted to Simplified Chinese, even if the spoken language is Traditional Chinese.

My Need/Suggestion: Given that Traditional Chinese is widely used in Taiwan, Hong Kong, and many overseas Chinese communities, I hope Shotcut can provide a more precise language option within the “Speech to Text” feature. This would allow users to directly select Traditional Chinese as the output language for subtitles.

Currently, I have to generate Simplified Chinese subtitles first, and then manually convert them to Traditional Chinese using a third-party tool (such as Python’s OpenCC library or an online converter). This adds an extra step to my workflow and impacts efficiency.

Suggested Improvement: I propose adding a clear option in the “Language” dropdown menu for “Speech to Text,” in addition to the existing “Chinese” option. For example, options like:

  • “Chinese (Traditional)”
  • Or even a more granular choice between “Chinese (Simplified)” and “Chinese (Traditional)”.

I believe this improvement would greatly enhance Shotcut’s usability and appeal within the Traditional Chinese user community, saving us time on subtitle post-processing.

Thank you again for all your efforts and contributions to Shotcut. I look forward to Shotcut becoming even more comprehensive in future versions.

Sincerely,

2

We do not make this. You need to ask OpenAI to update their model and Whisper.cpp to update their model based on that.
See https://platform.openai.com/docs/guides/speech-to-text/supported-languages/#supported-languages

3

This person found that switching to a larger model made Simplified Chinese work better for them. I doubt the same thing applies, but I thought I would mention it. Keep in mind that Shotcut does not currently include translation of subtitles, and this request is similar to that.

4

I found that by using specific prompts, you can control whether the model outputs Simplified Chinese or Traditional Chinese. For example, using a prompt like ‘以下是普通话的句子。’ (which means ‘The following are sentences in Mandarin.’ and uses Simplified Chinese) makes the model output Simplified Chinese. If you need to output Traditional Chinese, simply change the prompt to ‘以下是普通話的句子’ (the Traditional Chinese version of the same sentence). It would be great if Shotcut could add a ‘Custom Parameters’ or ‘Prompt’ field in the GUI, so users can easily set such prompts without having to wrap or replace the executable.

5

Is this the --prompt command line option of whisper-cli? Do you suggest one new UI field to target one CLI option, or a field to input any additional CLI options? Or, do you suggest to make it even easier for users by adding a new language option that automatically adds your suggested prompt?

6

Thank you for the confirmation. You’re exactly right – I was referring to the --prompt command line option of whisper-cli.

Regarding the implementation, my preference is to add a dedicated Prompt field (directly corresponding to --prompt) for better intuitiveness. It would be even better if you could also add a general “custom parameters” field, but that’s up to you.

One small UI suggestion: add a tooltip icon (ⓘ or ?) next to the Prompt field, with a message like:

Example usage: When a language has multiple variants (e.g., Simplified vs. Traditional Chinese), you can guide the model’s output variant by a prompt. Just enter the appropriate prompt text in the Prompt field.

Without such a hint, users are unlikely to realize that variant switching can be done via a prompt.

Due to time zone differences, my reply times may look odd – sorry about that. Thanks again for considering this feature!