[BUG] -

Created by: Waying13

Bug Description

When uploaded Chinese language audio, we default choose paraformer-zh, and Qwen2.5-7b model to do inference, but if uploaded the English language audio, I cannot see the other model option to process this language inference, but the backend actual could choose whisper model to do inference . Is there any inconsistency to handle different language of input audio? Or give some configuration to user to choose?

System Setup

Only see the fixed model option, cannot choose other model to match different input language audio

Reproducible Steps

1. Upload Chinese language audio file, get results
2. Upload English language audio file, cannot modify model selection

Root Cause Analysis

No response