可不可以添加一个inference.py，在指定一系列dir or文件名后就生成最终的音频呢？

抽离出pyqt的synthesis&generate按键绑定函数单独写一个python文件应该就可以了，这样可能对平时习惯用云端环境的人更友好一些。
更重要的是，这样就不用在自己的电脑上配置半天环境了。写个简单的colab就可以更好地普及啦。（不介意的话我可以负责来写www）

Start by locating the PyQt synthesis and generate button-binding functions mentioned in the issue and trace how they produce audio. Define the new inference.py entry point around those existing operations, accepting the requested directories or filenames and working without the GUI; completion should demonstrate generated audio from the specified inputs, with a Colab example only if included in scope.

Written by the indexing model from the issue text.