如何在v0.0.1版本下不使用參考樣本直接用模型做TTS

demo的功能都是錄一段額外的不同人聲並讓模型克隆新的聲音，但可不可以不使用參考樣本，直接調用預訓練的模型來生成語音呢。目前我是將gen_voice.py移到v0.0.1分支下執行，我想用類似的方式來使用訓練完成的模型，將我的聲音樣本訓練足夠後不透過克隆來使用，謝謝！

The request centers on gen_voice.py and the v0.0.1 branch. Read that entry point first to trace how the trained model is loaded and where reference audio is required; compare the available inference path with the requested no-reference workflow. Done criteria are not specified in the issue, so define the expected invocation and output before implementation.

Written by the indexing model from the issue text.