When infer() with multiple text, all audio are the same file size (当使用多个文本进行 infer() 时，所有音频的文件大小相同_

[ericfarng]

When calling infer() with multiple text, even if the text are very different length, then all the audio files are the same length
I am on Windows, using the demo code provided.

google translate:
当使用多个文本调用 infer() 时，即使文本的长度差别很大，所有音频文件的长度也相同
我在 Windows 上使用提供的演示代码。

[fumiama]

The Vocos will decode a batch of audio at the same time, who needs a matrix input, which means that the input length of all audio should match the longest sentence in infer array. User needs to remove the trailing zero values of their output by themself. The program cannot remove it because the stream mode also needs a same output length.