Unintentionally, I found that the voice converted from text to voice is a little flawed
Everyone knows that text-to-speech is more convenient, and often everyone uses tools for reading aloud to convert
But I do n’t know if it ’s my reading aloud female version problem or there is a bug in it. The converted voice has a bit of harsh sound at the end (sharp voice, it feels awkward)
It is not audible to play with a conventional player on the Windows PC, but under Linux, it is a bit obvious to use aplay
Also, using the voice transferred from Baidu, there is a "be de" voice in front, which is also very strange
At the beginning, it was suspected that the code was written incorrectly, for example, treating the file header as audio data. After checking, the code is OK, the reading position and length are normal
Later, I suspected that the problem of switching the player was changed to not switching the audio device. After the data is played, the problem still exists.
Therefore, he began to doubt the voice itself. But the problem is that Windows playback itself is no problem
Doubt is doubt, the experiment still has to be done
Downloaded a goldwave, edited the voice, changed the following data to mute data, and then saved
As soon as the result was played, the annoying abnormal sound was gone, which was nice.
Personal speculation, this is because the conversion software is free, it is estimated that some bugs were intentionally left. Fortunately, it can be repaired by other tools, of course, this operation is more troublesome