I have build HTS voice using HTS 2.2 version using 48KHz files. Following is the description of the data.
1. Number of sentences: 1000
2. Language: English
3. Length of the sentence: 5 to 20 words
4. Sampling frequency: 48KHz
Models are smoothly built and synthesis was done on the test sentences. But, during synthesis following error is thrown
x2x : warning: input data is over the range of type 'short'!
Spectrogram is showing dark frequency range from 0 to 24KHz (attached image shows the same).
Please let me know any mistake that might have done by me in the setup or any changes for the better quality.