Open salisbury-espinosa opened 4 years ago
Use np.log(f0.clip(min = minval)) where you can use an infinitesimally small value as a minval. About 1e-5 works well. If you still have any issues, let me know.
Voice_Converter_CycleGAN/preprocess.py:170: RuntimeWarning: divide by zero encountered in log f0_converted = np.exp((np.log(f0) - mean_log_src) / std_log_src * std_log_target + mean_log_target)