hm_learner

textGrid files are not exported to output path and MFCCs generating seems failed

1

hi,i use the mfa of 2.0.0.0rc5 version(windows). if i run the command such as: mfa align test_data mandarin_pinyin.dict mandarin.zip test_result there are some files generated in ~\Documents\MFA.TextGrids were exported to...

model cantnot fit to data, and test voice is too bad when i use the paper configuration

3

i removed the postnet(remove the code of model and loss about postnet ) and set the pitch_quantization="log",set features of pitch and enery = "frame_level", normalization="False",and other configuration is the same...

请问TTS的语音克隆效果如何？

2

我知道微调确实可以合成目标人相似的语音，但是对于只有一句目标说话人的情形，就只能使用融合了说话人编码器的模型，对于训练集中没有出现过的目标人的音频克隆效果好像不是很好，相似度很低，甚至出现提供同一个说话人的不同音频所克隆的音色不相同的情况，还出现提供女声克隆出男声的情况，请问你们训练的模型对于这种zero-shot场景的域外说话人的克隆效果如何？