Where is the hifigan vocoder of TTA released?

[yuyun2000]

Here is the bug report:

  File "D:\github\Amphion\models\tta\ldm\audioldm_inference.py", line 42, in __init__
    self.build_vocoder()
  File "D:\github\Amphion\models\tta\ldm\audioldm_inference.py", line 68, in build_vocoder
    with open(config_file) as f:
FileNotFoundError: [Errno 2] No such file or directory: 'D:/github/Amphion/ckpts/tta/hifigan_checkpoints/config.json'

[HeCheng0625]

Hi, you can download the vocoder checkpoint from https://huggingface.co/amphion/text_to_audio/tree/main/tta/hifigan_checkpoints

[yuyun2000]

3q for your answer and rename my question

[yuyun2000]

不对劲啊，这个gan的采样率是16000，nummel=80,我用对应的mel特征去推理为什么得不到原始音频