The default pre-trained model downloaded path and modification

[xiao11lam]

General Question

Hi, this is a question concerning path detection.

Like here:

paddlespeech tts --am fastspeech2_ljspeech --voc hifigan_ljspeech --lang en

It will download the fastspeech2_ljspeech and hifigan_ljspeechpre-trained model when it first runs.

According to my search in the other places, I only knew that once I installed the model, the path to the acoustic model may depend on the installation method and location. I need to check the documentation for the installation instructions or check the installation location to find the path. Still, after I tried those ways, I did not find where the pre-trained model's downloaded default path like for fastspeech2_ljspeech and hifigan_ljspeech here.

With my best thanks, is there could be a way that enables me to find it or even change the default saved model path? Any suggestion could be much appreciated to me.

[yt605155624]

The prtrained models will be downloaded in ${HOME}/.paddlespeech/models, please check

PaddleSpeech/paddlespeech/utils/env.py

Line 44 in 004a4d6

MODEL_HOME = _get_sub_home('models')