Fastspeech2 develop #4220
Fastspeech2 develop #4220
Conversation
|
zhangjinghong seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
| # See the License for the specific language governing permissions and | ||
| # limitations under the License. | ||
|
|
||
| import paddle |
There was a problem hiding this comment.
need to change to lazy_paddle for all python file
| @@ -0,0 +1,21 @@ | |||
| # copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve. | |||
| model_file: Optional[os.PathLike]=None, | ||
| params_file: Optional[os.PathLike]=None, | ||
| device: str='cpu', | ||
| # for gpu |
There was a problem hiding this comment.
Are these parameters all useful and can been changed in config file?
| format(precision, device)) | ||
|
|
||
| # after paddle 3.0, support new inference interface | ||
| if paddlespeech.utils.satisfy_paddle_version('3.0.0-beta'): |
There was a problem hiding this comment.
paddlex infer will not install paddlespeech. Also paddlex will install paddle>=3.0, therefore version check is unnecessary.
| elif precision in {"fp16", "bf16"}: | ||
| config.enable_mkldnn_bfloat16() | ||
| print("MKLDNN with {}".format(precision)) | ||
| if use_trt: |
There was a problem hiding this comment.
this parameter should came from config file, whether it can be changed?
| am_input_names = am_predictor.get_input_names() | ||
| get_spk_id = False | ||
| get_tone_ids = False | ||
| if am_name == 'speedyspeech': |
There was a problem hiding this comment.
speedyspeech is unsupported in your code.
| """ | ||
| super().__init__(*args, **kwargs) | ||
| self.model = self._build() | ||
| self.model_dir = "/home/zhangjinghong/PaddleSpeech/examples/csmsc/tts3/fastspeech2_nosil_baker_ckpt_0.4/inference" |
There was a problem hiding this comment.
model dir should came from outside.
| Returns: | ||
| type: The PwganResult class. | ||
| """ | ||
| return PwganResult |
There was a problem hiding this comment.
PwganResult return an audio? BaseResult can't return audio format, maybe you should new one.
No description provided.