DigitalPhonetics / IMS-Toucan

Controllable and fast Text-to-Speech for over 7000 languages!
Apache License 2.0
1.47k stars 166 forks source link

New issue after upgrading to new version #187

Closed uni-saurabh-vyas closed 3 months ago

uni-saurabh-vyas commented 3 months ago
  File "/mnt/efs/.../IMS-Toucan/ims_tts_src_jsonl_hindi.py", line 82, in <module>
    tts = ToucanTTSInterface(device="cuda" if torch.cuda.is_available() else "cpu", tts_model_path="Meta", language=lang)
  File "/mnt/efs/.../IMS-Toucan/InferenceInterfaces/ToucanTTSInterface.py", line 46, in __init__
    self.phone2mel = ToucanTTS(weights=checkpoint["model"], config=checkpoint["config"])
  File "/mnt/efs/.../IMS-Toucan/Modules/ToucanTTS/InferenceToucanTTS.py", line 61, in __init__
    cfm_filter_channels = config.cfm_filter_channels
AttributeError: 'DotWiz' object has no attribute 'cfm_filter_channels'
Flux9665 commented 3 months ago

it looks like you're using an old checkpoint with the new code. The old model checkpoints are no longer compatible.