[Bug] Error while loading model from path and save tts to file.

[kishcs]

Describe the bug

Error:

Traceback (most recent call last):
  File "/data/tts.py", line 42, in <module>
    tts.tts_to_file(text=text, file_path="sample.wav")
  File "/usr/local/lib/python3.10/dist-packages/TTS/api.py", line 332, in tts_to_file
    self._check_arguments(speaker=speaker, language=language, speaker_wav=speaker_wav, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/TTS/api.py", line 228, in _check_arguments
    if self.is_multi_lingual and language is None:
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1940, in __getattr__
    raise AttributeError(
AttributeError: 'TTS' object has no attribute 'is_multi_lingual'

When I am trying to load model from model path and converting tts to file.

tts = TTS(
    model_path="/data/models/tts_models--en--ljspeech--tacotron2-DDC/model_file.pth",
    config_path="/data/models/tts_models--en--ljspeech--tacotron2-DDC/config.json",
    vocoder_path="/data/models/vocoder_models--en--ljspeech--hifigan_v2/model_file.pth",
    vocoder_config_path="/data/models/vocoder_models--en--ljspeech--hifigan_v2/config.json",
    # vocoder_name="vocoder_models/en/ljspeech/hifigan_v2",
    progress_bar=True,
    gpu=True  # Set to True if running on GPU
)

tts.tts_to_file(text="A long sample text to generate wav from text", file_path="sample.wav")

To Reproduce

from TTS.api import TTS
tts = TTS(
    model_path="/data/models/tts_models--en--ljspeech--tacotron2-DDC/model_file.pth",
    config_path="/data/models/tts_models--en--ljspeech--tacotron2-DDC/config.json",
    vocoder_path="/data/models/vocoder_models--en--ljspeech--hifigan_v2/model_file.pth",
    vocoder_config_path="/data/models/vocoder_models--en--ljspeech--hifigan_v2/config.json",
    # vocoder_name="vocoder_models/en/ljspeech/hifigan_v2",
    progress_bar=True,
    gpu=True  # Set to True if running on GPU
)

tts.tts_to_file(text="A long sample text to generate wav from text", file_path="sample.wav")

Expected behavior

It should run similar to when loading model from model_name.

Logs

Traceback (most recent call last):
  File "/data/tts.py", line 42, in <module>
    tts.tts_to_file(text=text, file_path="sample.wav")
  File "/usr/local/lib/python3.10/dist-packages/TTS/api.py", line 332, in tts_to_file
    self._check_arguments(speaker=speaker, language=language, speaker_wav=speaker_wav, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/TTS/api.py", line 228, in _check_arguments
    if self.is_multi_lingual and language is None:
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1940, in __getattr__
    raise AttributeError(
AttributeError: 'TTS' object has no attribute 'is_multi_lingual'

Environment

TTS Version: 0.22.0
Torch Version: 2.7.0
Python: 3.10.12
OS: Ubuntu
CUDA: 12.8
GPU: H100
Pytorch installed via pip

Additional context

No response

[eginhard]

It works fine in our fork (available via pip install coqui-tts). This repo is not maintained anymore.