TTS_example.ipynb

tts_example.ipynb

I have tried the above command line but it does not work either. I think it is not the file path problem, but more like some argument parsing problem, that the audio.stats_path do not parsing correctly, because in the code the attribute "config.audio.stats_path" is "scale_stats.npy" not the "/content/TTS/scale_stats.npy ".

Sorrry, I missed that. You can double check if this "/content/TTS/scale_stats.npy exists or not if if doesn't. Don't use config.audio.stats_path aurguments.
@gbvssd

The file scale_stats.npy exists and the file path is correct. Is the -- coqpit.addio.stats_path argument set the state file path? And I track the error to the "utils/io.py" file and test the value of "config.audio.stats_path", I assume it should be "/content/TTS/scale_stats.npy " but it is "scale_stats.npy".
@Sadam1195

The file scale_stats.npy exists and the file path is correct. Is the -- coqpit.addio.stats_path argument set the state file path? And I track the error to the "utils/io.py" file and test the value of "config.audio.stats_path", I assume it should be "/content/TTS/scale_stats.npy " but it is "scale_stats.npy".

then may be you are not using the right version of repo. Let's get in touch on discussion or gitter as this conversation will spam the gist.
@gbvssd

Hello guys , I have some errors can anyone guide me to solve this
First Try:

Second Try:

Third Try :

What should I do Now ?

Thanks in Advance

all the community members

Anyone figure this out yet?

Anyone figure this out yet?

I encountered the same issue, instead of !git clone https://github.com/coqui-ai/TTS do this !git clone https://github.com/mozilla/TTS although you can make the json file, you still gonna have another problem.

Updated the notebook based on the latest version of 🐸TTS

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

I am receiving such an error:

  File "train.py", line 13, in <module>
    from TTS.datasets.TTSDataset import MyDataset
ModuleNotFoundError: No module named 'TTS'

Where I might be doing a mistake?

I am in TTS project directory, LJSpeech is in subfile in that directory

run it again it will be ok

training worked fine after replacing the name= argument to BaseDatasetConfig with formatter=:

--- a/tts_example.ipynb
+++ b/tts_example_new.ipynb
@@ -139,7 +139,7 @@
         "\n",
         "# init configs\n",
         "dataset_config = BaseDatasetConfig(\n",
-        "    name=\"ljspeech\", meta_file_train=\"metadata.csv\", path=os.path.join(output_path, \"/content/LJSpeech-1.1\")\n",
+        "    formatter=\"ljspeech\", meta_file_train=\"metadata.csv\", path=os.path.join(output_path, \"/content/LJSpeech-1.1\")\n",
         ")\n",
         "\n",
         "audio_config = BaseAudioConfig(\n",

Seems like this notebook is creating train-val splits in metadata_train.csv and metadata_val.csv, but then they're not using in the dataset config. Shouldn't it be

dataset_config = BaseDatasetConfig(
    formatter ="ljspeech", meta_file_train="metadata_train.csv", meta_file_val="metadata_val.csv", path=os.path.join(output_path, "/content/LJSpeech-1.1")
)

When I try to install -e, it gives me an error:

import os
os.chdir('/content/coqui-TTS')
!pip install -e .[all]

I got those error:
Failed to build TTS
ERROR: Could not build wheels for TTS, which is required to install pyproject.toml-based projects

Can anyone help?

@ZeLiu1 The error "Failed to build TTS" and "ERROR: Could not build wheels for TTS" during the installation with -e (editable) option indicates missing dependencies or build requirements for the TTS package. Ensure all dependencies are installed, check Python version compatibility, and have necessary build tools. Consider using a virtual environment or building from the source repository to resolve the issue.

Hi all,

I have launched the google colab: TTS_example.ipynb
But when Train Tacotron DCA starts i get the follow issue:

---

TypeError Traceback (most recent call last)
in <cell line: 18>()
16
17 # init configs
---> 18 dataset_config = BaseDatasetConfig(
19 name="ljspeech", meta_file_train="metadata.csv", path=os.path.join(output_path, "/content/LJSpeech-1.1")
20 )

TypeError: BaseDatasetConfig.init() got an unexpected keyword argument 'name'

try installing Visual Studio Build Tools 2022 Microsoft C++ Build Tools and make sure you check the box for the "Desktop development with C++" Install option