About Producing Manifest File

[nonmetal]

Hello, thank you for providing a wonderful repository. I would like to ask a question about how to obtain information in order to produce manifest file.

{
    "LJ001-0051.wav": {
      "text": "and paying great attention to the press work or actual process of printing,",
      "raw-text": "and paying great attention to the press work or actual process of printing,",
      "duration": 4.860090702947846,
      "phoneme": "æ|n|d|_|p|eɪ|ɪ|ŋ|_|ɡ|ɹ|eɪ|t|_|ɐ|t|ɛ|n|ʃ|ə|n|_|t|ə|_|ð|ə|_|\"|p|ɹ|ɛ|s|_|w|ɜː|k|\"|_|ɔː|ɹ|_|æ|k|tʃ|uː|əl|_|p|ɹ|ɑː|s|ɛ|s|_|ʌ|v|_|p|ɹ|ɪ|n|t|ɪ|ŋ|,"
    },
    "LJ001-0120.wav": {
    ...
    },
    ...
}

In the format of manifest file above, although there is a transcribed data of the audio file, getting duration and phoneme from the following data should need extra pre-processing, and the method nor code to create these information doesn't seem to presented in the repository.

Thus, I would like to ask: what libraries did you use for producing those information? Also, if you used codes to produce the manifest file in your experiments, I would like to carefully ask if you could provide the code.

Thank you.

[taras-sereda]

Hi @nonmetal thank you for the interest in our work!

We added necessary scripts in the repository to generate manifests for both model training and evaluation. Please consult the updated readme section