visilyromani / ttsmodelhelper Goto Github PK

TTSModelHelper is used to record audio training data for rhasspy Piper text to speach. Users can import a transcript for a model they create. Preview and listen to past audio they recorded and export all of the files into a compliant zip folder containing all of their training data

If you are interested in making changes fork and clone the repository. Make sure you have the prerequisites software to create a tauri application. Follow the quickstart guide from at Tauri website

Install the node requrements:

pnpm i

To launch the delopment enviroment with pnpm:

pnpm tauri dev

Currently this application does not support multiple speakers

Importing transcript data requires a json file in this spesific format:

[
    {"text":"This is an example transcript},
    {"text":"This is an example transcript2},
]

Metadata is structured with an id and transcript seperated by |

id|transcript

Exported files are in a zip format that follows the following structure

file.zip
│   metadata.csv
│
└───wav
    │   audiofile.wav
    │   audiofile2.wav