Using Own Database - Githubissues

Rayhane-mamah / Tacotron-2

DeepMind's Tacotron-2 Tensorflow implementation

MIT License

2.28k stars 905 forks source link

Using Own Database #420

Open letmeetran opened 5 years ago

letmeetran commented 5 years ago

Hello,

From my understanding, we need to write a build_from_path in wavenet_preprocessor.py in order for Tacotron 2 to accept our own database to train WaveNet. I currently have my database and the .csv file but I am struggling to write the build_from_path. Naturally, I tried copying and pasting the build_from_path function in preprocessor.py but this does not work. My .csv file is set up exactly like the LJ Speech .csv (Wavfile|Words).

Thank you @Rayhane-mamah for this amazing repo!

letmeetran commented 5 years ago

When I try to copy and paste the preprocessor.py code, I keep getting a directory error. In order to solve this, I created a new directory which satisfies the previous error but I still get a different directory error: home/Tacotron-2/a/mjdata/metadata.csv home/Tacotron-2/d/mjdata/metadata.csv