search

TimoBolkart / voca

This codebase demonstrates how to synthesize realistic 3D character animations given an arbitrary speech signal and a static character mesh.
https://voca.is.tue.mpg.de/en
1.17k stars 273 forks source link

Missing files from training data #101

Open mayerj711 opened 2 years ago

mayerj711 commented 2 years ago

Hi there,

I want to experiment and retrain voca. I downloaded the training data from the vocaset link mentioned. However, the following files were missing, and there is no script or format mentioned on how to generate them.

  1. init_expression_basis.npy
  2. raw_audio_fixed.pkl
  3. subj_seq_to_idx.pkl
  4. templates.pkl

Can you please provide a download link to these files, or a script to generate them? Even the correct format of these dictionaries would help me to write a script accordingly.

Thankyou!

TimoBolkart commented 2 years ago

Hi, I just downloaded the trainingdata.zip from the VOCA website and unpacked it. It seems to contain all files that you are missing (See screenshot of the folder below) image

akarshmishra001 commented 2 years ago

How can I make npy and pkl training data for my own dataset?

mhxlearn commented 2 years ago

Hi, I just downloaded the trainingdata.zip from the VOCA website and unpacked it. It seems to contain all files that you are missing (See screenshot of the folder below) image

How can I make npy and pkl training data for my own dataset?