Just realized there’s a parameter “–train_cached_features_path”. I assume this is to reuse the cached audio feature. But I failed to find the parameter to store the cache?
Could anybody show an example how to store and reuse the cache? Thanks!
Generation of the cache happens automagically: Just set this parameter to a (non-existing) file (in an existing directory). At the beginning of the training it will check for the file and - if not there - generate it for you. Next time it will just load it.
1 Like