about Training models about discrete_intonation HOT 1 CLOSED

This makes the experiment run on the provided test data. Morgana allows you to specify different steps for validation and testing using analysis_for_XXX_batch: e.g. calculate loss in validation, and run synthesis in testing.

There is also --train and --valid but these are True by default. Use --no-train, --no-valid, or --no-test if you want to ensure that stage is turned off.

You can always run python f0_phrase_VAMP.py --help for the model files in this repo, this will provide a description of all the arguments. This info is also available in Morgana's documentation, here is the link for the ExperimentBuilder arguments

Documentation for test: "If True, generation for --test_id_list will be performed after training"

from discrete_intonation.