fine-tuning text to speech

i want to fine-tune the model for Arabic common voice but its giving alot of errors related to pitch and non availability of train.txt file. how to generate this train.txt file. i already have train.tsv but is the format same and can i feed it to the model for training.

Start by locating the training entry point and compare the expected train.txt format with the existing train.tsv; then trace the reported pitch errors. Done means establishing whether train.tsv is accepted or what conversion is required for Arabic Common Voice fine-tuning.

Written by the indexing model from the issue text.