TTS Training corpus.
Use different folders for different typed text.
Working with bhashini.ai to create sanskrit tts, with the understanding that the resultant voices would be available for non-commercial use by us.
Prose and anuShTubh shloka training data (typed text which artists can read out) is easy and plentiful. What should be the most important other chandases to cover? Could you please help collect a corpus (defined above, bigger the better) for each of these meters, so that excellent voices can be trained?