Support for mantras

[mikkokotila]

Right now, looks like Sanskritization is ignored entirely, so basically, mantras are not phonetized.

Looks like a small addition with big value (as otherwise manual work is still needed in the end).

How would we implement this?

[eroux]

So, the main issue for mantras is a systematic-enough description of the Tibetan pronunciation of mantras, and I have never seen that anywhere. If you find a good documentation we can follow I'm happy to implement it! I doubt coming up with a documentation is something that can be achieved easily, but we can imagine at least making a list of the most common words, this is the approach that other phonetics software have used. I'm just a bit reluctant to do that because I don't think this is a very satisfying solution... but if I'm happy to merge a PR that adds such a list

In case you're interested in converting to proper Sanskrit phonology (and not phonetics), the timing of your question is interesting because I'm starting to work on a Tibetan -> IAST converter on https://github.com/eroux/Tibetan-to-IAST. I'll check in some initial code today I hope, perhaps that will be a good solution to your problem?

[mikkokotila]

Tibetan -> IAST converter on https://github.com/eroux/Tibetan-to-IAST. I'll check in some initial code today I hope, perhaps that will be a good solution to your problem?

Excellent. What I would like to do is contribute to that package in form of simplified schema, which will basically follows what you are doing with IAST, but takes away all the diacritical markers and such. What would be the simplest way to get started with that?

[eroux]

Thanks! Well, removing the diacritics is a simple regex, I think if you want to use bophono for the mantras the regex could just be a postprocessing of Tibetan-to-IAST in bophono. Something that could be a bit more challenging and live in Tibetan-to-IAST is Sanskrit normalization.

Some usual prakrit -> canonical Sanskrit normalizations are rcc -> rc where c is a consonnant or its aspirate version, for instance dharmma -> dharma, sarbba -> sarba, etc. This kind of thing can be found quite commonly in Tibetan renderings of Sanskrit and, I suspect, should be normalized for a Western audience. There are a few others, I'll try to find a proper list.

Two other interesting normalizations could be:

• merging syllables into words, which is going to be very tricky
• b -> v (ex: sarba -> sarva), which will require some research and should be based on the previous step