Implement Syriac / Assyrian Support

[esoleyman]

Description

Adds Syriac / Assyrian support

Type of PR

• Feature implementation

Testing

• pytest test/test_format_syr.py
• pytest test/test_parse_syr.py

Notes

Syriac support was implemented in glibc 2.36 under the syr locale. This may be a stumbling block to get Syriac tested appropriately. I can vouch that this works on my system currently.

[krisgesling]

Hi Emil,

First, congratulations on what I know is a lot of work!

Right now, we're in the midst of releasing the first production Mark II devices. So we're pretty swamped, but will definitely make some time to review this in detail over the next few weeks.

I'll also have a think about what we can do around the need for the latest glibc. This might just be documentation so that users know what's required.

Beyond Lingua Franca, do you have any recommendations for STT and TTS options?

[esoleyman]

Hi Emil,

First, congratulations on what I know is a lot of work!

Right now, we're in the midst of releasing the first production Mark II devices. So we're pretty swamped, but will definitely make some time to review this in detail over the next few weeks.

I'll also have a think about what we can do around the need for the latest glibc. This might just be documentation so that users know what's required.

Beyond Lingua Franca, do you have any recommendations for STT and TTS options?

Given that we have low resources for our language, we don't have any STT and TTS options at present. We are working on creating a language corpora among other initiatives including updating Unicode CLDR and getting Syriac onto Mozilla Common Voice.

[krisgesling]

Yeah, there are a lot of barriers to getting started unfortunately but it sounds like you are ticking them off one by one!

For TTS, if you have or know of any openly licensed voice datasets that we might be able to use, then we could look at training a Mimic 3 model.

[esoleyman]

There are no voice datasets — open or closed. We are working to create that as well. It’s a large undertaking but once we create that then mimic 3 is on my list.

…

On Mon, Oct 3, 2022, at 20:41, Kris Gesling wrote: Yeah, there are a lot of barriers to getting started unfortunately but it sounds like you are ticking them off one by one! For TTS, if you have or know of any openly licensed voice datasets that we might be able to use, then we could look at training a Mimic 3 model. — Reply to this email directly, view it on GitHub <#239 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABSTXE642GCZXFGUERQ756TWBODMFANCNFSM6AAAAAAQZHV3QU>. You are receiving this because you authored the thread.Message ID: ***@***.***>