LOCALIZATION REQUEST: Ossetic [os] about common-voice HOT 30 CLOSED

Hi, I am also a native speaker of Ossetian. I want to participate in the project. My Pontoon username is arturgudiev

from common-voice.

So far Ossetian only has 7 sentences contributed and approved. It will need 750 to launch.

from common-voice.

I've added Ossetian to Common Voice on Pontoon. If you could log in and suggest a translation, I will add you all as translators. Any questions, please feel free to contact us on Matrix.

from common-voice.

I've added @amikeco as a translator, if the other translators could also log in and make suggestions I will add them too. Also, there should be a manager for the locale, ideally it should be a native speaker.

In addition a next step will be to collect 750 public domain sentences in Ossetian for contributors to read out. I'm happy to provide advice on doing this. Please feel free to get in contact with me on Matrix.

from common-voice.

Dear @amikeco, thank you for your interest in Common Voice. Are you a native speaker of Ossetian or are you in contact with any members of the Ossetian-speaking community?

from common-voice.

Dear @amikeco, thank you for your interest in Common Voice. Are you a native speaker of Ossetian or are you in contact with any members of the Ossetian-speaking community?

I am actually not a native speaker of the language, but I grew up in Ossetia, and I am almost fluent and have experience in translating interfaces (VK, telegram, Wikipedia) and other activities for the language (e. g. presenting it at language festivals, and keeping a popular site about the language).

Yes, I am in contact with the community, I think I can reach thousands of views with my announcements. I also can contact personally a dozen of speakers who may be interested in the activity like translating the interface, collecting the texts and reading them for the Common Voice initiative.

from common-voice.

That sounds fine! It would be good to have speakers involved in the translation. Could you invite them to make Pontoon accounts and post them here? Thanks!

from common-voice.

That sounds fine! It would be good to have speakers involved in the translation. Could you invite them to make Pontoon accounts and post them here? Thanks!

All right, going to do that.

from common-voice.

I am a native speaker of Ossetian ( or Ossetic), I would like to take part in the project. My username at Pontoon is "chibupel".

from common-voice.

Hi! Could you please add ossetian language. I am a native speaker of ossetic too. My Pontoon username is geor_lolaev. Right now I have free time, so I can help in translating

from common-voice.

I am a native speaker of Ossetian ( or Ossetic), I would like to take part in the project. My username at Pontoon is "chibupel".

Hi! Could you please add ossetian language. I am a native speaker of ossetic too. My Pontoon username is geor_lolaev. Right now I have free time, so I can help in translating

Thank you for your support for the idea.

from common-voice.

Thanks @ArturGudiev, could you please make a few suggestions and I will add you as a translator. :)

from common-voice.

@ftyers What do you mean by suggestions? Should I translate something? Is it on Pontoon website?

from common-voice.

@ftyers I got it. I've suggested several translations on Pontoon

from common-voice.

@ArturGudiev Салам. Tell me your telegram please

from common-voice.

@theJosephGoodman Салам. It's arturgudiev

from common-voice.

There is an important correction: when we speak about the Ossetic/Ossetian version, we mean its bigger dialect known as Iron [ee-rohn]. The other dialect is known as Digoron, it has its own literary tradition, press, and even some people calling it a separate language, it may appear later with its own activists. Collecting both under one name seems not possible, since the differences are deeper than in reading same words differently.

I see in the list there are already many languages with narrowing descriptions, so it'd be nice to write ours as "Ossetic (Iron)", "Ossetian (Iron)", "Iron Ossetian" or in any other way that fits the project's tradition.

from common-voice.

It is the case for many languages on Common Voice (like Arabic, Portuguese, Armenian, Swahili, etc.) that have very different variants. Common Voice so far goes by ISO codes to distinguish languages that we support. We currently support variants for individual speakers (you can read about it here and we have a plan to support variants for text sentences too.

from common-voice.

It is the case for many languages on Common Voice (like Arabic, Portuguese, Armenian, Swahili, etc.) that have very different variants. Common Voice so far goes by ISO codes to distinguish languages that we support.

We have other levels of variation to call "variants", "subdialects" that are united by the writing system (there will be at least 4 in Iron Ossetic).

It's up to you how to present it so far: it could be code OS and name Ossetic, and when the Digor dialect (and literary norm) topic appears with the people ready to collect sentences and voices in it, we can come back to the discussion again.

from common-voice.

75% localization at Pontoon is ready.

from common-voice.

It is the case for many languages on Common Voice (like Arabic, Portuguese, Armenian, Swahili, etc.) that have very different variants. Common Voice so far goes by ISO codes to distinguish languages that we support. We currently support variants for individual speakers (you can read about it here and we have a plan to support variants for text sentences too.

I see 2 Armenian languages in the lists (Armenian and Armenian Western). So Armenian Western is a different variant with its own dataset. The same would be suitable for Digoron Ossetian. It has its own grammar and vocabulary. Although there are similarities with Iron Ossetian the differences are significant. (by the way, Digoron Ossetian is also sometimes called Western Ossetian).

from common-voice.

There are not two variants of Armenian deployed in Common Voice, as you can read in this blogpost and see in the datasets. In addition, any region specific language codes are a result of legacy inclusions. We are planning to merge them when the technical capacity exists.

from common-voice.

https://commonvoice.mozilla.org/os/ is running now, thank you very much.
We will start new topics for other issues about it, right?

from common-voice.

There is no button to contribute voice on os site. What is the reason? Should we add some os directory for that?

from common-voice.

Dear Francis,

Can we have an additional line of code somewhere in the Ossetic product? As Cyrillic Æ is poorly supported in fonts, people use the Latin code for it, Æ (absolutely identical visually, but other letter from the point of view of computers).
So either in the sentence collector or while packing the dataset, all Latin codes for Ææ should be changed to Cyrillic ones. Will it be possible?

Also the Main page of Common Voice uses some font that does not support Cyrillic letters, it is very obvious because of how Ossetic Æ's appear among the adjacent other letters:
https://commonvoice.mozilla.org/os
https://commonvoice.mozilla.org/os/review

from common-voice.

There are not two variants of Armenian deployed in Common Voice, as you can read in this blogpost and see in the datasets. In addition, any region specific language codes are a result of legacy inclusions. We are planning to merge them when the technical capacity exists.

There are essential differences in grammar, vocabulary, and pronunciation between the Iron and Digor languages. One object can have different words in Iron and Digor, correspondingly. You either need to provide 2 variants for every phrase (Iron and Digor) or create two separate datasets.

In Ossetia, books are printed in either Iron or Digor. Foreign books can have two translations (Iron and Digor). I still think it's better to create two different interfaces for the Iron and Digor languages. Otherwise, datasets will be mixed, which can result in lower quality speech recognition.

from common-voice.

Why do you think it would result in lower quality speech recognition? If ASR systems can be improved by adding data from unrelated languages, why not from related languages or variants of the same language? Anyway, in any case, this GitHub issue is not the place for this discussion, it would be better to take it to Discourse, and please feel free to link it here.

from common-voice.

Why do you think it would result in lower quality speech recognition? If ASR systems can be improved by adding data from unrelated languages, why not from related languages or variants of the same language? Anyway, in any case, this GitHub issue is not the place for this discussion, it would be better to take it to Discourse, and please feel free to link it here.

Sure, thanks. We can continue the discussion there.

from common-voice.

So far Ossetian only has 7 sentences contributed and approved. It will need 750 to launch.

Now there are 757 sentences in Ossetian. Should the voice recording option appear automatically? I don't see it right now.

from common-voice.

It should appear at the next release, so within 1-2 weeks.

from common-voice.