Kab team program about Mozilla's Projects: Session 01 Common Voice and Sentence collector

Common Voice
#1

A live training about Common Voice and Sentence collector was organized by the kab community with FB live today 21h00 and 22h00 Algerian Time.

The video will be shared later on Youtube.

We talked about the Dataset and licence. Motivation of the Mozilla’s project. Voice and sentence collections.

The Kab team has launched an alphabetization program about Mozilla’s projects: Localization, Bug report, beta tests…

Later if we could have a tech support, we can also talk about the tech aspects of Mozilla’s projetcs: TTS, DS, MDN, Firefox…
The Kab team is also involved in other open projects and open linguistics datasets.

2019-06-06%2022_49_07-(3)%20Sideg%20-%20Ifecka%20um%E1%B8%8Dinen%20-%20Asil%C9%A3%20%C9%A3ef%20Common%20voice%20d%20sentence%20collector%20-%20SRID_
2019-06-06 22_49_07-(3) Sideg - Ifecka umḍinen - Asilɣ ɣef Common voice d sentence collector - SRID_.jpg964×583 64 KB
2019-06-06%2022_43_59-(3)%20Sideg%20-%20Ifecka%20um%E1%B8%8Dinen%20-%20Asil%C9%A3%20%C9%A3ef%20Common%20voice%20d%20sentence%20collector%20-%20SRID_
2019-06-06 22_43_59-(3) Sideg - Ifecka umḍinen - Asilɣ ɣef Common voice d sentence collector - SRID_.jpg965×591 70.9 KB

3 Likes
#2

The kab team is impressive indeed, how do you manage to motivate people for voice donation and sentence collection, I will be soon in that stage with Abkhazian.

Also what other open projects and open linguistics datasets are you involved in? that might interest me for Abkhazian as well. I’m in particular interested in machine translation if you came across that by any chance.

Nart.

1 Like
#3

Hi @daniel.abzakh,

We were organizing training sessions about voice recognition and the impact of the voice assitants on our lives in the near future. So, we had to convince people to donate their voice. We have an FB group we use to recruit more volunteers. Recently, we are organizing live training sessions.
Almost we have one event per week (onsite or a live event).

Kabyle is also involved on other linguistics corpora. We invested Tatoeba, a corpus for sentences, translations but also voice recordings.

I’m using those tranlsations on Tatoeba to work on english/kabyle and french/kabyle for MT. I’m using OpenNMT to generate and test models.

I hope that Mozilla will launch such a corpora and a Machine learning based tool to work on MT. Like, voice, Machine translation is also a challenge for the next years. Free and opensource voice and Translation are the next challenge that Mozilla and it’s community should face.

3 Likes
#4

@belkacem77 Great information. Thanks for sharing.

2 Likes