Take part
Maloba is you
The corpus is built line by line, by speakers, annotators, validators and translators. Every contribution is traced, and every one counts. Here is how to join the work.
Lend your voice
A few minutes of Lingala speech on an everyday theme: family, trade, travel, health. All accents, all backgrounds. Recordings take place in group sessions, the Voice-a-thons, or at a partner organisation.
How consent is collected
On site, before recording, in the ODK Collect collection app. It covers the recording, its transcription and its publication in a corpus under the NOODL-1.0 licence, with no name and no contact details. It is kept with the recording. You may withdraw it at any time by writing to contact@congo-digital.com.
What the first Voice-a-thon produced
At the Grandes Endémies centre in Brazzaville, 10 people took part and 11 recordings were gathered. These recordings are a resource gathered outside the annotated corpus: no duration of speech is published.
Collections at closure
The Impfondo and Talangaï collections were not carried out.
Register as a voice donor
Annotate and review
Profile sought: a Lingala speaker at ease in writing, patient, precise. French is needed for the instructions and for translation.
Training: Lingala annotation guide and writing conventions (ɔ, ɛ, tones, French words), with corrected exercises.
Tools: Label Studio for transcription, Argilla for validation. Every annotation is reviewed; validators judge samples against three questions fixed in advance.
Translating
Our translation model learned more from 288 sentences translated by Lingala speakers than from 4,700 machine-produced sentences. Every translated sentence counts.
Join the language team
Open a site
Schools, health centres, radio stations, administrations, cooperatives: what the project lacks most, more than technical means, is access to places where the language is spoken. One morning at your organisation is enough to record dozens of voices.
The team comes with the equipment and the consent forms; you provide a room and let the participants know. The project lead replies within a week.
Offer a venue
Contributors
No name is published: the site names neither speakers, nor annotators, nor validators, nor translators, for want of written agreement to publish them. In the reports, validators are designated by a code.
Join the conversation
There is no messaging channel and no newsletter.
To write to the team: Contact page.