Lingala-TTS-Dataset

Language

Lingala is a Bantu language spoken primarily in the Democratic Republic of the Congo (DRC) and the Republic of the Congo (Congo-Brazzaville). It functions as a major language of wider communication across the northern and western DRC, including the capital Kinshasa, and across the northern half of Congo-Brazzaville, as well as among Congolese diaspora communities worldwide. Lingala has approximately 15 million native speakers, with an additional 10 million using it regularly as a lingua franca, predominantly in rural areas and the diaspora. Lingala emerged from Bobangi, a Bantu language historically used as a trade lingua franca along the Congo River in the precolonial period. Through contact with diverse linguistic communities and the influence of colonial and missionary intervention, Bobangi underwent extensive restructuring. Beginning in the early twentieth century, Catholic missionaries of the Congregatio Immaculati Cordis Mariae (the Scheutists) undertook a deliberate program of grammatical and lexical reform, drawing on indigenous languages including Iboko, Mabale, and original Bobangi to produce a normalized written form. This reformed variety was disseminated through mission schools, religious texts, and print media across the western Congo. The name Lingala itself was coined as part of this reform process, replacing the earlier glossonym Bangala. While the legitimacy of Lingala has been debated — with some characterizing it as an artificially constructed variety and others as a natural contact-induced koine — it is broadly acknowledged that its current geographical spread and sociolinguistic functions are substantially shaped by these missionary and colonial-era interventions.

Variants

Two broad varieties of Lingala are widely recognized. The first is the reformed or standard variety associated with the Scheutist missionary Egide De Boeck, whose 1904 grammar established a normalized form with restored Bantu nominal prefix systems and expanded vocabulary. This variety, sometimes referred to as book Lingala, school Lingala, or written Lingala, became the medium of formal education, Catholic liturgy, and missionary publications across much of the western Congo and gained acceptance among some Protestant missions as well. It remains in use today in formal registers, religious contexts, and media. The second is the spoken urban variety, particularly associated with Kinshasa (formerly Leopoldville). This variety retains features of the pre-reform contact language, including an eroded system of nominal prefixes and syntactic concordance, and was heavily influenced by Kikongo and the languages of diverse migrant communities. Residents of Leopoldville adopted the name Lingala for this variety as well, while distinguishing it informally from the missionary standard. This urban spoken Lingala nativized rapidly and became the dominant vernacular of the capital. A Kinshasa Bible translation in this variety was published in 2000, marking a tentative entry of spoken Lingala into domains previously reserved for the standard form. In the northeastern DRC, the older glossonym Bangala continues to be used alongside or instead of Lingala, reflecting the comparatively weaker penetration of the Scheutist reform in that region.

##Alphabet

Source

This dataset was created using self-audio recordings of one female and one male native speakers. The recordings were then transcribed the recordings. This task aimed to produce datasets suitable for developing text-to-speech models for the Lingala language. The speakers were guided through the process using open questions provided by the research coordinator.

Domain

The questions which prompted the speech recorded by the native speaker of Lingala covered a variety of domains relevant to the cultural practices of the indigenous community, and pertained to mostly the following genres: procedural, opinion and philosophical.

Size

Total size is 962,04 MB

Structure

• 8,572 audio clips totalling 4 hours, 25 minutes and 54 seconds;

• Five audio mapping files containing 8,572 lines. The dataset contains a visible stratum of French utterances and code-switched passages.

One recurring example of code-switching is the labelling of transcription slots that do not contain audible speech as "silence". Users of the dataset may wish to exclude these slots if they are using it for ASR or TTS tasks.

Sample

1. Lingala-TTS-Dataset_05_T709_T710.wav | Ko bunda pona ko zonguela abele na bango.

2. Lingala-TTS-Dataset_04_T211_T212.wav | Eko zala pondu oyo eko zala kaka bongo,

3. Lingala-TTS-Dataset_04_T327_T328.wav | Azala, asala na biso kaka nati asalaka na kala,

4. Lingala-TTS-Dataset_04_T105_T106.wav | Otambola malembe malembe kino oko koma na route nationale.

5. Lingala-TTS-Dataset_04_T191_T192.wav | Oko sokola ye, oko tika ye na pembeni, oko sala osaka na yo,

6. Lingala-TTS-Dataset_04_T97_T98.wav | Donc eko zala na bato ba kowuta Ollombo ba ko kende ko teka biloko na bango kuna après bazongi na bango.

7. Lingala-TTS-Dataset_05_T747_T748.wav | Biso kuna to zalaka boye, to vivre- to- to bikaka boye,

8. Lingala-TTS-Dataset_03_T1457_T1458.wav | ba todisaka yango na matiti ya ebele ya kitoko

9. Lingala-TTS-Dataset_04_T282_T283.wav | Eko sala été likama ekoki ko komela na ngonga nionso.

10. Lingala-TTS-Dataset_04_T610_T611.wav | Bo ko yaka côté bo moni bilembo, ba niama bazalaki wana,