Sample Fe’fe’-TTS-Dataset

Language

Fe'fe' (also written Ghomala') is a Grassfields Bantu language belonging to the Niger-Congo phylum, classified within the Mbam-Nkam (Eastern Grassfields) branch. It is spoken primarily in the Haut Nkam Division of the Western Region of Cameroon. According to Ethnologue (https://www.ethnologue.com/language/fmp/), Fe'fe' has a substantial speaker community in the Haut Nkam area. Despite its sociolinguistic significance within Cameroon, Fe'fe' remains substantially underrepresented in language technology resources.

Variants

According to the Administrative Atlas of Cameroon's Languages (Breton & Bikia Fohtung 1991), Fe'fe' comprises two geographically distributed dialectal areas:

• North Fe'fe' dialectal area, comprising four dialect varieties:

  • La'fi dialect (Balafi)

  • Tuŋi dialect (Foutouni)

  • Nkwet dialect (Fondjomekwet)

  • Ntii dialect (Fondanti)

• Central Fe'fe' dialectal area, comprising four dialect varieties:

  • Njəə-Poantu dialect (Bandja-Babountou)

  • Fa' dialect (Bafang)

  • Nka' dialect (Banka)

  • Nee dialect (Bana)

Writing System

The writing system used for the transcription of Fe'fe' in this dataset is the General Alphabet of Cameroon's Languages (AGLC), as adopted by the Ministry of Basic Education of Cameroon and regularly updated by the Direction de la Promotion des Langues Nationales. The AGLC provides a phonologically motivated orthographic standard for Cameroonian national languages and serves as the reference framework for Fe'fe' literacy materials.

1. Vowels

The vowel system attested in the dataset includes the following oral vowels:

a, e, i, o, u, α, ə, ʉ, ɔ

Where:

• α (Latin alpha / IPA ɑ): central to back unrounded vowel, distinct from a

• ə (schwa): mid-central vowel

• ʉ (barred u): high central rounded vowel

• ɔ (open-o): open-mid back rounded vowel

Long vowels are represented by vowel doubling (e.g., aa, αα, oo).

2. Consonants

The consonant inventory reflected in the dataset includes simple, prenasalised, and digraph consonants:

b, c, d, f, g, h, j, k, l, m, n, p, s, t, v, w, y, z, ŋ

Digraphs and trigraphs: gh, ph, sh, th, zh, mb, mf, nd, ng, nk, nc, nsh, nzh

Special symbols:

• ŋ (eng): velar nasal consonant

• ʼ (modifier letter apostrophe) and ꞌ (saltillo): glottal stop / glottal closure marker

3. Tone system

Fe'fe' is a tonal language with multiple contrastive pitch levels and contour tones. The dataset employs systematic tone marking on vowels in accordance with the AGLC convention. The following diacritics are attested in the dataset:

Level tones:

• High tone (H): acute accent — á, é, í, ó, ú, ά, ə́, ʉ́, ɔ́

• Mid tone (M): macron — ā, ē, ī, ō, ū, ᾱ, ə̄, ʉ̄, ɔ̄

• Low tone (L): grave accent — à, è, ì, ò, ù, ὰ, ə̀, ʉ̀

Contour tones:

• Rising tone (LH): caron — ǎ, ě, ǐ, ǒ, ǔ, α̌, ə̌, ʉ̌

• Falling tone (HL): circumflex — â, ê, î, ô, û

• Mid-rising tone (MH): macron-acute (᷇) — vowel᷇

• Mid-falling tone (ML): macron-grave (᷆) — vowel᷆

Additional diacritics attested in the dataset include the combining dot below (̣), combining tilde below (̰), and combining diaeresis (̈), reflecting fine-grained phonological distinctions in the Fe'fe' sound system.

Source

The dataset was compiled from scripted speech prompt lists read by native speakers of Fe'fe' in recording sessions held at the École Normale Supérieure de Yaoundé in June 2026, in the framework of the Mozilla Data Collective project. Sentences were selected to provide broad phonological coverage of Fe'fe' and were transcribed in accordance with the AGLC orthographic standard.

Domain

The dataset represents scripted speech in Fe'fe', covering a broad range of everyday sentence types drawn from a general-purpose TTS/ASR prompt list. All utterances are scripted rather than spontaneous.

Size

Total audio duration: 4,291 seconds (01h 11m 31s), distributed across 1,004 MP3 audio clips in 11 recording sessions.

Structure

The dataset is organised into 11 recording sessions:

• Session tts_dataset_fmp_01: 100 clips (07m 23s)

• Session tts_dataset_fmp_02: 100 clips (06m 40s)

• Session tts_dataset_fmp_03: 100 clips (06m 31s)

• Session tts_dataset_fmp_04: 100 clips (06m 21s)

• Session tts_dataset_fmp_05: 100 clips (06m 46s)

• Session tts_dataset_fmp_06: 100 clips (06m 16s)

• Session tts_dataset_fmp_07: 100 clips (06m 52s)

• Session tts_dataset_fmp_08: 100 clips (07m 45s)

• Session tts_dataset_fmp_09: 100 clips (08m 11s)

• Session tts_dataset_fmp_10: 100 clips (08m 21s)

• Session tts_dataset_fmp_11: 4 clips (00m 20s)

Each session folder contains:

1. MP3 audio clips (100 per session, except session fmp_11 which has 4 clips)

2. One per-session sentence-to-audio mapping file (mapping.tsv), with 4 columns

Description of columns (mapping.tsv)

• #audio_filename: filename of the audio clip (MP3)

• #key: unique hash identifier of the recording

• #sentence: sentence text as read by the speaker, transcribed in AGLC orthography

• #attempts: number of recording attempts before acceptance

Sample