Skip to content

About

Easy access to speech data across 142 African languages for training TTS and ASR models.

Topics

Resources

Stars

16 stars

Watchers

0 watching

Forks

Repository files navigation

AfriSpeech Selector

AfriSpeech Selector gives you access to 1,938,892 utterances · ~7,854 hours · 308 African languages from 4 public corpora for training TTS and ASR models.

Select languages by recorded hours (strength) — a country-balanced top-N or a hand-picked set, sized the way you want — and get the audio + metadata in the format your training pipeline expects. You take it from there; the tool doesn't do any text normalisation or cleaning (that's your framework's job).

It's clean read speech with aligned transcripts, so the natural fit is TTS — export WAVs + a manifest for LJSpeech, Piper, VITS, or MeloTTS. It works for ASR too (load_from_disk / Parquet, or stream with stream_dataset(...)), which is handy for supplementing low-resource languages.

The strength table is merged from four source corpora (see Data sources): AfriSpeech, Google WaxalNLP, Meta OmniLingual ASR, and Open Bible Speech (African). Use --dataset to select into any one of them.

Redistribution: building a local working set for your own training is fine. Redistributing copies of the audio (e.g. --push to a public repo) is not recommended, given the permissions of the underlying public data sources.

Available languages

369 language configs · 1,938,892 clips · ~7,854 hours · 45 countries, drawn from 4 corpora. Hours is the strength signal used for ranking. The --languages value is the name you pass to pick a language (e.g. --languages twi_twi, ajima_ajm…). List them anytime with afrispeech-select --list-langs, or see the full language catalog at the bottom. To restrict the pool to one source corpus pass --dataset, e.g. afrispeech-select --list-langs --dataset open-bible.

Install

git clone https://github.com/AfriSpeech/afrispeech-selector.git
cd afrispeech-selector
python3 -m venv .venv && source .venv/bin/activate
pip install -e .            # gives you the `afrispeech-select` command

(Use python3 — the code uses non-ASCII text and won't run under Python 2.)

Quickstart — one language, ready to train

Most people just want one language for TTS. Pick its name (afrispeech-select --list-langs, or the catalog at the bottom) and run one command. Clips are filtered to a 3–15 s window by default, so the result is training-ready.

# ~5 hours of Twi as an LJSpeech TTS dataset (wavs/ + metadata.csv), 22.05 kHz
afrispeech-select --languages twi_twi --total-hours 5 --out data/twi --format ljspeech

# …for Piper / VITS / MeloTTS instead, just change --format:
afrispeech-select --languages twi_twi --total-hours 5 --out data/twi --format piper

(Using it for ASR? See Using it for ASR below.)

That's it — data/twi now holds the audio + metadata in the right layout (omit --out and it defaults to data/<language>). Want more or less? Change --total-hours (or use --per-language N for a clip count). Want longer/shorter clips? Set --min-clip-sec / --max-clip-sec (defaults 3 / 15).

Asking for more than a language has? You get everything available — the hour/clip target is an upper bound; the tool never pads or repeats. When your request can't be met, the CLI tells you how much actually exists and asks you to confirm before downloading, e.g.:

Only ~2.37 h is available vs ~20 h requested. Proceed with all available? [y/N]

Pass -y/--yes to skip the prompt (for scripts/notebooks), or --dry-run to see the achievable amount up front. (The 3–15 s filter trims a language's usable hours below its catalog total, so e.g. a 50 h language yields somewhat less.)

Export one language for a TTS framework (WAVs + manifest)

TTS data-prep reads WAVs + a manifest from disk. Pick your framework with --format: ljspeech (generic / Coqui), piper, vits, or melo. Audio is 16-bit mono WAV at --target-sr (default 22050); transcripts are written verbatim — no normalisation.

# LJSpeech layout (wavs/ + metadata.csv: id|text|text)
afrispeech-select --languages twi_twi --total-hours 5 --out data/twi --format ljspeech

# Piper (metadata.csv: id|speaker|text)
afrispeech-select --languages twi_twi --total-hours 5 --out data/twi --format piper

# VITS (filelist.txt + speakers.txt) and/or MeloTTS (metadata.list)
afrispeech-select --languages twi_twi --total-hours 5 --out data/twi --format vits,melo

Each writes <out>/wavs/*.wav plus the manifest. Phonemisation / text cleaning is left to the framework's own preprocessor.

No install? python3 -m afrispeech_selector … works the same from the repo.

Using it for ASR

Being clean read speech with transcripts, it's well suited to supplementing low-resource ASR (less so as a stand-alone set for spontaneous/noisy audio). Export an on-disk 🤗 dataset, or stream it straight into a Trainer with no copy:

afrispeech-select --languages twi_twi --total-hours 5 --out data/twi \
    --format disk,parquet --target-sr 16000
from datasets import load_from_disk
ds = load_from_disk("data/twi").train_test_split(test_size=0.1)

# …or stream (no local copy) — `afrispeech-select … --recipe` prints this:
from afrispeech_selector import stream_dataset
ds = stream_dataset(["twi_twi"], split="train", max_seconds=5 * 3600, target_sampling_rate=16000)

Use it in a Colab / Kaggle training notebook

Open In Colab Open in Kaggle

The example notebook is meant to be dropped into your own training notebook: its cells install the tool and pull your selected language(s) straight into the running session — write wavs/ + a manifest for a TTS framework, or stream into ASR training — so your training script has direct access to the speech data needed for training.

Selecting multiple languages

Everything above works for several languages too — the format/output flags are identical, you just choose which languages. --total-hours is split evenly across them (so each gets its fair share).

1. Name the languages you want:

# Ghanaian languages, 15 h total (5 h each), LJSpeech
afrispeech-select --languages twi_twi,ewe_ewe,ga_gaa --total-hours 15 \
    --out data/gh --format ljspeech

2. Or pick across the dataset by strength / balance — top-N by hours, with optional country balancing and pool filters:

# Top 10 languages, at most 2 per country, 24 h total
afrispeech-select --top 10 --max-per-country 2 --total-hours 24 \
    --out data/multi --format ljspeech

# Narrow the pool first: only languages >= 20 h, only Ghana & Nigeria
afrispeech-select --top 8 --min-hours 20 --countries GH,NG --total-hours 16 \
    --out data/wa --format ljspeech

Preview before pulling — list what matches, or dry-run a selection:

afrispeech-select --list-langs --min-hours 20 --countries GH,NG   # what matches
afrispeech-select --top 10 --max-per-country 2 --dry-run          # what it would pick

All options

flag meaning
--top N select the top-N languages by hours
--languages a,b,c hand-pick specific subsets instead
--max-per-country N cap languages per country (balance)
--no-proportional pure hours ranking, ignore country balance
--min-hours / --max-hours / --min-clips filter the language pool by strength
--countries GH,NG restrict to these countries
--dataset afrispeech,waxal,omnilingual,open-bible restrict to these source corpora (default: all). afrispeech-select --list-langs --dataset waxal shows them
--total-hours H total audio across the selection, split evenly per language
--per-language N max clips per language
--max-hours-per-lang H duration budget per language (decimals OK, e.g. 0.5)
--min-clip-sec / --max-clip-sec per-sample length window, default 3 / 15 s (out-of-range clips skipped; --min-clip-sec 0 --max-clip-sec 9999 to disable)
--split train|val|test|all which split to draw from
--target-sr HZ resample audio (e.g. 16000 for ASR, 22050 for TTS)
--schema asr|whisper|common_voice reshape columns for a training framework
--recipe print a stream_dataset(...) snippet for this selection (no copy)
--out PATH output directory / base name
--format … HF: disk,zip,parquet,csv · TTS: ljspeech,piper,vits,melo
--push REPO_ID [--public] [--token …] push to an HF dataset repo (creates a copy)
--dry-run / --list-langs preview the plan / list matching languages
-y / --yes skip the confirmation prompt when the request exceeds what's available

Capped pulls stream from the Hub and only transfer the samples you ask for. An uncapped "full build" downloads whole shards and must be enabled with --allow-full.

Downloads work anonymously but the Hub rate-limits them; for faster, higher-rate downloads set a free read token — export HF_TOKEN=hf_… (picked up automatically) or pass --token.

Output schema

column meaning
audio decoded waveform (HF Audio: array + sampling_rate)
text transcription
language language label
country ISO 3166-1 alpha-2 code
length clip duration in seconds
iso, subset ISO 639-3 code and source config (traceability)

Load a result later:

from datasets import load_from_disk
ds = load_from_disk("data/twi")                  # from --format disk
# or: Dataset.from_parquet("data.parquet")
ds = ds.train_test_split(test_size=0.1)        # feed your trainer

Optional: the selection UI

A local browser helper for exploring languages and building the command (it does not download — it hands you the afrispeech-select line to run):

pip install -e ".[ui]"     # adds gradio
python app.py              # opens http://127.0.0.1:7860

Use as a library

One language:

from afrispeech_selector import build_dataset, export_tts, stream_dataset

# Stream into training — no copy
ds = stream_dataset(["twi_twi"], split="train", max_seconds=5 * 3600, target_sampling_rate=16000)

# …or materialise and export for a TTS framework
copy = build_dataset(["twi_twi"], split="train", max_seconds=5 * 3600, streaming=True)
export_tts(copy, out_dir="./twi", fmt="ljspeech", sampling_rate=22050)

Several languages — name them, or select across the dataset:

from afrispeech_selector import filter_catalog, select_top, stream_dataset

langs = select_top(filter_catalog(min_hours=20), 10, proportional=True, max_per_country=2)
ds = stream_dataset(langs, split="train", per_language=200, target_sampling_rate=16000)

Tests

pip install -e ".[dev]"
pytest tests/        # selection, builder, and CLI tests (mostly offline)

Keeping the catalog current

afrispeech_selector/data/catalog.tsv is the static strength table (fast, offline; bundled in the package). Regenerate it when the source datasets change:

python3 scripts/refresh_catalog.py              # rebuild all four sources
python3 scripts/refresh_catalog.py --sources waxal,open-bible   # a subset

The script reads per-split row counts from the datasets-server API and estimates hours from sampled durations (WaxalNLP has no duration column, so its hours are apportioned from the published corpus totals). No audio is downloaded. Language ISO codes / names / countries come from the curated afriso list of African languages; responses are cached in ~/.cache/afrispeech-selector/ so re-runs stay offline.

Project layout

afrispeech_selector/
  cli.py        the `afrispeech-select` command (workhorse)
  catalog.py    load the language table; country names
  selector.py   ranking, filtering, country-proportional top-N, plan
  builder.py    pull subsets → standard schema; stream_dataset (no copy) + apply_schema
  export.py     HF (zip/parquet/manifest/push) + TTS (ljspeech/piper/vits/melo)
app.py          optional selection UI (emits the CLI command)
  data/catalog.tsv  strength table (4 source corpora, ~369 configs) — bundled in the package
scripts/        refresh_catalog.py
tests/          selection, builder, CLI tests

Data sources

The catalog merges four public corpora of African speech. A dataset column in data/catalog.tsv (surfaced as the src column in --list-langs and the --dataset CLI filter) keeps each row traceable to its source.

Corpus House in catalog Configs Clips Hours Notes
AfriSpeech afrispeech 142 778,520 2267.9 Short, clean read speech; the original corpus (hours exact)
Google WaxalNLP waxal 34 467,430 1430.0 ASR (19) + TTS (15) configs; hours apportioned from published totals
Meta OmniLingual ASR omnilingual 174 140,035 2417.9 African configs only (matched via afriso to the curated African-language list); hours estimated from sampled clip durations
Open Bible Speech (African) open-bible 19 552,907 1738.0 African Bible readings; hours estimated from sampled clip durations

Hours for WaxalNLP, OmniLingual and Open Bible are estimates derived from row counts and sampled clip durations (WaxalNLP publishes no per-clip durations, so it is split proportionally to its 1,250 h / 180 h card totals). The exact download volume also depends on your --min-clip-sec / --max-clip-sec window.

Check each dataset's card on the Hub for license and reuse terms.

Language catalog

All 369 language configs, sorted by hours (strength), with the source corpus in the last column. The --languages value is what you pass on the CLI; combine with --dataset to restrict a selection to one source (e.g. afrispeech-select --top 10 --dataset omnilingual).

Expand the full 369-config catalog — 308 languages, ~7,854 hours, 1,938,892 clips
#LanguageISOCountryHoursClipsTrain/Val/Test--languages valueSource
1TigrinyatirEritrea143.5250,31540,856/4,425/5,034tir_asrwaxal
2ChichewanyaMalawi139.0230,32228,805/0/1,517Chichewaopen-bible
3WolayttawalEthiopia134.8947,29141,833/2,444/3,014wal_asrwaxal
4SidamosidEthiopia129.5145,40438,752/3,091/3,561sid_asrwaxal
5OromoormEthiopia128.4945,04538,185/3,078/3,782orm_asrwaxal
6AmharicamhEthiopia126.5144,35438,022/2,912/3,420amh_asrwaxal
7IgboiboNigeria119.0230,01128,510/0/1,501Igboopen-bible
8GamogmvEthiopia118.7830,20028,690/0/1,510Gamoopen-bible
9North NdebelendeBotswana111.1730,16128,652/0/1,509Ndebeleopen-bible
10LingalalinDR Congo108.6828,79027,350/0/1,440Lingalaopen-bible
11KikuyukikKenya107.7630,72229,185/0/1,537Kikuyuopen-bible
12DawrodwrEthiopia99.9429,57928,100/0/1,479Dawroopen-bible
13YorubayorBenin95.8430,62529,093/0/1,532Yorubaopen-bible
14OromoormEthiopia93.6830,41328,892/0/1,521Oromoopen-bible
15Swahili (macrolanguage)swaBurundi93.0630,63429,102/0/1,532Swahiliopen-bible
16GandalugUganda91.1130,44028,918/0/1,522Lugandaopen-bible
17EweeweGhana86.2430,16428,655/0/1,509Eweopen-bible
18HausahauBurkina Faso83.6430,71729,181/0/1,536Hausaopen-bible
19ShonasnaBotswana82.4230,68529,150/0/1,535Shonaopen-bible
20TwitwiGhana78.9230,56529,036/0/1,529Twi (Asante)open-bible
21GofagofEthiopia78.7630,38828,868/0/1,520Gofaopen-bible
22FulahfulNigeria68.1223,88119,132/2,358/2,391ful_asrwaxal
23Luo (Kenya and Tanzania)luoKenya67.3630,42928,907/0/1,522Dholuoopen-bible
24MalagasymlgMadagascar65.8523,08618,526/2,270/2,290mlg_asrwaxal
25MalagasymlgMadagascar61.3220,28718,900/746/641malagasy_mlgafrispeech
26KabuverdianukeaCabo Verde58.0719,72818,156/699/873kabuverdianu_keaafrispeech
27TwitwiGhana55.2830,27028,756/0/1,514Twi (Akuapem)open-bible
28Southern DagaaredgaGhana53.8418,87415,071/1,893/1,910dga_asrwaxal
29EweeweGhana53.8018,86115,054/1,916/1,891ewe_asrwaxal
30ShonasnaZimbabwe53.3017,99615,954/1,161/881shona_snaafrispeech
31KabiyekbpTogo53.1918,21116,328/825/1,058kabiye_kbpafrispeech
32Eastern Yiddish (Hebr script)yddSouth Africa51.803,0001,950/300/750ydd_Hebromnilingual
33Northern Betsimisaraka MalagasybmmMadagascar51.762,9982,998/0/0bmm_Latnomnilingual
34OrmaorcKenya51.762,9981,500/900/598orc_Latnomnilingual
35Southern Betsimisaraka MalagasybzcMadagascar51.752,9972,997/0/0bzc_Latnomnilingual
36Tesaka MalagasytkgMadagascar51.752,9972,099/598/300tkg_Latnomnilingual
37Plateau MalagasypltMadagascar51.712,9952,995/0/0plt_Latnomnilingual
38LingalalinDR Congo51.6618,11014,400/1,844/1,866lin_asrwaxal
39PularfufGuinea51.512,9832,983/0/0fuf_Latnomnilingual
40Egyptian Arabic (Arab script)arzEgypt51.452,9802,384/298/298arz_Arabomnilingual
41Libyan Arabic (Arab script)aylEgypt51.452,9802,384/298/298ayl_Arabomnilingual
42IkposokpoGhana51.4018,02014,415/1,760/1,845kpo_asrwaxal
43Bagirmi FulfuldefuiCentral African Republic51.182,9642,964/0/0fui_Latnomnilingual
44DagbanidagGhana50.8317,81914,231/1,750/1,838dag_asrwaxal
45Tunisian Arabic (Arab script)aebAlgeria50.502,9252,925/0/0aeb_Arabomnilingual
46TwitwiGhana50.3217,14015,794/828/518twi_twiafrispeech
47ShonasnaZimbabwe50.1617,58514,109/1,727/1,749sna_asrwaxal
48Bassa (Cameroon)basCameroon48.5816,26914,981/620/668bassa_cameroon_basafrispeech
49Masikoro MalagasymshMadagascar46.572,6972,697/0/0msh_Latnomnilingual
50Antankarana MalagasyxmvMadagascar43.962,5461,803/450/293xmv_Latnomnilingual
51Mauritian CreolemfeMauritius43.5715,08813,115/822/1,151mauritian_creole_mfeafrispeech
52NyanekanykAngola42.9414,73113,430/858/443nyaneka_nykafrispeech
53PulaarfucGambia42.652,4702,470/0/0fuc_Latnomnilingual
54GunguwBenin39.9813,34012,157/494/689gun_guwafrispeech
55SwahiliswaTanzania38.7313,10111,781/703/617swahili_swaafrispeech
56KilubalubDR Congo38.3112,94911,553/679/717kiluba_lubafrispeech
57IgboiboNigeria38.0712,63911,379/556/704igbo_iboafrispeech
58ZuluzulSouth Africa36.9812,43411,540/571/323zulu_zulafrispeech
59AkanakaGhana36.3712,75210,107/1,123/1,522aka_asrwaxal
60Changana (Mozambique)tsoMozambique35.7712,39411,369/478/547changana_mozambique_tsoafrispeech
61KirundirunBurundi35.6212,39911,116/550/733kirundi_runafrispeech
62KinyarwandakinRwanda35.5712,01210,517/606/889kinyarwanda_kinafrispeech
63TsongatsoSouth Africa35.2511,83810,710/682/446tsonga_tsoafrispeech
64GagaaGhana35.1211,93910,703/632/604ga_gaaafrispeech
65Bara MalagasybhrMadagascar33.701,9521,952/0/0bhr_Latnomnilingual
66AmharicamhEthiopia33.5911,41510,340/603/472amharic_amhafrispeech
67FonfonBenin33.1910,91210,039/459/414fon_fonafrispeech
68Moroccan Arabic (Arab script)aryAlgeria33.191,9221,488/0/434ary_Arabomnilingual
69XhosaxhoSouth Africa33.1511,48510,073/658/754xhosa_xhoafrispeech
70OtetelatllDR Congo31.5311,30110,316/463/522otetela_tllafrispeech
71Tsimihety MalagasyxmwMadagascar31.081,8001,800/0/0xmw_Latnomnilingual
72Tanosy MalagasytxyMadagascar30.991,795898/448/449txy_Latnomnilingual
73LingalalinDR Congo30.7110,0809,076/414/590lingala_linafrispeech
74EweeweGhana29.419,7608,633/611/516ewe_eweafrispeech
75ChitumbukatumMalawi29.3110,3659,303/623/439chitumbuka_tumafrispeech
76RongarngMozambique29.1110,2859,138/642/505ronga_rngafrispeech
77KambakamKenya28.869,5678,603/348/616kamba_kamafrispeech
78KikuyukikKenya28.829,6408,808/413/419kikuyu_kikafrispeech
79HausahauNigeria28.379,3678,364/509/494hausa_hauafrispeech
80MacuavmwMozambique28.0210,1448,966/682/496macua_vmwafrispeech
81KongokonDR Congo27.829,9808,828/474/678kongo_konafrispeech
82MatengomgvTanzania27.327,7927,402/0/390Matengoopen-bible
83IsokoisoNigeria26.849,4748,544/447/483isoko_isoafrispeech
84KriokriSierra Leone26.169,2308,293/553/384krio_kriafrispeech
85EdobinNigeria26.049,0288,323/346/359edo_binafrispeech
86BomubmqBurkina Faso25.901,5001,500/0/0bmq_Latnomnilingual
87Maasina FulfuldeffmBurkina Faso25.901,5001,500/0/0ffm_Latnomnilingual
88Sakalava MalagasyskgMadagascar25.881,4991,499/0/0skg_Latnomnilingual
89Algerian Arabic (Arab script)arqAlgeria25.711,4891,489/0/0arq_Arabomnilingual
90UrhobourhNigeria25.678,9447,946/447/551urhobo_urhafrispeech
91OromoormEthiopia25.668,9337,778/557/598oromo_ormafrispeech
92Central-Eastern Niger FulfuldefuqMali25.481,476600/431/445fuq_Latnomnilingual
93Saidi Arabic (Arab script)aecEgypt25.351,4681,168/150/150aec_Arabomnilingual
94MasaabamyxUganda24.468,5746,867/849/858mas_asrwaxal
95EsanishNigeria24.198,4997,356/708/435esan_ishafrispeech
96NyankolenynUganda24.178,4726,783/831/858nyn_asrwaxal
97FulahfulNigeria23.913,8783,100/380/398ful_ttswaxal
98FrafragurGhana23.827,9607,140/398/422frafra_gurafrispeech
99Sesotho (South Africa)sotSouth Africa23.328,1497,583/332/234sesotho_south_africa_sotafrispeech
100SenasehMozambique22.847,8497,019/341/489sena_sehafrispeech
101Sesotho (Lesotho)sotSouth Africa22.767,9447,052/390/502sesotho_lesotho_sotafrispeech
102LenjelehZambia22.528,0997,198/412/489lenje_lehafrispeech
103SogaxogUganda22.477,8776,219/841/817sog_asrwaxal
104Liberian EnglishlirLiberia22.067,7337,112/335/286liberian_english_lirafrispeech
105FantefatGhana22.047,6316,942/277/412fante_fatafrispeech
106DagaaredgaGhana21.537,4776,575/375/527dagaare_dgaafrispeech
107NzemanziGhana21.147,3196,413/532/374nzema_nziafrispeech
108Pidgin (West Africa)wesCameroon21.097,4056,785/227/393pidgin_west_africa_wesafrispeech
109KisikssLiberia20.997,7396,967/377/395kisi_kssafrispeech
110Borgu FulfuldefueBenin20.671,1971,197/0/0fue_Latnomnilingual
111FantifatGhana20.671,1971,197/0/0fat_Latnomnilingual
112MooremosBurkina Faso20.516,7856,293/196/296moore_mosafrispeech
113AhantaahaGhana20.306,9086,038/441/429ahanta_ahaafrispeech
114GusilaygslSenegal20.251,173775/238/160gsl_Latnomnilingual
115RomborofKenya20.201,170653/241/276rof_Latnomnilingual
116DoualaduaCameroon20.126,5215,832/145/544douala_duaafrispeech
117LuoluoKenya19.666,8295,959/367/503luo_luoafrispeech
118SepedinsoSouth Africa19.597,0166,435/396/185sepedi_nsoafrispeech
119Bissau Guinean CreolepovGuinea-Bissau19.356,5735,899/375/299bissau_guinean_creole_povafrispeech
120GandalugUganda19.276,7575,455/664/638lug_asrwaxal
121Sudanese Arabic (Arab script)apdCentral African Republic18.941,097816/156/125apd_Arabomnilingual
122NobiinfiaEgypt18.871,093668/262/163fia_Latnomnilingual
123Swahili (Congo)swcDR Congo18.676,4935,670/503/320swahili_congo_swcafrispeech
124Western ManinkakanmlqGambia18.561,075621/200/254mlq_Latnomnilingual
125Judeo-Moroccan Arabic (Hebr script)ajuMorocco17.531,0151,015/0/0aju_Hebromnilingual
126SehwisfwGhana17.486,0245,488/199/337sehwi_sfwafrispeech
127RunyankorenynUganda17.105,6054,736/432/437runyankore_nynafrispeech
128BouloubumCameroon16.935,6925,153/231/308boulou_bumafrispeech
129KwanyamakuaNamibia16.666,1075,467/239/401kwanyama_kuaafrispeech
130ToorottjDR Congo16.54958674/110/174ttj_Latnomnilingual
131YorubayorNigeria16.102,6112,233/201/177yor_ttswaxal
132KonzokooDR Congo15.94923654/146/123koo_Latnomnilingual
133JuladyuBurkina Faso15.855,4865,019/258/209jula_dyuafrispeech
134Western Niger FulfuldefuhBurkina Faso15.64906906/0/0fuh_Latnomnilingual
135Tandroy-Mahafaly MalagasytdxMadagascar15.54900450/150/300tdx_Latnomnilingual
136TshwatscMozambique14.995,2214,651/209/361tshwa_tscafrispeech
137AcoliachUganda14.705,1554,108/519/528ach_asrwaxal
138MandjakmfvGambia14.28827488/179/160mfv_Latnomnilingual
139YorubayorNigeria14.194,9604,504/216/240yoruba_yorafrispeech
140RangilagTanzania14.16820477/180/163lag_Latnomnilingual
141Balanta-GanjabjtGuinea-Bissau13.95808576/109/123bjt_Latnomnilingual
142ChigacggDR Congo13.90805559/126/120cgg_Latnomnilingual
143SetswanatsnSouth Africa13.844,8284,232/344/252setswana_tsnafrispeech
144Kikongo ya LetaktuDR Congo13.814,7544,232/262/260kikongo_ya_leta_ktuafrispeech
145Liberian EnglishlirLiberia13.76797650/71/76lir_Latnomnilingual
146Gambian WolofwofGambia13.21765548/116/101wof_Latnomnilingual
147ChichewanyaMalawi12.813,8843,407/231/246chichewa_nyaafrispeech
148AkebukeuTogo12.67734494/91/149keu_Latnomnilingual
149SangosagCentral African Republic12.604,1933,645/261/287sango_sagafrispeech
150AcoliachUganda12.522,0311,621/218/192ach_ttswaxal
151KikuyukikKenya12.492,0261,602/210/214kik_ttswaxal
152GandalugUganda12.482,0241,608/211/205lug_ttswaxal
153MashishrDR Congo12.414,2633,855/273/135mashi_shrafrispeech
154Nigerian PidginpcmNigeria12.291,9931,590/199/204pcm_ttswaxal
155Luo (Kenya and Tanzania)luoKenya12.261,9891,552/228/209luo_ttswaxal
156NyankolenynUganda12.251,9871,597/199/191nyn_ttswaxal
157Seychelles CreolecrsSeychelles12.164,2693,983/72/214seychelles_creole_crsafrispeech
158HausahauNigeria12.151,9711,572/202/197hau_ttswaxal
159BayotbdaGambia11.83685450/158/77bda_Latnomnilingual
160IgboiboNigeria11.781,9111,552/159/200ibo_ttswaxal
161FangfanEquatorial Guinea11.683,9693,630/184/155fang_fanafrispeech
162TugentuyKenya11.52667407/104/156tuy_Latnomnilingual
163FipafipMalawi11.46664410/106/148fip_Latnomnilingual
164GwenogweKenya11.34657401/123/133gwe_Latnomnilingual
165LugandalugUganda11.253,6173,237/75/305luganda_lugafrispeech
166KitubaktuDR Congo11.243,9353,639/195/101kituba_ktuafrispeech
167TshilubaluaDR Congo11.243,5323,234/158/140tshiluba_luaafrispeech
168KeiyoeyoKenya11.17647398/109/140eyo_Latnomnilingual
169FarefaregurBurkina Faso11.08642476/66/100gur_Latnomnilingual
170Swahili (macrolanguage)swaKenya10.961,7781,387/192/199swa_ttswaxal
171ChopicceMozambique10.803,7893,414/144/231chopi_cceafrispeech
172WarjiwjiNigeria10.62615443/98/74wji_Latnomnilingual
173CibembabemZambia10.563,8233,333/170/320cibemba_bemafrispeech
174Bago-KusuntubqgTogo10.52609352/99/158bqg_Latnomnilingual
175AnaanganwNigeria10.41603369/115/119anw_Latnomnilingual
176Tahaggart Tamahaq (Tfng script)thvAlgeria10.29596298/149/149thv_Tfngomnilingual
177NgangelanbaAngola10.283,6383,348/166/124ngangela_nbaafrispeech
178GolagolLiberia10.24593347/120/126gol_Latnomnilingual
179Ndau (Western)ndcMozambique10.223,6213,235/239/147ndau_western_ndcafrispeech
180GbarigbyNigeria10.01580409/97/74gby_Latnomnilingual
181PhimbiphmMozambique10.003,5613,172/121/268phimbi_phmafrispeech
182GbagyigbrNigeria9.82569396/94/79gbr_Latnomnilingual
183DũyaldbNigeria9.76565401/74/90ldb_Latnomnilingual
184Dera (Nigeria)knaNigeria9.69561376/77/108kna_Latnomnilingual
185IgoahlGhana9.58555376/87/92ahl_Latnomnilingual
186MansoankamswGambia9.58555379/88/88msw_Latnomnilingual
187Samba LekondiCameroon9.58555349/134/72ndi_Latnomnilingual
188SoninkesnkGambia9.57554385/91/78snk_Latnomnilingual
189PeropipNigeria9.50550322/124/104pip_Latnomnilingual
190BondeibouTanzania9.43546373/103/70bou_Latnomnilingual
191Guduf-GavagdfCameroon9.43546376/102/68gdf_Latnomnilingual
192SwatisswSouth Africa9.423,2362,767/170/299swati_sswafrispeech
193Cakfem-MushereckyNigeria9.41545307/123/115cky_Latnomnilingual
194EweeweGhana9.361,5191,215/152/152ewe_ttswaxal
195BurakbysNigeria9.31539360/103/76bys_Latnomnilingual
196IdomaiduNigeria9.24535348/115/72idu_Latnomnilingual
197KalabariijnNigeria9.24535341/94/100ijn_Latnomnilingual
198VaivaiLiberia9.20533299/87/147vai_Latnomnilingual
199IsekiriitsNigeria9.19532369/80/83its_Latnomnilingual
200NgizimngiNigeria9.13529352/104/73ngi_Latnomnilingual
201Eastern KrahnkqoCôte d'Ivoire9.12528280/106/142kqo_Latnomnilingual
202EloyiafoNigeria9.10527357/70/100afo_Latnomnilingual
203NabamneChad9.08526251/176/99mne_Latnomnilingual
204AlagoalaNigeria9.05524341/86/97ala_Latnomnilingual
205LijilimgiNigeria9.00521353/70/98mgi_Latnomnilingual
206UbagharabycNigeria9.00521349/86/86byc_Latnomnilingual
207KingazgaMalawi8.98520520/0/0zga_Latnomnilingual
208ToupourituiCameroon8.963,0702,588/292/190toupouri_tuiafrispeech
209AbronabrCôte d'Ivoire8.94518302/87/129abr_Latnomnilingual
210ElemeelmNigeria8.91516350/90/76elm_Latnomnilingual
211MbemfoNigeria8.87514341/80/93mfo_Latnomnilingual
212KhanaogoNigeria8.79509332/77/100ogo_Latnomnilingual
213DegemadegNigeria8.77508339/90/79deg_Latnomnilingual
214EjaghametuCameroon8.77508325/95/88etu_Latnomnilingual
215IkwereikwNigeria8.70504318/91/95ikw_Latnomnilingual
216GoemaiankNigeria8.68503325/107/71ank_Latnomnilingual
217IsokoisoNigeria8.68503325/79/99iso_Latnomnilingual
218Bura-PabirbwrNigeria8.67502352/82/68bwr_Latnomnilingual
219VendavenSouth Africa8.652,9902,721/135/134venda_venafrispeech
220Basa (Nigeria)bzwNigeria8.63500336/72/92bzw_Latnomnilingual
221YaceekrNigeria8.63500351/74/75ekr_Latnomnilingual
222ItoitwNigeria8.58497341/79/77itw_Latnomnilingual
223MiyamkfNigeria8.58497342/67/88mkf_Latnomnilingual
224Nigerian FulfuldefuvCameroon8.56496347/74/75fuv_Latnomnilingual
225AbuaabnNigeria8.55495317/86/92abn_Latnomnilingual
226NgamonbhNigeria8.53494311/73/110nbh_Latnomnilingual
227KuanyamakuaAngola8.48491312/95/84kua_Latnomnilingual
228KwambikwmNamibia8.48491328/75/88kwm_Latnomnilingual
229WajawjaNigeria8.48491345/73/73wja_Latnomnilingual
230GejigyzNigeria8.46490315/92/83gyz_Latnomnilingual
231BadebdeNigeria8.44489314/101/74bde_Latnomnilingual
232TsotsoltoKenya8.44489316/98/75lto_Latnomnilingual
233LogooliragKenya8.43488306/72/110rag_Latnomnilingual
234Dijim-BwilimcfaNigeria8.41487334/69/84cfa_Latnomnilingual
235WapanjukNigeria8.41487307/75/105juk_Latnomnilingual
236BoghombuxNigeria8.39486273/118/95bux_Latnomnilingual
237CibakcklNigeria8.39486299/78/109ckl_Latnomnilingual
238KohumonobcsNigeria8.39486347/70/69bcs_Latnomnilingual
239Mom JangoverCameroon8.39486323/68/95ver_Latnomnilingual
240AsheahsNigeria8.36484329/84/71ahs_Latnomnilingual
241NyungwenyuMozambique8.352,9762,744/63/169nyungwe_nyuafrispeech
242WangalwgKenya8.34483297/76/110lwg_Latnomnilingual
243KimbundukmbAngola8.322,8942,468/249/177kimbundu_kmbafrispeech
244MafamafCameroon8.32482325/88/69maf_Latnomnilingual
245OdualoduNigeria8.32482307/90/85odu_Latnomnilingual
246EsanishNigeria8.31481308/98/75ish_Latnomnilingual
247GusiiguzKenya8.29480287/117/76guz_Latnomnilingual
248TurkanatuvEthiopia8.27479285/106/88tuv_Latnomnilingual
249YekheeetsNigeria8.25478320/79/79ets_Latnomnilingual
250CencenNigeria8.24477313/90/74cen_Latnomnilingual
251HwanahwoNigeria8.24477327/68/82hwo_Latnomnilingual
252KomakmyCameroon8.24477326/65/86kmy_Latnomnilingual
253IkposokpoGhana8.22476293/90/93kpo_Latnomnilingual
254TangaletanNigeria8.22476323/69/84tan_Latnomnilingual
255KarekarekaiNigeria8.18474331/69/74kai_Latnomnilingual
256LamanghiaNigeria8.18474324/85/65hia_Latnomnilingual
257Cross River MbembemfnNigeria8.17473284/89/100mfn_Latnomnilingual
258Idakho-Isukha-TirikiidaKenya8.17473305/100/68ida_Latnomnilingual
259KinandennbDR Congo8.152,8712,637/44/190kinande_nnbafrispeech
260HubahbbNigeria8.12470288/63/119hbb_Latnomnilingual
261Marghi CentralmrtNigeria8.12470297/80/93mrt_Latnomnilingual
262Waci GbewciBenin8.12470308/95/67wci_Latnomnilingual
263KamokcqNigeria8.10469321/66/82kcq_Latnomnilingual
264NyankpayesNigeria8.10469310/68/91yes_Latnomnilingual
265Nsenga (Mozambique)nseMozambique8.092,8652,385/313/167nsenga_mozambique_nseafrispeech
266BaoulebciCôte d'Ivoire8.082,6572,307/125/225baoule_bciafrispeech
267BokyibkyCameroon8.08468297/93/78bky_Latnomnilingual
268JibajuoNigeria8.06467321/68/78juo_Latnomnilingual
269Piya-KwoncipiyNigeria8.06467324/76/67piy_Latnomnilingual
270DadiyadbdNigeria8.05466289/70/107dbd_Latnomnilingual
271IzonijcNigeria8.05466290/80/96ijc_Latnomnilingual
272PökootpkoKenya8.03465298/70/97pko_Latnomnilingual
273Lala-RoballaNigeria7.99463301/91/71lla_Latnomnilingual
274GokanagknNigeria7.982,8082,567/179/62gokana_gknafrispeech
275JjukajNigeria7.96461272/91/98kaj_Latnomnilingual
276KabraslkbKenya7.96461284/72/105lkb_Latnomnilingual
277LongudalnuNigeria7.96461299/63/99lnu_Latnomnilingual
278NzanyinjaCameroon7.94460302/87/71nja_Latnomnilingual
279TarokyerNigeria7.94460298/70/92yer_Latnomnilingual
280BudumabdmCameroon7.93459301/90/68bdm_Latnomnilingual
281NgasancNigeria7.91458314/65/79anc_Latnomnilingual
282KushikuhNigeria7.89457308/77/72kuh_Latnomnilingual
283Marghi SouthmfmNigeria7.89457315/78/64mfm_Latnomnilingual
284NdongandoAngola7.89457285/68/104ndo_Latnomnilingual
285Samba DakaccgCameroon7.89457292/95/70ccg_Latnomnilingual
286SayasayNigeria7.89457260/103/94say_Latnomnilingual
287EggonegoNigeria7.87456305/68/83ego_Latnomnilingual
288EmbuebuKenya7.87456272/71/113ebu_Latnomnilingual
289GlavdaglwCameroon7.87456322/65/69glw_Latnomnilingual
290KanembukblNiger7.87456306/79/71kbl_Latnomnilingual
291Kirya-KonzəlfkkNigeria7.87456290/88/78fkk_Latnomnilingual
292Kulung (Nigeria)bbuNigeria7.84454281/67/106bbu_Latnomnilingual
293MerumerKenya7.80452279/105/68mer_Latnomnilingual
294Maba (Chad)mdeChad7.68445445/0/0mde_Latnomnilingual
295TulatulNigeria7.68445263/112/70tul_Latnomnilingual
296AfadeaalCameroon7.67444310/64/70aal_Latnomnilingual
297BacamabcyNigeria7.65443278/99/66bcy_Latnomnilingual
298HereroherAngola7.65443281/95/67her_Latnomnilingual
299Ndebele (Zimbabwe)ndeZimbabwe7.622,5522,308/81/163ndebele_zimbabwe_ndeafrispeech
300ZarmadjeBurkina Faso7.61441316/61/64dje_Latnomnilingual
301HavuhavDR Congo7.592,6192,447/172/0havu_havafrispeech
302BangwinjibsjNigeria7.58439286/84/69bsj_Latnomnilingual
303DghwededghCameroon7.58439263/109/67dgh_Latnomnilingual
304TedagatuqLibya7.56438287/87/64tuq_Latnomnilingual
305TerattrNigeria7.53436294/75/67ttr_Latnomnilingual
306BaoulébciCôte d'Ivoire7.501,216972/122/122bau_ttswaxal
307IbindayomDR Congo7.462,6722,508/59/105ibinda_yomafrispeech
308DazagadzgLibya7.39428261/65/102dzg_Latnomnilingual
309AwakawoNigeria7.37427265/71/91awo_Latnomnilingual
310FantifatGhana7.221,171953/117/101fat_ttswaxal
311Liberia KpellexpeGuinea7.22418266/84/68xpe_Latnomnilingual
312AgwagwuneyayNigeria6.84396227/87/82yay_Latnomnilingual
313DinkadinSouth Sudan6.842,4962,178/99/219dinka_dinafrispeech
314ItsekiriitsNigeria6.782,3872,387/0/0itsekiri_itsafrispeech
315KwangalikwnNamibia6.752,2582,139/81/38kwangali_kwnafrispeech
316TwitwiGhana6.741,093872/117/104twi_ttswaxal
317ChitongatoiZambia6.612,4872,220/195/72chitonga_toiafrispeech
318Bassa (Liberia)bsqLiberia6.072,2842,135/12/137bassa_liberia_bsqafrispeech
319CinyanjanyaMalawi5.791,8831,765/42/76cinyanja_nyaafrispeech
320NdongandoNamibia5.771,9201,745/63/112ndonga_ndoafrispeech
321AjaajgBenin5.711,9421,765/105/72aja_ajgafrispeech
322KpellexpeLiberia5.701,9331,743/112/78kpelle_xpeafrispeech
323Réunion CreolercfRéunion5.251,8031,718/23/62r_union_creole_rcfafrispeech
324Eastern Egyptian Bedawi Arabic (Arab script)avlEgypt5.18300300/0/0avl_Arabomnilingual
325Algerian Saharan Arabic (Arab script)aaoAlgeria5.15298298/0/0aao_Arabomnilingual
326NdebelenblSouth Africa5.151,9851,819/134/32ndebele_nblafrispeech
327Chadian ArabicshuCameroon5.13297297/0/0shu_Latnomnilingual
328AbbeyabaCôte d'Ivoire5.031,7761,574/166/36abbey_abaafrispeech
329YombeyomDR Congo4.881,6341,390/149/95yombe_yomafrispeech
330KikongokwyAngola4.871,7691,623/103/43kikongo_kwyafrispeech
331UmbunduumbAngola4.701,6301,340/227/63umbundu_umbafrispeech
332ChiyaoyaoMozambique4.611,6471,423/103/121chiyao_yaoafrispeech
333LomalomLiberia4.521,5401,350/41/149loma_lomafrispeech
334WolaitawalEthiopia4.281,4761,250/133/93wolaita_walafrispeech
335Chitonga (Malawi)togMalawi4.161,4801,363/65/52chitonga_malawi_togafrispeech
336TivtivNigeria4.031,3741,220/119/35tiv_tivafrispeech
337LarildiCongo3.851,3591,284/42/33lari_ldiafrispeech
338MerumerKenya3.831,2771,139/138/0meru_merafrispeech
339EwondoewoCameroon3.711,3051,134/32/139ewondo_ewoafrispeech
340KabylekabAlgeria3.651,1881,068/64/56kabyle_kabafrispeech
341KhanaogoNigeria3.621,256921/139/196khana_ogoafrispeech
342GitongatohMozambique3.601,3191,228/4/87gitonga_tohafrispeech
343TewetwxMozambique3.391,2511,076/148/27tewe_twxafrispeech
344DangmeadaGhana3.371,177941/154/82dangme_adaafrispeech
345NdaundcMozambique3.341,2311,147/68/16ndau_ndcafrispeech
346GuérégxxCôte d'Ivoire3.011,057953/67/37gu_r_gxxafrispeech
347WolofwolSenegal2.50846803/10/33wolof_wolafrispeech
348DamaranaqNamibia2.46789698/62/29damara_naqafrispeech
349Swahili (Katanga)swcDR Congo2.40866798/32/36swahili_katanga_swcafrispeech
350YacoubadafCôte d'Ivoire2.26787656/0/131yacouba_dafafrispeech
351ManyawamnyMozambique1.94697697/0/0manyawa_mnyafrispeech
352Makhuwa-MarrevonexmcMozambique1.87704635/41/28makhuwa_marrevone_xmcafrispeech
353Makhuwa-MeettomghMozambique1.83671552/85/34makhuwa_meetto_mghafrispeech
354CinamwangamwnZambia1.66553500/53/0cinamwanga_mwnafrispeech
355Chitonga (Zimbabwe)toiZimbabwe1.41486452/0/34chitonga_zimbabwe_toiafrispeech
356AttiéatiCôte d'Ivoire1.40482477/5/0atti_atiafrispeech
357LundalunZambia1.11423322/101/0lunda_lunafrispeech
358LomwenglMozambique1.07378373/5/0lomwe_nglafrispeech
359ChuabochwMozambique1.03390373/0/17chuabo_chwafrispeech
360Mambwe-LungumgrZambia0.93331264/34/33mambwe_lungu_mgrafrispeech
361IjawijcNigeria0.88312292/0/20ijaw_ijcafrispeech
362Ngbandi (Northern)ngbDR Congo0.78255222/33/0ngbandi_northern_ngbafrispeech
363Makhuwa-ShirimavmkMozambique0.77277277/0/0makhuwa_shirima_vmkafrispeech
364HereroherNamibia0.61192192/0/0herero_herafrispeech
365ChokwecjkAngola0.56177169/0/8chokwe_cjkafrispeech
366TaabwatapDR Congo0.56195163/0/32taabwa_tapafrispeech
367KisongesopDR Congo0.40141141/0/0kisonge_sopafrispeech
368KanyokknyDR Congo0.28109109/0/0kanyok_knyafrispeech
369LuvalelueZambia0.072020/0/0luvale_lueafrispeech

License

CC-BY-4.0

About

Easy access to speech data across 142 African languages for training TTS and ASR models.

Topics

Resources

Stars

16 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages