Speak
natively.
For filmmakers, creators, and anyone who needs a Nigerian language voice at scale.
Pass a sentence with tone marks. Get back native-speaker audio with the correct pitch contour, vowel quality and rhythm — at 22 kHz.
Selected from 13,000+ applicants · iDICE Startup Bridge Founders Lab · Backed by Ekiti State Ministry of Digital Economy
Voice AI for African languages · Pilot: Yorùbá
African languages finally
have a voice.
Laban is voice infrastructure built for the 1.4 billion people global AI quietly skipped. Text-to-speech, transcription, voice cloning, and real-time interpretation — engineered for tone, not for English.
Ẹ kú àárọ̀ — Good morningNdeewo — Hello in IgboSannu — Hello in Hausa50,000,000 Yorùbá speakers3 tones · 12 vowels · 1 voice engineẸ kú àárọ̀ — Good morningNdeewo — Hello in IgboSannu — Hello in Hausa50,000,000 Yorùbá speakers3 tones · 12 vowels · 1 voice engine
The Problem
Every voice AI platform was built for English. When they attempt African languages, they don't just sound robotic — they say the wrong thing entirely.
Get the tone wrong, and it’s not a mispronunciation — it’s a different word.
In Yorùbá, the same letters carry four different meanings depending on tone and vowel quality. Click any card to hear the pitch contour change the word.
Today on these platforms, Yorùbá is either absent or wrong: ElevenLabsOpenAI VoiceGoogle TTSAzure SpeechAmazon Polly
Watch
Sixty seconds
in motion.
Four primitives
One stack. Four ways to put African voice into your product.
Speak
Transcribe,
with tone.
For transcribing Yorùbá field recordings, phone calls, and media content with full tonal accuracy.
Fine-tuned Whisper-large-v3 returns clean Yorùbá orthography — diacritics, underdotted vowels and all — from phone calls, field recordings, video.
Ẹ kú àárọ̀ o
Preserve
a voice.
For studios dubbing films and creators producing audio content in their own voice.
A 30-second reference clip yields a speaker embedding. Film studios dub. Authors narrate audiobooks. Every clone carries an audible watermark and an explicit consent record.
Translate,
live.
For multilingual meetings, classrooms, and earpiece interpretation across African language pairs.
Streaming ASR → NLLB-200 → TTS, end-to-end under 1.05 seconds. The foundation for multilingual classrooms and pan-African meetings.
ENHow are you today?
↓YOBáwo ni o ṣe wà lónìí?
Who it’s for
For filmmakers
Dub your production into any Nigerian language for a fraction of the current cost. No studio. No re-casting. Broadcast quality.
For content creators
Generate voiceovers in your own voice in any Nigerian language in minutes. One recording. Unlimited output.
For developers
Drop a tone-aware African language voice layer into your product with a single API call.
Pilot language
Why we started with Yorùbá.
Native speakers
0M+
across Nigeria, Bénin, Togo and the diaspora.
Contrastive tones
0
High, mid, low. Tone changes the word, not the mood.
Distinct vowels
0
Seven oral + five nasal. ẹ and ọ are not e and o.
Studio hours, target
0hr
Balanced across Èkìtì, Òyó, Òǹdó, Lagos and Ìjẹ̀bú.
A system that ignores tone isn't speaking imperfect Yorùbá. It is speaking a different word entirely.
— Laban Technical Framework v2.0 · §3
Yorùbá · liveIgbo · Q2 '26Hausa · Q2 '26ÈfìkSwahili
Build with us
Be first when Laban opens.
We are accepting a limited number of early partners across film, edtech, healthcare, and broadcast — for early API access, sample audio packs, and a direct line to the team building this.
Limited cohort · No spam, just a heads-up when your language goes live.