Platform

One voice stack. Built for how Africa speaks.

The first products from the Waridi lab: models that listen, speak and hold a conversation in Swahili, available through one API.

Customer call · Billing support

sw-TZSpeakers
00:00

Habari, nataka ku-check balance yangu na kulipa bill ya umeme kabla ya Ijumaa.

00:07

Sawa kabisa. Salio lako ni shilingi elfu kumi na mbili. Nikulipie bill sasa hivi?

00:12

Ndiyo, tafadhali. Asante sana…

Speech recognition

Waridi Sikia

Hears Swahili the way people speak it.

Sikia (“listen”) transcribes everyday Swahili on real phone lines and in noisy places, and treats code-switching as normal speech rather than an error.

  • Tanzanian Swahili first, Kenyan and regional varieties next
  • Swahili–English code-switching and Sheng
  • Batch files or real-time streaming
POST /v1/stt
Karibu sana! Agizo lako limefika na litawasilishwa kesho asubuhi.
Zawadi · Warm · DarGenerate
0:04
Text to speech & voice cloning

Waridi Sauti

Voices that sound like home.

Sauti (“voice”) speaks Swahili with natural stress, rhythm and pronunciation, from customer-service prompts to long-form narration. Custom voices are only cloned with the verified consent of the voice owner.

  • sauti-standard and sauti-pro model tiers
  • Expressive, accent-true preview voices
  • Consent-verified voice cloning
POST /v1/tts
A customer on a call with a Waridi agent
✦
Issue resolvedLine restored in 42s
Waridi agentSpeaking Kiswahili
● 00:42

“Pole sana! Nimeona tatizo kwenye laini yako na nimeirekebisha. Tafadhali washa upya simu.”

TransferEnd call
Conversational voice agents

Waridi Agents

Retire “press one”.

Agents answer the phone in Swahili, understand what callers need and resolve it in real time, instead of routing people through menus they never asked for.

  • Real-time turn-taking and barge-in
  • Telephony via SIP and WebRTC
  • Hand-off to human agents with full context
POST /v1/agents
A saxophonist performing in a studio🎬 Dubbing · EN → SW
AI dubbing & localization

Waridi Dub

Bring every story home.

Dub films, series, courses and podcasts into Swahili with voices that keep the speaker’s intent, timing and emotion, for broadcasters, studios and creators.

  • Timing-aware translation and voicing
  • Multiple speakers per programme
  • Review and edit in Waridi Studio
POST /v1/dub
  1. Sikia listens
  2. Your app responds
  3. Sauti speaks
Speech infrastructure

Waridi API & Studio

Bring your own model.

Waridi handles the voice: Sikia listens, Sauti speaks. Your team keeps the language model it already uses, connected through REST and streaming endpoints, Python and Node SDKs, or Waridi Studio without code.

  • Python (waridi) and Node (@waridi/sdk) SDKs
  • Streaming over WebSockets
  • Usage-based pricing with a free sandbox
wss://api.waridi.io/v1/stream
Private beta · Now onboarding

Kila sauti ina thamani.
Every voice matters.

Build with the voice stack made for African languages.

Request early accessRead the docs