Research Engineer, Speech Synthesis
Build Waridi Sauti voices with natural Swahili stress, rhythm and expression.
About the role
You’ll work across modelling, data and listening tests to make Sauti voices sound unmistakably Tanzanian — warm, clear and expressive — and fast enough for live conversation.
What you’ll do
- Train and fine-tune TTS and voice-cloning models
- Work with linguists on pronunciation, prosody and text normalisation for Swahili
- Run listening panels with native speakers and act on the results
- Optimise models for low-latency streaming inference
What we’re looking for
- Experience with modern TTS or audio generation models
- Solid Python and PyTorch
- A good ear, and care for how speech sounds
Nice to have
- Native or fluent Swahili
- Experience with consent-based voice cloning
- Audio DSP background
How we hire
A short intro call, a practical exercise close to the real work, a conversation with the team, and a decision within two weeks. We don’t expect you to tick every box — if this excites you, apply.