tzvaita/voiceai-zw-lab

Voice support AI for Shona, Ndebele, and Zimbabwean code-switched speech. # VoiceAI — Conversational support AI for Zimbabwean languages Building a voice-based customer support assistant that works in Shona, Ndebele, and code-switched Zimbabwean English. **Status:** Phase 1 — retrieval backend (in progress) ## Why Voice assistants in Zimbabwe serve English speakers well and everyone else badly. ASR for Shona is thinly researched; for Ndebele, barely at all. This project builds a working support assistant end to end, then tackles the speech layer for the languages people actually speak. ## Architecture Speech → ASR → retrieval (pgvector) → LLM → TTS → Speech ## Roadmap - [ ] **Phase 1** — RAG support chatbot (FastAPI, pgvector, Ollama) - [ ] **Phase 2** — WhatsApp channel - [ ] **Phase 3** — English voice loop (Whisper → RAG → TTS) - [ ] **Phase 4** — Shona/Ndebele ASR: dataset, baselines, fine-tuning ## Stack Python 3.11 · PyTorch 2.13 · FastAPI · PostgreSQL 17 + pgvector · Ollama (Qwen2.5) · faster-whisper · HuggingFace transformers ## Log Build notes and benchmarks in `docs/`. --- Tennyson Zvaita · Harare, Zimbabwe