riselle-r/south-african-accent-tts

# 🇿🇦 Gemini Live Multimodal Speech & Language Studio (South African Accent Edition) A high-performance, modular Streamlit web application showcasing **Google Gemini's Multimodal Live API** with continuous, low-latency bidirectional voice conversation, dynamic accent and dialect personalization tailored specifically for **South African accents and languages**, real-time audio visualizers, and synchronized live transcription. --- ## 🌟 Key Features 1. **Continuous Low-Latency Bidirectional Streaming**: - Web Audio API capturing 16kHz linear PCM mono from browser mic. - 24kHz linear PCM streaming playback with low-latency audio buffer queue. - Natural barge-in / speech interruption handling with instant audio purge. 2. **South African Dialect & Accent Personalization**: - **Primary Languages**: English (South Africa), Afrikaans, isiZulu, isiXhosa, Sesotho, Setswana, Sepedi. - **Regional Accent Presets**: - *General South African (GSA / Neutral)*: Polished, corporate non-rhotic articulation. - *Afrikaans-influenced English (Pretoria / Platteland)*: Rolled/tapped 'r's, distinctive vowel shifts, direct and warm. - *KZN / Durban English (Coastal cadence)*: Laid-back coastal lilt, relaxed vowel lengths, melodic rising inflection. - *Cape / Coloured English (Western Cape cadence)*: Vibrant rhythmic staccato cadence, dynamic intonation. - *isiZulu / African-influenced English (Gauteng / Urban township cadence)*: Syllable-timed, open vowels, resonant pitch variations, Tsotsitaal flourishes. - *Custom Dialect*: User-defined free-form phonetic and dialect directives. - **Colloquialism & Idiom Intensity (1 to 5)**: From Corporate / Subtle to Authentic & Mzansi Expressive (*Howzit, Lekker, Yebo, Now-now, Sharp sharp*). - **Gemini Voice Profiles**: Puck, Charon, Kore, Fenrir, Aoede. 3. **Synchronized Live Transcript & Visualizer**: - Real-time side-by-side transcription of user speech and Gemini speech. - Live animated dual-channel waveform visualizer (User Mic & Gemini Output …