Enterprise meetings, telehealth consultations, university lectures, and customer support calls generate thousands of hours of audio and video every day. Audio-only transcription misses crucial information shown on screen—such as financial tables on presentation slides, shared code windows, or clinical imagery.
If you are looking to hire a freelance multimodal video, audio transcription, and speech AI programmer, I am Aldian Fazrihady—creator of Transcribe Copilot and Interview Copilot.
Multimodal Speech & Video AI Systems I Build
1. Ultra-Fast Audio Diarization & Cost-Optimized Transcription
By combining Voice Activity Detection (VAD), FFmpeg audio normalization, Groq Whisper Large-v3, and Gemini 2.5 Flash (Replacing Expensive SaaS Transcription with a Self-Hosted Gemini & Groq Copilot on Cloud Run), I deliver speaker-diarized transcripts 90% cheaper than off-the-shelf APIs.
2. Visual Scene Detection & Slide OCR Fusion
My multimodal video pipelines (Beyond Voice: Building Multimodal Video Transcription with AI Slide Detection & Visual OCR) detect slide changes via perceptual hashing and merge visual OCR with spoken words into a single searchable knowledge base.