Audio-sensitive learners blocked by music — transcripts + muteable captions

Executive Summary

Many learners and viewers are effectively blocked when spoken content is mixed with background music: language learners, hearing-impaired students, and busy professionals trying to study from recorded lectures or webinars struggle to read captions or parse transcripts when music masks speech. Institutions and creators — roughly 10 million potential customers globally — are under pressure to provide accurate captions and searchable transcripts for accessibility and pedagogical reasons, creating an addressable market of about $8.0B (10M x $800 ACV). A viable product would produce time‑aligned transcripts and captions with per-segment audio classification (speech vs. music), let users toggle or attenuate music in playback and captions, and provide a lightweight editor and API for human corrections and institutional workflows. Key features would include music-aware ASR, source-separation to preserve voice while reducing music, live and batch processing, and exportable, searchable transcripts; pricing and packaging would target creators, universities, and enterprise accessibility teams around the $800 ACV benchmark. This market is attractive now because modern ASR models are approaching human-level accuracy, regulations and enforcement around accessibility are increasing, and the volume of recorded, asynchronous learning continues to rise. The product could stand out by combining robust music detection and source separation with tight caption alignment and institutional integrations, but practical challenges include handling overlapping music and speech, potential legal/licensing complexity around modifying copyrighted music, and competing against established caption vendors — success will require investment in model fine-tuning, platform partnerships, and clear compliance guarantees.