The moment a trainer or educator tries to play a video clip or audio-backed slide deck for a remote class, they discover their conferencing tool strips the system audio entirely — or the quality is so degraded that learners can't follow. The complaint appears verbatim across products: 'not able to upload videos with audio and sharing them with students,' 'Connect cannot share videos directly or system sounds from shared content; you must download the video and then upload it to the Meeting room,' and 'it is less straightforward to share video content.' This is a structural problem, not a bug. Conferencing vendors mix audio streams through their own codec pipeline, and system audio is an afterthought bolted onto a voice-first architecture. The buyer is the training team or L&D department — they sign the conferencing contract — but the users are instructors who suffer the failure live in front of a class and have no escalation path that matters to the vendor.

What makes this worse is that the workaround vendors suggest — download the video, re-upload it to a media pod, play it from there — is fragile, slow, and breaks spontaneous teaching moments. You can't just drop in a YouTube clip mid-lesson. And when the workaround fails, the instructor loses the room.

The need recurs every single training session where media is involved. Corporate L&D teams run dozens of live virtual instructor-led sessions per month. Each broken audio moment is a direct hit to learner satisfaction scores, which are tracked and reported upward. That creates a budget owner with a measurable, recurring pain — not someone hoping for a nice-to-have.

What to build

Build a lightweight desktop process that intercepts system audio output, re-encodes it as a virtual microphone input, and routes it through any conferencing tool's existing audio channel — with a browser extension that lets the instructor mark which audio source is 'media' so it gets normalized separately from voice, preventing ducking or clipping.

Where to start

Start with Adobe Connect customers specifically — they are the most vocal about this exact gap ('Connect cannot share system sounds') and their users are disproportionately enterprise L&D teams who already have a training budget line and are actively searching for workarounds in forums.

The hard part

Virtual audio device drivers behave differently on Mac versus Windows and can conflict with existing audio software; getting reliable, low-latency audio routing on both OSes without requiring admin privileges is technically fiddly and the first demo will likely fail on at least one common hardware configuration.

How it makes money

Per-seat annual license billed to the L&D team, priced below what an hour of wasted instructor time costs — roughly $8-15 per instructor per month, with volume tiers for teams over 20.

See the evidence. The complaints behind this idea, the products they came from, and similar ideas in Video Conferencing.

More ideas in Video Conferencing