Feature Request: Support direct upload of audio/video files for transcription and summarization

[wchemz]

Is your feature request related to a problem? Please describe.
Currently, LMA (Live Meeting Assistant) only supports real-time meeting transcription via Chrome extension, audio streaming, or virtual participant. There is no way to directly upload pre-recorded audio or video files for transcription. Users who want to transcribe a past meeting must use a workaround — playing the recording in a separate browser tab and streaming the audio to LMA in real time. This means the user has to sit through the entire recording duration, defeating the purpose of automation.

Describe the solution you'd like
Add an upload feature to the LMA portal that allows users to:

• Upload a pre-recorded audio file (and eventually video) directly through the UI.
• Provide meeting metadata (title, date, participants, etc.) at upload time.
• Have the uploaded recording processed asynchronously — transcribed, summarized, and displayed in the meeting list alongside live meetings.
• Support common audio formats (e.g., MP3, WAV, M4A) and video formats (e.g., MP4, WebM) on the roadmap.

Describe alternatives you've considered

• Streaming workaround: Play the recording in a separate browser tab and use the LMA audio streaming feature to capture it. This works but requires the user to be present for the full duration of the recording and doesn't allow skipping irrelevant sections easily.
• Direct Amazon Transcribe: Users could upload audio files directly to Amazon Transcribe, but this bypasses the LMA portal entirely — losing the UI, summary templates, sharing, search, and meeting management features.
• Custom development: Fork the LMA repo and build the upload feature independently using Kiro or a development partner, but this adds maintenance burden and diverges from the upstream project.

Additional context
N/A

[admin-fsw]

Good request. Tools like livesuggest.ai show there's demand for lightweight alternatives that skip the heavy infrastructure and just work in the browser with system audio, no upload flow needed. Would be useful to see more options in that direction here too.