Platform
Screen + webcam face, captured simultaneously
FFmpeg pulls both streams off mediasoup PlainTransport into a single split-screen MP4 stored on your MinIO or S3. Configurable retention, encryption at rest, signed-URL playback only.
1080p
Default recording resolution
AES-256
Encryption at rest
Signed URLs
Playback never public
0 browser CPU
Server-side recording only
Each session's screen producer and webcam producer are routed to dedicated PlainTransport endpoints inside the mediasoup worker. A separate recorder process (Python + FFmpeg) consumes the RTP, transcodes to H.264 + AAC, composites the side-by-side or PiP layout, and writes the final MP4 to your object store. The agent's browser does no recording work — every byte is server-side.
Choose MinIO for fully on-prem, AWS S3 Mumbai, GCP Cloud Storage Mumbai, or any S3-compatible target. Configure retention by queue, client, or session type. Lifecycle policies move recordings to cold storage and delete them on schedule.
Every recording is indexed by session ID, agent, queue, customer ID (if provided), tags, and QA score. Find any call in seconds. Recordings are linked from QA scorecards, agent dashboards, and the supervisor floor view.
QA reviewers click a recording link and play it back inline — no download, no separate Windows Media Player workflow. Speed controls (0.5×, 1×, 1.5×, 2×), waveform scrubber, and bookmark-anchored comments make review fast.
No. Recording happens entirely on a separate FFmpeg recorder host that consumes a forwarded copy of the RTP from the SFU. The agent's machine sees no extra CPU or bandwidth.
Yes — per-queue config. Audio-only is roughly 5× smaller and skips the FFmpeg compositor.
An automated job runs daily, scans recordings past retention, and deletes them via the S3 API. The deletion is logged in the audit trail and includes a cryptographic proof-of-deletion when using MinIO.