Skip to content

Audio deduplication is keyed only by seq, causing cross-sender drops #16

Description

@airadier

Summary

audio_pipe_receive() deduplicates packets using only the 16-bit seq value. In a real group, multiple devices can legitimately emit the same sequence number.

Impact

Audio from one rider can be incorrectly dropped as a duplicate of another rider's packet if their sequence numbers collide inside the dedup window.

Evidence

  • Dedup window stores only uint16_t seq in main/audio_pipe.c:31-78
  • RX path checks only seq in main/audio_pipe.c:129-143
  • Packet format in main/group_mgr.h and docs/ARCHITECTURE.md has no sender identity for dedup purposes

Expected

Dedup should be scoped per sender/session, not globally per sequence number. That likely requires adding sender identity or session identity to the packet format and dedup key.

Notes

This is a functional issue for any multi-emitter broadcast mesh, not just an optimization.

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions