self-hosted SIP/SIPREC capture with audio reconstruction and a NOC copilot

Commercial
1

Hi everyone,

I’m Moy Amiga, the founder of VoxyWatch — posting in Commercial since it’s a
commercial product, though there’s a free tier I’ll mention below.

Short version: VoxyWatch is a self-hosted SIP/SIPREC capture and monitoring
portal. It ingests HEP directly from your SBC/proxy (Asterisk via res_hep,
plus Kamailio, OpenSIPS, FreeSWITCH, and most commercial SBCs), so there’s
no SPAN port or network tap to set up.

What it does:

  • SIP ladder diagrams per call, with SDP/codec analysis
  • SIPREC audio reconstruction (stereo, caller/callee split) playable in the browser
  • Per-trunk quality metrics: ASR, NER, ACD, MOS, PDD, jitter, loss
  • CDR database (searchable/exportable) + per-call PCAP/WAV export
  • Per-trunk auto-baselines and rule-based anomaly detection
  • An optional AI copilot that explains likely root cause — bring your own
    LLM key; it only reads your capture, never touches your SBC
  • Embedded SNMP agent + traps for PRTG/Zabbix/Nagios

It runs entirely on your own server: no cloud, no call-home, no telemetry.
Two systemd services, ~60s install.

For sources that can’t emit HEP natively, there’s an optional lightweight
capture agent (voxywatch-probe, Go + libpcap, amd64/arm64) that sniffs the
NIC and forwards HEP to the portal — not needed for Asterisk, but handy in
mixed environments.

A note on trying it: the free tier (up to 50 concurrent calls, all features)
is already enough to evaluate everything end to end — no license needed.

If you run more than that and want to test at your real scale, I’m running a
small early-feedback program while we onboard the first wave: write me (forum
DM or contact@voxywatch.com, your 32-hex HWID handy) and I’ll set you up with
a full-capacity license at no cost, in exchange for honest feedback on what
works and what’s missing for an Asterisk shop. Limited slots.

Portal — install + docs: GitHub - VoxyWatch/publish · GitHub
Capture probe (optional): GitHub - VoxyWatch/voxywatch-probe · GitHub
Site: https://voxywatch.com

I built this after years operating voice networks, frustrated that HOMER OSS
doesn’t reconstruct audio and passive sniffers need taps. Happy to answer
technical questions here, and genuinely interested in feedback from people
running Asterisk in production — what’s missing, what would make it useful
for you.

Thanks,
Moy

2

For anything that is routinely capturing clear text RTP, or can otherwise recover audio, you need to build in reliable measures to block the original capture during payment card transactions (in particular, CVV exchange), in order for user to comply with PCIDSS. Otherwise you may force users into changing the way they handle such transactions. to avoid them being captured by your system.