troubleshooting & file map

Permissions checklists, firewall rules, where everything lives on disk, and what talks to the network.

permissions checklist

macOS

Microphone — recording.
System Audio Recording Only — for capturing system audio; prompted on first use, needs macOS 15+.
Local Network — finding your phone and syncing over the wifi.

Windows / Linux — microphone permission where the OS enforces one; system-audio capture needs no extra permission.

Phone — see the phone permissions table.

firewall

If phone pairing can't find the desktop (or vice versa), a firewall is the usual suspect. yapyap needs, on your local network only:

TCP 42700 inbound on the desktop — pairing and recording transfer.
UDP 5353 — how phone and desktop find each other.

The built-in language-model engine only ever runs locally — it's never reachable from the network.

what talks to the internet

For an app this local, the outbound list is short and fully enumerable:

destination	when	what for
model CDNs (incl. Hugging Face)	model download/first use	speech & enrichment models
`github.com`	update checks; first AI start	signed app updates; the built-in engine (a small one-off download)
`api.polar.sh`	activation & background licence re-checks	license
your BYO provider	only if you configured one	lens/enrichment requests you routed there

Recordings, transcripts, voiceprints, search indexes: never. Phone sync is LAN-only, end to end.

where your data lives

Everything sits in one per-user app folder:

OS	path
macOS	`~/Library/Application Support/com.xiduzo.yapyap/`
Windows	`%APPDATA%\com.xiduzo.yapyap\`
Linux	`~/.local/share/com.xiduzo.yapyap/`

Inside:

recordings/ — your audio files.
yapyap.db — one database file: transcripts, speakers, lenses, search index. Copy these two and you've backed up everything that matters.
models/ — your speech and language models.
bin/ — the built-in engine.

Backup recipe: quit yapyap, copy recordings/ + yapyap.db. Models and the engine re-download themselves; your data doesn't.

the log

yapyap studio writes what it does to a log file on your computer. Attach it when you report a problem. See find the log for where it is on each OS.

common hiccups

"System audio capture is unavailable" — OS too old (macOS needs 15+), or not available on some Linux systems. Mic recording is unaffected.
Antivirus quarantined something (Windows) — the freshly downloaded built-in engine under bin/ is a common false positive. Restore or allow it; it comes from an official source.
Summaries/lenses stopped, transcription fine — the language-model engine is down; check settings → compute and local AI troubleshooting.
Phone says "desktop offline" — desktop asleep, different wifi, or firewall (see above). Recordings wait on the phone; nothing is lost.
A processing step failed on one recording — hit the inline retry on the step; failures are per-step and re-runnable.
AppImage crashes on open / "looks corrupt" (Linux) — the download is almost certainly fine; run it from a terminal to see the real error. A message like version 'GLIBC_2.39' not found means the distribution is older than yapyap supports (Ubuntu 24.04+, Debian 13+, Fedora 40+ — see system requirements); the .deb/.rpm will refuse for the same reason. A FUSE/fusermount error means the fuse3 package is missing (rare — every supported distro ships it): install it, or run ./yapyap*.AppImage --appimage-extract-and-run.
Blank white window or instant crash at launch (Linux, usually Nvidia proprietary drivers) — a WebKitGTK graphics issue, not yapyap data. Launch with WEBKIT_DISABLE_DMABUF_RENDERER=1 ./yapyap*.AppImage; if that cures it, add the variable to your desktop entry.
App closes instantly at launch or during onboarding, no error (Windows/Linux) — versions before v0.6.8 required a CPU with modern vector instructions and a working Vulkan driver; older processors, basic display adapters, and some virtual machines crashed silently. Fixed in v0.6.8 — install the latest from yap-yap.app. See system requirements.