RemSound — user manual

RemSound is a Windows program for sending live audio between two or more computers with very low delay. You can think of it as a private, peer-to-peer audio link: each machine picks what it wants to capture and what it wants to play out, and audio flows directly between them over UDP.

It was built for music collaboration over the internet (a guitarist on one machine, a vocalist on another, listening to each other in real time), but works just as well for low-latency monitoring across rooms in a house, podcast co-hosting, or any other “send audio from this PC to that PC, fast” use case.

Table of contents

1. What RemSound does
2. Quick start
3. Profiles
4. The main window: menu bar + three tabs
5. Menus (File, Record (Alt+O), and Help)
6. Connectivity tab
7. Audio inputs and outputs tab
8. Audio profile tab
9. ASIO and WASAPI
10. Peers — finding and connecting
11. How the network works (LAN, WAN, Tailscale)
12. Latency and audio quality
13. Keyboard shortcuts
14. Global hotkeys (mute, volume, tray)
15. Remote control: adjusting a peer's listening volume from your end
16. Startup behaviour
17. Updating RemSound
18. Recording to a file
19. Logs and diagnostics
20. Troubleshooting
21. Glossary

1. What RemSound does

RemSound moves audio from one PC’s microphone or sound card to another PC’s speakers or audio interface, in close-to-real-time, over your network. It’s symmetric: each side runs the same program and chooses independently whether to send, receive, or both.

The basic flow

There is no central server, no account, nothing in the cloud. Packets go straight from one machine to the other.

2. Quick start

Assume you and a friend both have RemSound running and a way to reach each other on the network (same Wi-Fi, same Tailscale account, etc.).

1. Launch RemSound. The first thing you’ll see is the profile picker. On a fresh install your only option is (Blank template) — select it and press Enter or click OK. Once you’ve saved a profile or two later on, this dialog is how you pick which one to load. See Profiles for the full story.
2. Once the main window opens, use Ctrl+Tab to move to the Audio inputs and outputs tab. Tick Receive audio (Alt+R), then tick the device you want incoming audio played through in WASAPI outputs for received sound (Alt+3).
3. On the same tab, tick Send my audio (Alt+S) and tick your microphone in WASAPI inputs to send (Alt+5).
4. Move to the Connectivity tab (Ctrl+Shift+Tab to go back left, or Ctrl+Tab from elsewhere). Find your friend in the Discovered peers (Alt+D) list and tick them. Or, if they aren’t appearing, click Add peer by IP (Alt+A) and type their address.
5. Have your friend do the same with you on their machine.
6. Within a second or two, both sides will hear each other.
7. Open the File menu (Alt+F) and choose Save as to give your setup a name. Next time you launch, picking that profile from the startup dialog restores all your settings, device ticks, peers, and connections in one go.

3. Profiles

RemSound saves your entire setup — device ticks, send / receive states, codec, packet size, smoothness, latency targets, hotkeys, ASIO driver choice, remembered peers, currently-connected peers — into one JSON file per profile. You pick which profile to load every time you launch the program. Tick a different combination of devices for “morning podcast” vs “evening jam session” and switch between them in two clicks.

The startup picker

When RemSound launches, the first thing you see is the profile selection dialog. It’s a listbox of profile names plus a synthetic (Blank template) entry at the top. The keys are deliberate:

The first profile in the list is highlighted by default, so a fresh install where you only have (Blank template) is just Enter to proceed.

What “Blank template” means

Blank template is a one-shot session with all defaults: nothing ticked in any device list, neither Receive nor Send checked, default codec, no ASIO driver selected, no remembered peers, default hotkeys. You’d pick it for a one-off session you don’t intend to save, or as the starting point for a new profile. The Save and Update button is intentionally hidden when on Blank template — there’s nothing to update, only to save-as-new.

Saving and updating

The File menu has two save items:

The window title bar shows which profile is currently active: RemSound — Active profile: My session name. NVDA reads this on Alt+Tab.

Switching, renaming, and deleting

Where profiles live on disk

One JSON file per profile. By default, stored at:

<RemSound folder>\profiles\<your machine name>\<profile name>.json

The per-machine subfolder keeps each computer’s profiles separate. If you used the Browse… button on the startup dialog to pick a different folder (e.g. inside Dropbox), profiles are stored directly under that folder — no per-machine subfolder, so two machines pointing at the same shared folder see exactly the same list.

You can also copy a profile JSON between machines: drop it into the other machine’s profile folder and it appears in their startup dialog. If the new machine doesn’t have the same hardware (different sound cards, different ASIO drivers), those device ticks are silently skipped at load time — RemSound won’t error or warn, the relevant lists just won’t have those entries ticked.

What’s NOT in profiles

A short list of things that intentionally aren’t profile-saved:

• The folder profiles are read from. Picked via the Browse… button on the startup dialog and saved in a small machine-local remsound.config.json next to the executable.
• Connection health / heartbeat RTTs — runtime telemetry, not setup.
• Auto-tune learning state — recomputed from scratch each session.
• Window position and size — Windows handles these via standard window-manager memory.

4. The main window: menu bar + three tabs

The main window has three things stacked vertically:

1. A menu bar at the top with two menus — File (profile management, preferences, keyboard shortcuts, minimise to tray, exit) and Help (open the user manual, check for updates, About). See Menus.
2. A tab strip with three tabs — Connectivity, Audio inputs and outputs, Audio profile. Use Ctrl+Tab and Ctrl+Shift+Tab to move between them, or arrow left/right when focus is on the tab strip. Each tab has its own Alt+letter shortcuts that fire only when that tab is the visible one — the same letter can mean different things on different tabs without conflict.
3. A status footer updating once a second with connect uptime, peer count, send/receive activity, and heartbeat health.

5. Menus (File, Record, and Help)

Three top-level menus on the main window: File (Alt+F), Record (Alt+O), and Help (Alt+H). The Record menu uses Alt+O rather than Alt+R so it doesn’t clash with the Receive audio (Alt+R) checkbox on the main form.

File menu

Press Alt+F to open the menu, then the underlined letter for the item you want, or use the single-press shortcut from the table below.

Record menu

Press Alt+O to open the menu, then the underlined letter for the item you want. (Alt+O rather than Alt+R because Alt+R is taken on the main form by the Receive audio checkbox.) The recording feature taps the audio path at the right place to capture either what you’re sending, what you’re receiving, or both, and writes it to disk as a WAV, MP3, OGG-Opus, or FLAC file. See Recording to a file for the full chapter; this is the menu summary.

Help menu

Press Alt+H to open the menu, then the underlined letter for the item you want.

6. Connectivity tab

This is where peers and logging live. Tab order on this tab:

7. Audio inputs and outputs tab

This tab controls everything to do with which audio devices are involved. The ASIO driver picker at the top decides whether ASIO is in the pipeline at all; the Receive and Send sides each have their own master checkbox and device lists.

All device lists are checkable list boxes — arrow up and down to browse, spacebar to toggle a tick. Profiles save device ticks; Blank template starts with everything unticked.

Receiving

Receiving has two requirements: Receive audio ticked, and at least one output device ticked. Without an output device, even if packets arrive there’s nowhere for them to go.

Tick as many outputs as you like across both the WASAPI and ASIO output lists — the same received audio plays out of all of them. Common combinations:

• One output: just your monitors / headphones.
• Studio monitors via ASIO + a wireless headset via WASAPI for moving around the house.
• Two physical outputs for left/right room monitoring.

Sending

Sending requires Send my audio ticked, plus at least one capture source ticked across the three send lists.

Tick any combination across the three lists. RemSound mixes them into a single stream and sends that to all selected peers. So you can send a mic plus a guitar plus your system audio at once, mixed together, and your friends hear all three.

8. Audio profile tab

Everything that shapes the audio quality / latency trade-off lives here. The first thing you land on when you Tab into the tab is the Priority mode checkbox — kept ungrouped at the top because it's the single biggest knob for cold-start audio feel. Below that, two NVDA-announced groups: Audio send parameters first, Audio receive parameters below. As you tab through, NVDA announces which group you’ve entered.

Use CPU and Windows performance settings in high priority mode (Alt+U)

The first control on the tab. When ticked, RemSound asks Windows for high-priority treatment across the board, the entire time RemSound is open under this profile. Specifically:

• No automatic CPU downclock when the machine looks idle.
• No migration of audio threads onto efficiency cores on hybrid CPUs (12th-gen Intel and later).
• No deep CPU-sleep idle states.
• 1 ms scheduler quantum honoured even when other apps haven’t asked for it.
• High process priority class.
• Process memory priority asserted as normal, so Windows doesn’t demote our pages under memory pressure.
• Working-set minimum locked at 32 MB so the OS’s idle-time trimmer can’t page out our hot code and audio buffers.
• MMCSS Pro Audio thread priority bumped one notch from the default High to Critical on every hot thread (network listener, mix loop, render).

Net effect: the cold-start “first few seconds sound terrible, then it warms up” behaviour goes away. Nothing in the OS coasts while RemSound is sitting between bursts of activity.

It’s a per-profile setting, so you can have one profile for live sessions where this is on and another for casual background-listening profiles where it stays off. Toggling marks the profile as having unsaved changes; save the profile to keep the choice.

Cost on a desktop: a couple of watts more while RemSound is open. Cost on a laptop on battery: more aggressive battery drain over the session because the cores park in a higher power state instead of dropping to deep idle — RemSound’s actual CPU usage doesn’t change, the cores just don’t sleep. The override is automatically reversed when RemSound closes (or when you untick it), so leaving the app running with the box on for a session is fine — turning it off mid-session also works.

None of the other apps on your machine are affected. RemSound only asks Windows to keep itself at high priority; the OS keeps throttling everything else normally, so your screen reader, browser and background services still get the usual power-saving treatment.

Audio send parameters

Audio receive parameters

The shape of this section depends on whether an ASIO driver is selected on the Audio inputs and outputs tab. With no ASIO driver, you see one latency row (the WASAPI one, labelled simply “Audio latency”). With an ASIO driver selected, you see two latency rows — one per lane — each with its own auto-tune toggle. The two lanes are independent: a WASAPI underrun doesn’t pull the ASIO target up, and vice versa.

Most people only need to pick a codec and a smoothness level and let the rest sit on defaults.

9. ASIO and WASAPI

RemSound speaks two audio backends. Which one is in play depends on the ASIO driver (Alt+D) listbox at the top of the Audio inputs and outputs tab.

WASAPI (the default)

Windows’ standard audio engine. Every device in your Windows Sound control panel is reachable via WASAPI. Latency from a WASAPI capture or playback is typically 10–30 milliseconds. Anyone running RemSound has WASAPI; no special hardware needed.

ASIO (needs a driver)

A separate audio path used by professional audio interfaces. ASIO drivers bypass the Windows audio engine and talk directly to the hardware, giving sub-5-millisecond hardware latency.

The ASIO driver picker is omitted entirely on a machine with no ASIO drivers installed. Common drivers that do appear:

• Audient USB Audio ASIO Driver — for EVO 4 / 8 / 16, iD-series interfaces.
• Komplete Audio ASIO Driver — for Native Instruments interfaces.
• Focusrite USB ASIO — for Scarlett, Clarett, Red.
• RME ASIO — for Babyface, Fireface, UCX.
• Realtek ASIO — bundled with some Realtek drivers. Avoid this one — it’s a generic wrapper that can grab whatever Windows default device it finds, often clashing with screen reader audio.

How the driver picker chooses the pipeline

One control, two outcomes:

Default on a fresh install is (none). If you have an ASIO driver and want to use it, arrow down to it in the picker. To go back to WASAPI-only, arrow back up to (none).

ASIO channel pairs

ASIO doesn’t expose “devices” the way Windows does. Instead it gives you a list of channels (typically 2, 4, 6, 8, or more depending on the interface), grouped into stereo pairs. RemSound labels each pair with the driver name, the pair number, and the channel names the driver itself reports. For an Audient EVO 8 you’d see entries like:

Audient USB Audio ASIO Driver — Pair 1 (channels 1/2): Mic | Line | Instrument 1 / Mic | Line 2
Audient USB Audio ASIO Driver — Pair 2 (channels 3/4): Mic | Line 3 / Mic | Line 4
Audient USB Audio ASIO Driver — Pair 3 (channels 5/6): Loop-back 1 (L) / Loop-back 2 (R)

Buffer size for ASIO

RemSound doesn’t expose a buffer-size control of its own. To change ASIO buffer size, open the audio interface’s own control panel application (NI’s Komplete Audio Control Panel, Audient EVO Control software, etc.) and set it there. The driver remembers its buffer size between sessions; RemSound just uses whatever the driver is configured to do.

Same driver, sender and receiver, on one machine

RemSound supports this — you can capture from your audio interface and play received audio out of the same interface simultaneously, on the same machine. Most modern pro audio drivers handle this fine.

10. Peers — finding and connecting

“Peer” means another machine running RemSound that you want to talk to. Peers are managed on the Connectivity tab. The tab has three lists, all of them checkable:

Plus the Add peer by IP (Alt+A) button which opens a small prompt for a hostname or IP. Useful for first-time connection over a VPN where discovery hasn’t reached them yet.

Audio only plays from peers you’ve ticked

Even if a peer is sending audio in your direction, you won’t hear it until you’ve ticked their checkbox. This is deliberate — connecting is a consent step. A peer’s name shows up in Discovered the moment they come online, but they can’t make sound on your speakers until you say yes.

Heartbeat indicator

For each connected peer, the status read-out at the bottom of the window shows a small status: their latest round-trip time in milliseconds, or pending, stale, or unreachable if heartbeat replies have stopped. RemSound plays a connect cue (a short sound) when a peer transitions to healthy and a disconnect cue when one becomes unreachable. Both cues can be silenced via Mute connect/disconnect sounds in the Preferences dialog (File → Preferences, or Ctrl+P).

11. How the network works (LAN, WAN, Tailscale)

RemSound uses two UDP ports:

One canonical audio port number is used for everything — Tailscale, LAN peer-to-peer, and any relay server. You never need to type a port after a hostname or IP; the default is implied. Both sides of a connection need to use the same audio port number.

Heartbeat packets ride on the same UDP port as audio, so if your audio reaches the peer, your heartbeat does too — one firewall rule covers both.

Network packet priority (qWAVE / Voice / WMM)

RemSound asks Windows’ built-in Quality-of-Service service (qWAVE — Quality Windows Audio/Video Experience) to mark its outbound audio packets as Voice priority (DSCP 46, Expedited Forwarding). This happens automatically on every launch — there’s no toggle — and falls back silently if the service is unavailable (some stripped-down Windows images).

What it buys you, in order of how much you’ll actually notice:

• Wi-Fi: the DSCP marking maps to WMM Voice access category, the shortest medium-contention window the Wi-Fi standard defines. On a busy access point with other clients fighting for airtime (someone streaming, downloading, video-calling), RemSound’s packets get to transmit first. This is the biggest real-world win.
• Local NIC scheduling: the Windows network stack sends RemSound’s packets ahead of best-effort traffic from the same machine (browser downloads, OS background transfers, cloud-sync uploads). When the upstream is saturated, audio still goes out cleanly.
• Wired LAN switches and routers that honour DSCP: most consumer kit does, on the local side at least. The marking can give RemSound’s packets priority through the home network.

What it doesn’t buy you: priority across the public internet. Most ISPs strip or rewrite DSCP markings at the network edge, so the qWAVE bits rarely survive past your local ISP’s first hop. LAN and same-house Wi-Fi: tangible benefit. Across the internet: neutral — no harm, no help. RemSound also requests 1 MB kernel send and receive buffers on its UDP sockets so a momentary stall (GC, scheduler hiccup) up to roughly 30 ms doesn’t cause the kernel to drop datagrams on either side.

LAN — same Wi-Fi or Ethernet

On a normal home network, discovery and heartbeat both work without configuration. Launch RemSound on two machines and they’ll see each other within a second or two via the discovery broadcast. No firewall changes are usually needed because UDP broadcast is allowed by default.

WAN — different physical locations

Direct internet peer-to-peer connections require either:

• A VPN that puts both machines on the same virtual network — Tailscale is the recommended option. Each machine gets a Tailscale IP (in the 100.x.y.z range) and they can reach each other directly.
• Port forwarding on each end’s router (more involved; not covered here).

Discovery on Tailscale and other VPNs

UDP broadcast doesn’t traverse a VPN — Tailscale’s virtual interface ignores broadcast packets entirely. RemSound works around this by also sending discovery announcements directly (unicast) to every IP in your Remembered peers list. So:

1. One time only: each side adds the other’s Tailscale IP to its Remembered peers list (via the “Add peer by IP” button).
2. From then on, RemSound auto-unicasts announcements at those IPs every 1.5 seconds.
3. The receiving side’s discovery hears the announcement, adds the sender to its own unicast list, and announces back.
4. Within seconds, both sides see each other in Discovered peers without further typing.

So the rule is: only one side has to type the other’s IP once. After that, discovery is bidirectional automatically.

Round-trip time and what it means

12. Latency and audio quality

Five controls together shape the latency / audio-quality trade-off, all on the Audio profile tab:

• Audio latency in milliseconds (Alt+L) — the headline target buffer size at the receiver.
• Buffer smoothness (Alt+B) — how aggressively the receiver protects against jitter spikes.
• Packet size (Alt+P) — Standard or Small. Smaller packets shave a couple of milliseconds off send-side delay at the cost of doubling packet rate.
• Lock to audio clock (Alt+K) — ties packet emission to the audio device’s hardware clock end-to-end, eliminating Windows scheduler jitter.
• Continuous auto-tune — lets the receiver choose the latency target for you, every few seconds.

Plus the codec choice (PCM / Opus 20 ms / Opus 10 ms), also on the Audio profile tab. Most people only need to pick a codec and a smoothness level and let the rest sit on defaults.

Audio latency slider

The Audio latency spinner tells the receiver how much audio to keep buffered as a cushion against network jitter. Bigger buffer = more delay, fewer clicks. Smaller buffer = less delay, more clicks when the network wobbles.

Smaller is better when the network can handle it. If you don’t want to think about this number, switch on continuous auto-tune (below) and leave it.

Buffer smoothness

The Buffer smoothness listbox sets a 1-to-10 scale that controls how patient the receiver is when network jitter spikes. The default is 3.

Smoothness and the Audio latency spinner work together — smoothness controls how the receiver responds when audio runs late; the latency value controls how big a head-start it builds. Real-world tip: if you can hear clicks, try raising smoothness by one or two before reaching for a bigger latency target.

Packet size — Standard or Small

Two options: Standard (default) and Small. Controls the frame size each network packet carries:

The latency win is small — at most a few milliseconds end-to-end. Small packets are useful when you and your collaborator are on the same LAN and want to chase every last millisecond; for any internet path it’s a false economy because doubling the packet rate also doubles your odds of running into jitter at exactly the wrong moment, which translates into clicks.

Lock to audio clock

The Lock to audio clock checkbox locks RemSound’s send timing path to the audio device’s own hardware clock instead of letting Windows’ general thread scheduler set the pace. Off by default. The label tells you what it does in your current setup:

Continuous auto-tune

The Continuous auto-tune latency checkbox hands the latency value over to RemSound itself. When on, RemSound observes incoming-packet jitter every few seconds and nudges the latency target up if it’s seeing late packets, or down if the network has been calm. The companion Auto-tune latency interval (Alt+I) combo sets how often it re-evaluates — 3, 5, 10, 15, or 30 seconds. Faster values respond quickly to a network change but can also feel “twitchy”. Treat continuous auto-tune as a hands-off way to keep the cushion right-sized as your network mood changes.

If you turn auto-tune off, the latency value stays wherever it last was.

Artefact sound type

When the playback buffer comes up empty for a moment (an underrun), RemSound has to fill the gap. The Artefact sound type listbox decides what that gap sounds like:

• Noise burst (default) — a brief broadband shhh that blends into music. Easy on the ear; tells you something happened without being jarring.
• Click — raw zero-fill, no concealment. Produces an obvious click on every underrun. Use this when you want to hear every problem (e.g. when tuning latency lower).

Opus packet-loss recovery (built in, no setting)

Both Opus modes ship with built-in forward error correction: each packet carries a small redundant copy of the previous packet’s audio, and the receiver uses it to reconstruct any single packet that goes missing in transit. The result is that single-packet losses become inaudible — no click, no glitch — instead of the small pop you’d otherwise hear. Two-packet-in-a-row losses still produce one click; that’s a fundamental Opus limitation, not something configurable.

This is automatic — there’s no switch. PCM mode does not have it.

Codec choice

PCM gives the absolute best audio with no encoding artefacts but uses about 30× the bandwidth of Opus. Over the open internet, Opus is almost always the right call.

13. Keyboard shortcuts (within the main window)

Each tab has its own Alt+letter shortcuts. The same letter can mean different things on different tabs without conflict — mnemonics fire only on the active tab. Move between tabs with Ctrl+Tab and Ctrl+Shift+Tab.

Connectivity tab

(Logging controls — Enable logs and Write logs now — live in the Preferences dialog; reach them via File → Preferences or Ctrl+P, then Alt+L / Alt+W within the dialog.)

Audio inputs and outputs tab

Audio profile tab

Some of these shortcuts shift depending on whether an ASIO driver is selected. When one is selected the ASIO-lane controls take the simpler Alt+L / Alt+T mnemonics; the WASAPI-lane controls move to Alt+W / Alt+Y so they don’t collide.

File menu shortcuts (work from any tab)

Always-available

14. Global hotkeys (work even when minimised)

Configurable in the Keyboard shortcuts dialog (Ctrl+K, or File → Keyboard shortcuts). The dialog is a single list of every bindable shortcut: arrow up and down to move between rows, press Enter to rebind the highlighted row, press Del to clear it (back to not set), press Escape (or Tab to the Close button) to close. Defaults:

You can change any of these to whatever combination you prefer. Each accepts modifiers (Ctrl, Shift, Alt) plus one regular key.

15. Remote control: adjusting a peer's listening volume from your end

The use case: you're on your laptop, listening to audio coming from your desktop, and you've got NVDA Remote open so you can drive the desktop with your laptop's keyboard. Every keystroke you make goes to the desktop — including any local volume hotkey on the laptop, which now never reaches the laptop. There's no way from inside that NVDA Remote session to nudge the laptop's listening volume without breaking out of the session.

RemSound's remote control feature gives you a way: configure a hotkey on the desktop (the machine your keyboard is talking to) that sends a command across the audio link telling the laptop's RemSound to raise / lower / mute its own listening volume. You stay in NVDA Remote; the laptop responds.

Two flavours of remote command

You get two independent sets of remote-control hotkeys, both controlled by the same opt-in toggle on the receiver. Pick whichever fits the situation, or bind both:

Both sets target the receiver. Neither changes anything on the sending machine.

How to set it up

1. On the machine that should respond to remote commands (the one you're listening on — the laptop in the example): open Preferences (Ctrl+P) and tick Accept remote volume commands from peers. Save the profile (Ctrl+S) so the choice sticks. (One toggle covers both flavours of remote command.)
2. On the machine that should send remote commands (the one your keyboard is driving — the desktop in the example): open the Keyboard shortcuts dialog (Ctrl+K). Bind whichever of the six remote-control rows you want:
   • Send remote volume up / down to peers — nudges the receiver's RemSound app slider.
   • Send remote receive mute toggle to peers — toggles the receiver's RemSound mute.
   • Send Windows global volume up / down to peers — nudges the receiver's Windows master volume.
   • Send Windows global mute toggle to peers — toggles the receiver's Windows mute.
   Bind whatever key combinations you prefer (e.g. Ctrl+Shift+Up / Ctrl+Shift+Down for one set, Ctrl+Alt+Up / Ctrl+Alt+Down for the other). These are global hotkeys: they fire as long as RemSound is running, no matter which app has focus.
3. That's it. Press the configured hotkey on the desktop — the laptop responds the way it would if you'd pressed the corresponding key on the laptop directly, and you hear the change without leaving the NVDA Remote session. Hold the Windows-volume hotkey down for a steady ramp; the OS auto-repeat fires the step over and over.

Mutuality

The feature is symmetric: both machines can both send and accept. If you configure hotkeys on both ends and tick “Accept remote volume commands” on both ends, either side can drive the other's volume. There is no “controller” / “controllee” designation.

What it does not touch

• It changes RemSound's receive volume slider on the target machine, not Windows system volume. So your peer can't accidentally turn down a video call's audio or your screen reader.
• It only works between peers who are already connected (i.e. each has the other ticked in their connected-peers list). The audio allow-list is the gate — an unticked peer can't drive your volume.
• The opt-in tick is per-profile, so a setup you've designated as “trusted home pair” can have it on while a one-off jam-session profile keeps it off.
• Remote commands travel on the same UDP audio port as audio and heartbeat (47830 by default), so there's no extra firewall rule to add.

16. Startup behaviour

Open the Startup behaviour dialog from File → Preferences → Startup behaviour (or Ctrl+P, then Alt+S). It has three independent toggles, plus a profile picker that appears when the third one is on. Tab cycles between the three checkboxes, the profile list (when visible), and the Close button. Esc closes the dialog. Each tick is saved immediately — there's no OK/Apply button.

Profile to start with (Alt+L) — the listbox of saved profiles. Visible only when the third toggle is on. Arrow up/down to pick, the choice is saved as you move; double-click to pick and close the dialog at the same time.

Combining the three for a hands-off boot

1. Save a profile with the device ticks, peers, and audio settings you want for “always-on” use.
2. Open Startup behaviour. Tick all three: Start minimised, Start automatically when this user logs in, Start with a specific profile — and pick the profile you just saved.
3. Close the dialog. Reboot or log out and back in to test — RemSound launches itself, loads the profile, and goes straight to the tray. Audio starts flowing as soon as the peer is reachable.

17. Updating RemSound

RemSound can check its own GitHub Releases page for a newer build on a schedule you choose, prompt you to install, and either ask first or do it silently. There’s also a one-press “check now” button so you don’t have to wait for the timer.

Settings in Preferences

Open File → Preferences (or Ctrl+P). The update settings sit above the logging row:

What happens during an install

RemSound can’t overwrite its own .exe while it’s running, so an install spawns a small one-shot helper that finishes the job after RemSound exits:

1. RemSound downloads the new release’s RemSound-vX.Y.zip from GitHub into a staging folder next to the running .exe.
2. It writes a tiny detached _apply-update.cmd alongside it that watches the RemSound process.
3. RemSound exits.
4. The detached cmd notices the exit, copies the staged files over the install folder, deletes the staging folder, and relaunches RemSound.exe.
5. The cmd deletes itself.

You’ll see the window close, then reopen on the new version within a second or two. Anything that was unsaved on the old session (a profile in mid-edit, for instance) is lost — RemSound will not auto-save before installing. Save first if you’ve been editing.

If install fails

The downloader is best-effort: a flaky network, a locked install folder, or a temporarily-unavailable release will pop a MessageBox saying it couldn’t complete and leave the running version untouched. The release page URL is in the popup so you can grab the zip in a browser and install by hand if needed. If you installed RemSound into Program Files without granting your user write permission to the install folder, the install helper’s file-copy step will fail too — either fix the permission or move RemSound to a writable folder (e.g. somewhere under your user profile).

The About dialog and release notes

To see what version you’re currently on without checking for updates, open Help → About RemSound (Alt+H, A). The dialog shows the version number and the release notes for the build you’re running, in a scrollable read-only text box you can tab into and arrow through under NVDA. Close (or Esc) dismisses.

18. Recording to a file

RemSound can write the audio passing through it to a file on disk — useful for keeping a copy of a music session, capturing a long jam for later editing, or just saving a one-off voice exchange you want to refer back to.

What gets recorded

Recording taps the audio path at fully-mixed, fully-processed points: for the received side, after volume / mute / limiter have been applied (so the file mirrors what you hear); for the sent side, the raw captured float audio just before it’s encoded for the wire (so the file is independent of which codec you chose). The three source options:

• Record all received audio — the full mix of everything coming in from connected peers. Default.
• Record all sent audio — what your microphones, loopbacks and ASIO inputs are sending out. Useful for confirming what your collaborators are actually hearing from your end.
• Record both sent and received audio — a single file containing both directions, soft-mixed and limiter-protected. The right pick for capturing a complete two-way exchange.

File formats

All four formats record at 48 kHz — that’s the engine’s internal mix rate — and labels in the attributes list make the rate explicit so it’s never ambiguous.

Crash resilience. All four formats are designed to leave a playable file behind even if RemSound crashes mid-recording. WAV re-patches its RIFF / data-chunk size fields every 5 seconds; MP3, OGG-Opus, and FLAC all use stream formats where every frame is self-decoding, so a truncated file is automatically a valid shorter file of the same format. You lose at most ~5 seconds of recently-captured audio on a crash, never the whole session.

Start / stop sound cues

RemSound plays a short ding when a recording starts and another when it stops, so you have an audible confirmation that the toggle actually took effect. The cues are record start.wav and record stop.wav in the RemSound install folder — you can drop your own files in to customise them, as long as they keep the same filenames.

Each cue can be turned off independently via Preferences → Cue sounds (Alt+N). The same checklist also controls the older connect / disconnect cues. The cues play through the default Windows output device using System.Media.SoundPlayer, which sits outside the recording pipeline — so a normal RemSound recording does not contain the cue. The exception is if your sender side is using WASAPI loopback of the very same output device the cue plays to as one of its capture sources, in which case the loopback grabs the cue alongside the rest of the device’s output; in that case you’ll have to live with the start cue appearing in the file unless you disable it.

Where recordings go

By default, recordings live in <RemSound install folder>\recordings\<machine name>\. Each recording session creates a new file named RemSound-yyyy-MM-dd_HH-mm-ss.<ext>, so files never collide.

You can change the folder via Record → Change recordings folder. Picking a different folder writes that path to your current profile, so the choice rides along with the rest of your settings — switching profiles can switch recording destinations too. If a saved profile points at a folder that doesn’t exist on the loading machine, the recorder silently falls back to the default location for the current machine.

Record → Open current recordings folder launches Windows Explorer on whatever the configured folder currently is, creating it on the fly if a recording hasn’t been made there yet.

Starting and stopping

Three ways to start or stop a recording:

• Ctrl+R from anywhere in the main window — toggle. The menu item text flips between “Start recording” and “Stop recording” so NVDA announces the current state when you walk into the menu.
• Record → Start recording (or Stop, when one is in progress).
• Closing RemSound while a recording is running cleanly finalises the file — you don’t lose data if you forget to stop manually.

The recorder runs on its own background thread with a generous queue, so neither the audio nor the network path is affected by disk speed. If the disk ever can’t keep up with the audio rate, the recorder drops the oldest queued audio (never the newest) and notes it in the log; in practice you’ll only see that on a fully-saturated USB stick or a very slow network drive.

Recording settings dialog

Reached via Record → Recording settings. Three keyboard-navigable listboxes:

OK saves the choices to the current profile. Cancel (or Esc) discards. Settings persist via the normal profile-save mechanism — changes here flag the profile as having unsaved changes and you’ll be prompted on exit if you haven’t saved.

19. Logs and diagnostics

If logging is enabled (the Enable logs checkbox in the Preferences dialog — File → Preferences, or Ctrl+P — default on), RemSound writes a tab-separated log file per session into publish\logs\ (relative to wherever RemSound.exe lives). One file per launch, named RemSound-<machine>-<PID>-<date>-<time>.log.

The file contains two kinds of rows:

The Write logs now button in the Preferences dialog (Alt+W within that dialog) writes a "user requested write logs now" marker into the log so you can find the moment in the file later.

Logs are plain UTF-8 text and openable in any spreadsheet (paste-as-tab-separated) or text editor. Most useful columns when something feels wrong:

• BufferMs — how much audio is queued for playback. Should sit close to your latency target.
• Underruns — count of moments the playback buffer ran dry. Each one is a tiny click. A few per minute is normal on WAN; hundreds per second means something is broken.
• Drops — packets discarded because the buffer overflowed. Should stay near zero in normal use.
• Heartbeat — RTT to each connected peer. pending / unreachable / stale indicate problems.
• OpusFecRecoveries — running total of single-packet drops that were silently filled in by Opus packet-loss recovery. A non-zero number that’s growing means FEC is saving you from clicks. Only meaningful when an Opus codec is in use.
• OpusUnrecoveredGaps — running total of multi-packet drops that FEC couldn’t repair. Each one corresponds to an audible click. Stays at 0 on a clean line; small numbers are normal on WAN.

20. Troubleshooting

I don’t hear my friend

1. Connectivity tab: is your friend in Connected peers with a green RTT (e.g. “192.168.1.5: 27 ms”), not unreachable or pending?
2. Audio inputs and outputs tab: is Receive audio ticked?
3. Same tab: is at least one output device ticked in the WASAPI or ASIO output list?
4. Same tab: is the volume slider above zero?

My friend doesn’t hear me

1. Audio inputs and outputs tab: is Send my audio ticked?
2. Same tab: is at least one capture source ticked across the three send lists?
3. If you’re using a microphone: is Windows’ microphone privacy setting allowing apps to use it? (Settings → Privacy → Microphone.)
4. Have they ticked your name in their Discovered peers list?

I can hear them but the audio crackles

• First, try raising Buffer smoothness by 1 or 2 on the Audio profile tab (Alt+B). It usually fixes crackles for a smaller delay cost than bumping the latency.
• Then, try raising the Audio latency value (Alt+L). WAN connections often need 30–80 ms.
• If using PCM, switch to Opus — Opus has built-in single-packet-loss recovery that PCM doesn’t. Much more tolerant of network instability.
• If on Wi-Fi, try wired Ethernet — Wi-Fi adds 5–20 ms of unpredictable jitter.
• If you previously set Packet size to “Small (LAN only)” but you’re now on the internet rather than the same-house LAN, switch it back to Standard. Small packets double your packet rate, which makes WAN clicks more likely.

Audio is fine but the delay feels long

• Lower the relevant latency value on the Audio profile tab in steps until clicks just start, then nudge back up by one step. With an ASIO driver selected this is two independent knobs (one per lane).
• Or, turn on Continuous auto-tune for the lane that feels slow and let RemSound find the right level.
• Drop Buffer smoothness down towards 3 if you’d been running it high “just in case”.
• If only WASAPI matters for your session, set the ASIO driver picker to (none) — that disables ASIO entirely and removes its lane from the pipeline.

The ASIO lane sounds grainy or constantly micro-clicks

Most likely your ASIO latency target is below the platform’s real-world jitter floor. The receiver fights to hold the buffer at the target by drift-dropping frames and click-trimming — both are audible. Raise ASIO latency in milliseconds (Alt+L) on the Audio profile tab to 25 ms or more and the grain should vanish. Even pure-ASIO setups don’t safely sustain sub-15-ms receive buffers over real networks; aim higher on Wi-Fi.

I can’t see my friend in Discovered peers

• If on the same LAN: are both machines on the same Wi-Fi or Ethernet segment? Some guest networks isolate clients from each other.
• If on Tailscale: type their Tailscale IP into “Add peer by IP” (Connectivity tab, Alt+A) once. After that, both sides see each other automatically.
• Check Windows Firewall isn’t blocking RemSound. The first launch usually prompts; if you said no, you’ll need to allow it manually.

One side says “unreachable” even though audio is flowing

Heartbeat shares the audio port, so if audio gets through, heartbeat should too. If one side is showing “unreachable” while audio plays, make sure both machines are running the same RemSound build — an older version on either end can speak a different heartbeat dialect.

NVDA went silent or crackly when I selected an ASIO driver

You probably picked Realtek ASIO. It’s a generic wrapper, not bound to Realtek hardware, and tends to grab whatever Windows’ default audio device is — which is usually the same one NVDA is using. Arrow back to (none) in the ASIO driver picker (or pick a different ASIO driver).

Audio device list shows old devices that are no longer plugged in

RemSound refreshes its lists every second. If a device has truly been unplugged it should disappear within a few seconds. If it’s lingering, restart RemSound — Windows’ device cache occasionally needs poking.

21. Glossary