Everything Onion Board does

Pads: add by button, drag-and-drop (files or folders), or by dropping files into the sounds folder (Sounds folder opens it; they join the board by themselves). Plays mp3, wav, ogg, flac, m4a and more, and pulls the audio out of video files. Search, reorder, resize, set colours. Pictures on pads: right-click → Add picture…, or drop or paste (Ctrl+V) a picture onto a pad; a link's thumbnail or a file's cover art is used by itself. Pads from videos have a Video button that shows the video in step with the sound (only you see it). Pick several pads with Ctrl+click, Shift+click or Ctrl+A to change their colour, volume, fades or categories, export them or remove them together (one Undo). From the keyboard: arrows move between pads, Enter or Space plays, F2 renames, Alt+Enter opens Edit, Delete removes (the picked pads, or the one you're on) and Ctrl+F jumps to the search box. Right-click a pad for Rename, Duplicate (a second pad with its own file, for its own effects or hotkey) and Open file location.
Record a sound with your mic, or from what's playing: Add → Record. Record your own voice, your voice through the voice changer while it's on, or What's playing: a bit of a sound, a web search result or the radio as you hear it (the window stays out of the way, so you can play it while it records); after Stop the quiet ends are already cut off, drag the start and end to cut more, Preview plays it in your headphones only, then name it and Save to add the pad (it joins the category showing). Up to 15 minutes, spooled to disk as it records; nobody hears you while you record.
Categories: the tabs above the pads (the + after the last tab makes one). Right-click a pad → Categories to put it in any number of them; right-click a category to rename, export or delete it. The in-game overlay shows the same category, and its R key (numpad *) switches to the next one. Right-click a category to give it a random-sound hotkey. With Use hotkeys per category (Settings → Hotkeys) each category is its own set of keys: one key, a different sound per category.
Remove can be undone: Removed “…” · Undo stays up for 10 seconds. After that the sound waits in Recently deleted (the Backup menu, or the bin button that shows while it holds sounds) for 30 days, and Bring back returns it exactly as it was. Only when it ages out does its audio file go to the Recycle Bin.
Backup / share (Backup button, or Settings → General): Export everything writes every sound (picture, effects, hotkey, categories) and your settings to one .zip; import it on a new PC (which audio devices and where sounds go stay per PC). A category or a single pad exports as a sound pack to share; importing skips sounds you already have, and asks before a pack's hotkeys take keys your sounds already use. Backup → My sound packs lists the packs you imported: reset one to how it came (its pads' edits undone, removed ones back) or remove it in one go. Import from another soundboard brings sounds over from Soundpad, Resanance, Soundux or EXP Soundboard. The format is a plain zip of JSON and the original audio files — see BACKUP-FORMAT.md.
Per sound: global hotkey (works in-game) or MIDI pad, volume, loop, what pressing does: Solo, the default (one sound at a time: it stops every other sound first), Overlap (layers over the others), Toggle, Queue (waits for the sounds playing to finish; right-click any pad → Play next does it once), Only others hear it (not in your headphones), a wait before it plays, a cooldown against spamming, fade in / fade out, and Hold to play (plays only while you hold its key or pad down, like an air horn).
Effects on any sound (right-click a pad → Edit → Effects): speed and pitch (separately, or together like a record player with Tape mode), trim (drag the start and end on the sound's waveform, or type exact times — keep one line out of a 4-minute video), a 7-band EQ, a boost up to +36 dB that clips on purpose, play backwards, and every voice effect (echo, reverb, distortion, radio, robot, add-on effects too). One-click presets: Deep fried, Bass boosted, Slowed + reverb, Nightcore, Chipmunk, Demon, Fast / Slow-mo (same pitch), Old radio, Reversed. Preview plays it to you only; Save changes that pad, Save a copy keeps the original and adds the edited version as its own pad. Pads with effects show FX; Reset goes back to the original. The original file is never changed.
Your mic on or off: send your voice with the sounds, or sounds only.
Where your sounds go (Setup → Devices → Send my sounds to, also in Settings → Audio): straight into your mic (the default: one click and one Windows permission prompt, then Discord and games hear your sounds through the mic they already use, with nothing to pick there; voice changer, mic volume, gate, ducking and mute all still work, and the Setup tab offers a one-click repair if a Windows update takes it off; after the first permission prompt, repairs and changes of mic don't ask again), the virtual cable (the backup: Use the virtual cable instead on the Setup tab, or the installer's box, unticked; Discord or the game uses its CABLE Output as a mic. Once your sounds are in your mic, the Setup tab offers to remove the cable), another device (any output but your headphones: Voicemeeter, a mixer, a capture card, a second sound card, any output OBS captures as an Audio Output Capture) or nowhere (only you hear your sounds, and the stream output if you set one). With another device the app never puts the cable back, never asks you to install it, and the header pill gets a tick once it's sending. With nowhere nothing nags. The choice is per PC and isn't in backups.
Stream output for OBS (Setup → Devices → Also play on → a device set to Clean, for streaming; one at a time, with its volume and Include my voice under it): what others hear, without the voice chat shaping, on a device of its own (a second virtual cable such as VB-Cable A+B, or any output you don't listen on). In OBS add it as an Audio Output Capture (or, for a cable, Audio Input Capture of its Output end) and your sounds and screen triggers are their own track, with their own volume. Your voice goes with them unless you untick it.
Transport bar: play/pause, stop, a seek slider, and speed, pitch and live effects while it plays (the 1x button: 0.25×–2×, ±12 semitones, keep the pitch or not, plus bass, treble, muffle, reverb, echo, distortion and presets on every sound playing). That's for listening and isn't saved; use Effects to keep a version.
Cards or a list: two icons at the top right of the Sounds, Apps and Voice tabs. The list fits many more on the screen: sounds as one-line rows, each program on one line, voices as a small picture beside the name. Each tab remembers its choice.
Any window size: less important controls tuck away as it gets smaller and come back when it grows. Small enough (below roughly 440 × 380) it becomes a mini player with your pads and the play / stop controls, down to 260 × 120.
Global hotkeys (set in ⚙ Settings → Hotkeys, the overlay key in ⚙ Settings → Overlay, the most-used ones from the keyboard button on the Sounds tab; all work in-game):
Stop all, and pause/resume all.
A random sound (from the category showing), the last sound again, next / previous category (beeps tell you which), sounds louder / quieter, your mic on / off, the voice changer on / off or only while a key is held, and all hotkeys off / on for when you need to type.
In-game overlay (by default the key left of 1: ` on a US keyboard; where that key types a letter, like ö or ñ, it's Alt + that key): a small panel of your sounds over the game. Number keys play them, and the game keeps your mouse and keyboard. Settings → Overlay: tap to open or hold to show, number row or numpad, hide after picking or after a while, which monitor and where (or drag it), its size and background, and Show preview.
Auto push-to-talk: holds your game's PTT key while a sound plays.
Instant replay: set its key and the last 30 seconds of everything your PC plays (a friend in Discord, the game, a video; not Onion Board's own sounds) are kept in memory. Press it after something funny and it becomes a pad. Nothing is saved or sent anywhere until you press it; clear the key to switch it off. Needs Windows 11, or Windows 10 version 2004 or newer.
Hotkeys can beep in your headphones (only you hear it), so you know they worked.
A game that runs as administrator: Windows doesn't let other programs hear the keys pressed in it, so hotkeys (and auto push-to-talk) work there only when Onion Board runs as administrator too. When a full-screen game like that comes to the front, the board says so, with a Restart as admin button.
Radio tab: internet radio from all over the world, from the free Radio Browser directory. Click a dot on the world map to tune in (hover one for its country, genres, quality and how popular it is; drag to move, scroll to zoom: zoomed in, the cities and towns with stations are named), or search by name, genre, country or city. Star stations for ★ Favorites; Popular and Recent list the most played and the ones you had on. One click plays a station; it plays in your headphones, goes out through your mic when you press Send, and can Record it as a pad. The bar's live controls change the pitch (up to ±36 semitones with Redline) and add bass, treble, muffle, reverb, echo, distortion or a preset.
Apps tab: send one program's sound (a music player, a browser, a video, a call in another app) to whoever's listening, without touching any other program. Each program is a card with its level, a Send switch, its own volume, Hear it myself and Record (waits for the program to make a sound, then adds it to your Sounds as a pad). Programs you switched on are picked up again next time they run. Under each card, the Clip editor keeps that program's last minute as a live waveform: freeze it, drag across the bit you want, Space to hear it, Send to play it to others, or Save to keep it in Saved clips at the bottom of the tab (double-click plays, F2 renames, right-click adds it to your Sounds, sends, copies or deletes it with Undo). Cut, copy, paste, fades, louder / quieter, reverse and undo are in its Edit menu; Ctrl+C there and Ctrl+V on the Sounds tab makes a new sound; the big-view button gives one program the whole tab. The tab's live controls change the pitch and effects of every program you send. A closed editor doesn't listen or keep anything. A program you removed comes back, settings and all, from Forgotten programs.
Voice tab: voice changer (25 ready voices, from Deep, High and Male voice and Demon to Autotune, Talkbox, Masked caller, Dark lord, a Dalek and a singing monk; pitch with a natural-sound mode and a voice-size control, autotune, mic clean-up (noise gate + hiss removal), monster growl, robot / talkbox, harmony, ring modulator, helmet, shout blowout, radio with walkie-talkie clicks, flanger, wobble (vibrato and tremolo), echo, reverb (with pre-delay and early reflections), distortion (oversampled, so it crunches without fizz), a studio polish (de-esser, warmth and a lookahead limiter), plus add-on effects such as an 8-bit bitcrusher; Save as a voice keeps your own mix as a voice button (Edit voice tweaks it, a share code copies it to a friend and Import a code adds theirs; removed ones wait in Recently deleted), and the dice picks a random one; it shows how much delay the voice adds, starts off every time the app opens, and a big ON / OFF button shows which it is), AI voices (optional: Get AI voices downloads about 55 MB, needs Python 3.12+; 12 characters turn your voice into someone else's, live on your own CPU, about 75 ms behind you. All voices shows each as a card with Hear it, sorted under Men, Women and In between & fun, with chips for low, middle or high voices; Make your own voice blends two and sets how deep and high it is; Remove AI voices takes it all off again), text-to-speech with Windows' built-in voices or your own (More options → Custom voices: a TTS server on your PC such as Kokoro or AllTalk — any OpenAI-style /v1/audio/speech address — a TTS program, or Piper voice packs dropped into the voices folder), and live voice-to-speech: press Start talking as the voice and each sentence you say is spoken by a computer voice instead of yours. Speech recognition runs on your PC (the first time, the Voice tab's Install speech recognition button downloads it, about 300 MB; needs Python 3.12+). Speak in makes the voice say it in one of 35 languages (Chinese, Spanish, French, German, Japanese, Portuguese…): talk in your own language and it's translated on your PC. Each language is an add-on you download only if you pick it (65–196 MB), and it needs that language's Windows voice (Settings → Speech → Add voices, free).
Triggers tab: plays a sound when a picture shows up in your game: a game's "YOU DIED", a rare spawn, a queue popping. It's the Onion Watch add-on: the tab's Add Onion Watch to this tab button downloads it from its GitHub release (checked against the SHA-256 GitHub lists) and it appears right there, no restart. Pick the game's window (it's watched even while other windows cover it, and two copies of a game are told apart) or a screen, cut the thing to watch for straight out of it (or paste a Win+Shift+S cut, or add a file), pick sounds from your board, and choose the wait, how soon it may play again and how close a match must be, with the live match % next to it. Test plays a trigger's sound now (it shows in Playing now, and its button says Stop while it plays). A trigger plays once each time the picture appears, or rings until you stop it. Its sounds play through the board like a pad; a ringing one loops in your headphones. When a newer Onion Watch is out, the tab offers it (only while Check for updates is ticked). Remove Onion Watch in its More menu uninstalls it (after asking); your triggers are kept for when you get it again. Everything happens on your PC: the screen is never saved or sent anywhere. If a game still shows up black, set it to Borderless or Windowed fullscreen.
Sound modes (Sounds tab's Listening box, Setup → Who's listening, Settings → Audio): Game and Voice chat shape your sounds for the voice chat on the other end, working out which one by themselves; Clean sends them as mixed; Advanced picks an exact voice chat or your own custom mode. Only what's sent is shaped: your own playback stays as mixed, unless you turn on Hear what they hear. What do these do? explains each.
Mute my mic while a sound plays (Setup → Who's listening): others hear only the sound, clean, and your mic comes back the moment it ends. Or the opposite, While I talk, lower my sounds.
Volumes: sounds → them, your voice → them, your headphones. Exact % boxes go up to 1000%; a soft limiter stops hard clipping. "Level volumes" makes every sound equally loud.
7-band equalizer on your voice, your sounds, or both, with presets.
Test mode: Hear what they hear plays your mic plus the sounds exactly as others get them (its button stays lit while it's on).
Connect your chat (Setup tab): Make it sound clean in Discord (Bun sets Discord up himself, see Discord set-up below), Set up game voice chat, Zoom, Teams or a browser call, Game has no microphone setting?, and the Step-by-step guide with Bun (your mic and headphones, where your sounds go, then Discord). Also play on (Setup → Devices) gives extra outputs a copy of what others hear, each with a − to remove it.
Search the web for sounds from the Sounds tab: Enter in Search sounds lists results (thumbnail, title, length) in place of the pads, from YouTube, YouTube Music, SoundCloud, Myinstants (short meme sound buttons) or TikTok sounds (reposted on YouTube); Play plays one once, Add keeps it as a pad (only the audio is downloaded, via yt-dlp). Instagram and most other sites have no search without an account, so paste a link to the video instead. The Search button's menu also has Trending on Myinstants: the meme sounds people play most there right now (the ones for your country), with nothing to type. YouTube's m4a/webm audio needs FFmpeg, like a dropped m4a file does. Settings → Updates → Helper programs lists FFmpeg, yt-dlp, the JavaScript engine and Tor, with Update now and Reinstall for when downloads start failing, and an opt-in box to update yt-dlp automatically (off by default). A yt-dlp update only installs a version this project approved (pins/yt-dlp.json, checked against yt-dlp's signed PyPI release and added by a reviewed pull request) once it has been out for 3 days, and never one older than the built-in copy. When a site breaks and no approved version fixes it yet, Force update installs PyPI's newest yt-dlp right away, after a warning that it isn't approved yet. FFmpeg and Tor copies the app downloaded follow the app: an app update that names a newer version fetches it (checked against its SHA-256) at the next start.
Discord check: while it runs, Onion Board reads Discord's own voice settings and shows a bar across the window when one would cut your sounds out (Noise Suppression, the Studio profile on your mic, Echo Cancellation, automatic input sensitivity, Advanced Voice Activity); Fix Discord has Bun set it up.
Discord set-up: Bun saves how Discord is set for talking, then switches off what chops up your sounds (Custom profile, noise suppression, echo cancellation, automatic gain and sensitivity off). Discord has to be closed while he writes: if it's open, he asks first, closes it, and opens it again on its voice settings. A copy of its settings folder is kept first. Put Discord back (Settings → Audio → Mic effect) undoes it, and Put Discord back for talking on close does it whenever Onion Board closes. On the virtual cable you still pick CABLE Output as Discord's mic; its guide shows where. With Straight into my mic, an app that opens your mic in "raw" mode (and so skips your sounds) is named in a warning bar too.
Your language: Settings → Appearance → Language opens a searchable list of English and 32 other languages, each in its own name (Arabic right to left). Windows set to one of them gets a one-time offer to switch.
Tabs: a new install starts with Sounds, Voice and Setup; the rest wait under + More tabs and are added when clicked. Settings → Tabs switches any of Radio, Apps, Triggers or Voice off so it doesn't load at all.
Community: Join the Discord (Settings, the tray menu, What's new) opens the Onion Board Discord for help and news.
Data & quality (Settings): Use less data, download quality, the radio's bitrate limit, Slow or patchy connection (waits longer before giving up), Also save the video for links, and Show pictures and like counts in search results.
Screen readers: buttons that only show an icon are read out by name.
⚙ Settings (a search box above the categories keeps only the cards with the word you type; changes apply at once):
Privacy & security: a switch for each thing the app does online, and Offline mode.
Connection: direct, a proxy or Tor, Network activity, and Keep an app log (off stops onionboard.log being written and deletes it).
Data & quality: download sizes, radio quality and search extras.
General: the window (Keep window on top, Play pads with one click), tips, starting with Windows, backups and Start over (Reset just the parts you pick, with a restore point first; Restore points undoes it).
Tabs: switch the Radio, Apps, Triggers or Voice tab off so it doesn't load at all.
Appearance: 32 themes in four groups, switched live, Halloween first: Classic (Dark, Light, true-black Midnight, High Contrast…), Colourful, Wild (Synthwave, Hacker, Amber Terminal…) and Meme (Flashbang, Deep Fried, Retro 98, Comic Sans…); your own highlight colour, how live tabs show (a tint or a dot), the language, the interface size.
Audio: devices, Also play on, the stream output and buffering (Low or Safer, with a count of drop-outs).
Hotkeys, Overlay, Updates (when to check, maintenance), Add-ons (Onion Watch and Onion Pocket), Remote, and About (version, links, the Discord, feedback, bug reports and support).
MIDI pads and macro keypads: a pad controller (Akai LPD8 / MPD, Launchpad, any USB MIDI keyboard) works without extra software: set a hotkey and hit a pad instead of pressing a key. Pads work for sounds, stop / pause, random sounds and the overlay. A controller is only opened while a hotkey uses it, so your music app can have your other MIDI gear. On Windows only one program can use a MIDI device at a time: if one is busy, Onion Board says so and picks it up once it's free. A cheap USB macro keypad works too: set its keys to F13–F24 (keys nothing else uses) and use those as hotkeys.
Stream Deck and scripts (optional, off by default): Settings → Remote → Remote control lets programs on this PC play your sounds: a Stream Deck (with the Onion Board plugin, a Play sound key per sound picked from a list, lit while it plays; or Bitfocus Companion, Touch Portal, website buttons), AutoHotkey or a script. It only listens on this PC and needs the key shown there (Copy an example link gives a ready-made "play a random sound" link). Besides playing and stopping sounds it can mute you (a panic button), switch the voice changer on or straight to a voice, say a typed line in the computer voice, queue sounds or play a whole category in a row, turn the mic and Hear what they hear on or off, change the volume and category, save the instant replay, change the live speed, pitch and effects, switch who's listening, run the radio, send a program from the Apps tab, put on one of My voice combos, switch every hotkey off while you type, and take chat commands (!sound, !random, !say… with cooldowns and who may use each) and alerts and donations (follows, subs, bits, raids, tips, channel points, TikTok gifts) from Streamer.bot, Firebot, TikFinity and the like; /api/help lists it all and CONTROL-API.md explains it. New to it? The Streamer guide there walks through Stream Deck keys, channel points and chat commands (Streamer.bot), and Copy AI prompt gives ChatGPT / Claude everything it needs (the links, your sounds) to set up whatever tools you use with you.
Your pads on your phone (Onion Pocket add-on, optional, off by default): Settings → Remote → Get Onion Pocket downloads and installs it (from its GitHub release, checked first); then its card is there. Scan the code with your phone's camera and your pads show up in its browser (iPhone or Android, nothing to install); a tap plays the sound on the PC. It has its own key (Forget phones makes a new one), only answers phones on your home network, and can play, stop and mute sounds, never touch your mic. Meant for your home Wi-Fi, not public Wi-Fi. When a newer Onion Pocket is out, its card offers Update Onion Pocket to …: one click and the new one runs, phones still paired.
Privacy: Settings → Privacy & security has a switch for each thing the app does online, and Offline mode switches them all off (the installer can set it before the first start). Everything can go through a SOCKS5 / HTTP proxy or Tor (Get Tor downloads the Tor Project's own, checked first), and never quietly goes direct if that fails. Network activity (Settings → Connection) lists every connection the app makes and why. By itself it only checks for updates and sends an anonymous daily count (version, tabs opened, and a random ID that shows how people use the app over time); each has its own switch. Little touches that follow the time of day read your PC's clock and never send it.
Runs in the background: closing the window keeps it in the tray (hotkeys and the overlay keep working; right-click the tray icon for Open Onion Board, Stop all sounds or Quit). Optionally starts when you sign in, straight to the tray (Settings → General).
Updates itself: each time it starts, and every hour after, it checks GitHub for a new version (untick it in Settings → Updates). One found as it starts installs itself behind the loading screen (Updating to X… 42%): downloaded, checked against the file GitHub lists, installed, and the new version opens, keeping your sounds and settings (untick Install updates by themselves to always be asked). Found later, Update now downloads it and Restart now installs it. An important fix (a release whose notes have an Urgent: … line, for Onion Board or Onion Watch) shows as a banner across the window with the reason and an Update now button; it can be hidden until the next start, but not skipped for good.

How it works

Straight into your mic (the default):

🎤 your mic ──► Windows' audio engine ──► Onion Board's effect ──► Discord / game (same mic)
                                              ▲         │ clean mic
🔊 your sounds ─► Onion Board mixes them ─────┘         ▼ (your voice changer, gate, volume…)

Onion Board puts a small Windows audio effect on your mic (like Equalizer APO or Soundpad do). It hands Onion Board your clean voice, and puts back what you send: your processed voice plus your sounds, about 20 ms later. Close Onion Board and your mic is just your mic again.

It works with Equalizer APO on the same mic, whichever went on first: your EQ still runs, Onion Board's sounds still get through, and taking either off leaves the other working.

With the virtual cable:

🎤 your mic ── send ✓ / ✗ ──┐
                            ├─►  Onion Board mixes them  ─►  CABLE Input ═══ pipe ═══► CABLE Output
🔊 your sounds ─────────────┘                                                          (Discord / game mic)

The virtual cable is a free audio driver that works like a pipe: Onion Board plays into one end, and Discord or the game uses the other end as a microphone. You hear the sounds in your own headphones separately.

Setup → Devices → Send my sounds to picks between these, any other output by name (Voicemeeter, a mixer, a capture card, OBS) or Nobody (your headphones and the stream output only).