For bands, jam groups & VRChat performers

Play together online.
Perfectly in sync.

VRC Downbeat lets your band jam over the internet with sample-accurate timing — no matter how bad your ping is. Everyone plays to the same backing track, the app lines every performance up to the exact sample, and the host streams the full band mix straight into VRChat.

Get it from Microsoft See how it works
  • Free & open
  • Windows 10/11
  • End-to-end encrypted
  • Join with a room code
  • ASIO & WASAPI

The problem

Latency killed the online jam.
So we removed it from the equation.

Playing music together over the internet normally fails for one simple reason: sound arrives late. Even 30 ms of network delay is enough to make a band fall apart — and most tools try (and fail) to fight this with ever-smaller buffers.

VRC Downbeat takes a different approach: nobody listens to each other live over the network. Instead, every musician plays along to the same backing track locally — with zero latency — and the app reassembles all performances on the host's machine, sample by sample, exactly as they were played. The result sounds like the whole band was in one room.

How it works

What actually happens in the background

One app, two roles: one of you is the Host, everyone else joins as a Musician. Here is the full journey of a jam, step by step.

1

The host builds a song library

The host adds songs as stems — separate tracks for drums, bass, guitar, vocals and so on (platforms like Karaoke-Version sell songs in exactly this format — more on that in the guided tour). Then the host opens a room via Connect › Host Downbeat Server… and gets a short room code like D29UQS to share with the band — public, or private with a password. No port forwarding, no exposed IP address. (Prefer full control? Connect › Host on this PC… still offers classic direct hosting on your own port.)

2

Musicians connect and download the songs

Each musician types in the room code, picks a name and connects — the connection is end-to-end encrypted from the first byte. Songs and their stems travel over the same connection — once the host has allowed it and the musician has accepted it — and are kept in the musician's own Music folder, so the next session with the same song needs no download at all. Every musician gets their own personal mix of the backing track — the drummer can mute the drum stem, the singer can push the vocals up, and nobody else is affected. Everyone also sets up their own inputs once: a mic, a guitar, keys or a VST instrument, each with its own strip in the mixer.

3

Soundcheck: “Prepare Jam”

The host picks a song and presses Prepare Jam. If someone is missing the song, the host chooses who may receive it, each of those musicians is asked whether they want it, and a Transfers window shows every download with progress and remaining time. Everyone loads the song and reports ready (a musician who declines simply sits this jam out). Meanwhile the host runs the soundcheck: switch Send Input on for one musician at a time, hear their inputs almost live, and set their level like a sound engineer at the desk. Those levels carry over into the jam.

4

Everyone plays locally — with zero latency

After a 3-2-1 countdown — “GO!” — playback starts at sample 0 of the backing track on every machine. You hear the backing track and your own instrument directly through your audio interface — there is no network audio in your monitoring at all, so it feels exactly like practicing at home.

While you play, the app records your inputs and applies latency compensation like a DAW would: the first few milliseconds (your interface's input + output latency) are discarded, so what remains lines up 1:1 with the backing track. The audio is then sent to the host, tagged with its exact sample position.

5

The 3-second gate makes network lag irrelevant

The host does not play anything immediately. It waits until it has at least 3 seconds of audio buffered from every musician, then starts its own playback: backing track and all incoming performances, read in lock-step from sample 0.

Because every incoming block carries its exact sample position, network jitter, slow connections or short dropouts can never push a musician out of time — a lost block simply becomes a moment of silence, and the stream continues exactly where it belongs. The mix at the host is sample-accurate, always.

6

The band mix goes live — straight into VRChat

The host hears the full band about 3 seconds behind the musicians and mixes it live: every musician has a fader, every stem has a fader. Through the built-in virtual microphone stream, that mix is fed into VRChat (or Discord, or OBS) — so your audience hears a tight, mixed band performance in real time. The host can play or sing along on top of the delayed mix: their own inputs go straight into it. Hosts with a streaming invitation can also send the same mix to a public listener link for VRChat video players — no OBS needed.

Features

Everything a remote band needs

From the first soundcheck to the final multitrack recording.

Sample-accurate sync

Every performance is placed at its exact sample position. Network lag, jitter and dropouts can never make the band drift apart.

Jam on slow internet

Every musician picks their own stream quality in the quick bar: uncompressed PCM, lossless FLAC (half the data) or Opus down to ~0.09 Mbit/s — up to 9× less than PCM. Mixed choices in one band are fine; the timing stays sample-accurate. Change it any time before the jam starts.

Songs with stems

Backing tracks are made of individual stems. Each musician mixes them locally — mute your own instrument's track and take its place. A master fader moves the whole backing track at once.

Per-channel FX chains

Low cut, high cut, EQ, noise gate, compressor, distortion, chorus, phaser, delay, reverb and limiter — 11 effects, freely ordered, on every stem, every musician and every input. Recordings stay raw; the FX are re-applied (and still editable) on playback and export.

Drag & drop library

Drop audio files or whole folders straight onto the song list — a folder becomes one song with all its files as stems. The pencil opens Edit Song: rename the song or its stems, drop extra stems in, remove others. Connected musicians get the update instantly.

MON and MIX on every input

Each input has its own switches: MON — you hear it; MIX — the band and the host get it. Hear yourself (or don't) without changing the take, or keep a mic out of the jam entirely. Sound Activation keeps room noise out while you're quiet.

DAW-grade latency compensation

ASIO support reads your interface's exact latency and compensates it automatically. USB mics, headsets and other Windows Audio devices work too — the app estimates their delay, and a manual offset fine-tunes it.

VST3 instruments and effects built in

Load a VST3 drum or synth instrument (e.g. Superior Drummer 3), plug in your MIDI kit and play — or run your guitar through an amp sim and your voice through a vocal chain. Everything is rendered inside the app: no DAW, no audio routing, no bleed from the backing track.

Create Input: tell it what's plugged in

Mic, guitar jack, keyboard, MIDI controller, line-in gear or a custom VST insert — pick the type and the Create Input window shows only the options that type needs. Find your jack by playing: the channel whose meter moves is yours.

VRChat virtual mic

Route every stem, musician, your own inputs and any Windows app independently to You or the VRChat mic — including Spotify with You off. Works with the free VB-CABLE driver (the app links you to the official download and picks the cable up automatically); no Voicemeeter needed.

Windows app mixer inside

Route any program — Spotify, a YouTube tab, your DAW — into the virtual mic with per-app volume and mute, right from the app. Like EarTrumpet, but built in.

Talk with “Send Input”

No separate talk mic: between songs your normal inputs are your voice. In the band list you switch Send Input on per person and decide exactly who hears you — the host for the soundcheck, a bandmate for a quick word. Text chat included.

Multitrack recording

Record every jam as separate, sample-aligned tracks: one per musician, the host's inputs and the backing stems. Remix in the Recordings window and export the whole mix or single stems as WAV, FLAC, MP3, AAC or OGG.

Share recordings

Send a finished recording — as an MP3 mix or as FLAC stems — to one musician or the whole band over the same encrypted connection. The recipient confirms with one click; nothing leaves your PC on its own.

End-to-end encrypted

Every connection is sealed with modern authenticated encryption (X25519 + XChaCha20-Poly1305). Chat, songs, audio — nobody in between can read any of it.

Join with a room code

Host a session on the Downbeat Server and share a short code like D29UQS — public, or private with a password. No port forwarding, no router setup, no IP addresses exchanged.

Live stream to a link

Hosts with a streaming invitation can send the same band mix to a public listener link — for VRChat video players and worlds, no OBS needed. Pick the quality, press Go Live, share the link.

Privacy by default

You appear under a name you choose — never your computer name. IP addresses are hidden in the app by default and never written to chat or logs.

Privacy & security

Jam together, without giving anything away

Older jam tools force the host to open a router port and hand out their public IP address to the whole band. VRC Downbeat was rebuilt so that nobody has to share anything:

Room codes instead of IP addresses

Both the host and the musicians connect outbound to the Downbeat relay server — like any normal website visit. The relay pairs everyone up by room code and then simply forwards the data. Nobody sees anybody else's address, and nobody touches their router.

The relay can't read your session

Encryption is end-to-end: right after connecting, musician and host exchange ephemeral keys (X25519) and every single message — chat, song transfers, jam audio, voice, recordings — is sealed with XChaCha20-Poly1305 authenticated encryption. The relay only ever forwards ciphertext; not even its operator can listen in. Direct-IP connections get the exact same encryption.

The full details — what the relay sees, what is logged and for how long, which optional features contact third parties — are in the Privacy Policy. The rules for what you may share in a session, and how to report abuse, are in the Terms & Content Guidelines.

Guided tour

From download to your first jam

A walkthrough of a real session, step by step — click any image to enlarge it. The diagrams show what is new in 1.1.0; some screenshots still show the previous layout, but the steps are the same. Inside the app, Help › Guide has the full illustrated manual.

Step 1

Choose your role

Start the app and pick a side: Host a jam if you manage the band's songs and want to hear (and stream) the finished mix — Join as musician if you want to plug in and play. One person hosts, everyone else joins. You can switch roles anytime with Change Role in the menu bar.

VRC Downbeat startup screen with Host a jam and Join as musician buttons
Step 2

Host: build your song library

The Library panel on the left is the band's songbook, and only the host manages it. Click Add song… and select all the audio files that belong to one song in a single go — each file becomes one stem (WAV, MP3, FLAC, AIFF and OGG all work). Give the song a name, confirm, and the app copies the files into its own library and reads their length and sample rate automatically. Or simply drag a folder onto the list: it becomes one song with every file inside as a stem.

The list shows a stem counter next to each song — “5 stems” means bass, drums, guitar, piano and vocals arrive as separate faders for everyone. Only have a normal full mix? That's fine too: one file simply becomes a one-stem song. Named something wrong? The pencil opens Edit Song — rename the song or its stems, add or remove stems (outside of a jam, so everyone always has the same stems). Your musicians never touch any of this — when you press “Prepare Jam” the app asks you who may receive the song, and sends it to them.

Where are the files? Everything the app keeps for you — Songs, My Songs, Downloads, Recordings, Exports and Settings — lives in Music\VRC Downbeat in your Windows user folder. Settings › General › Data folder lets you pick another place (an external drive, for example) and offers to move your files there.

Where do stems come from? Karaoke-Version sells custom backing tracks where every instrument is a separate download — perfect for this app. Moises can split any regular song into stems (or remove just your instrument) using AI. VRC Downbeat is not sponsored by or affiliated with either of these services — they're simply what works well in practice.

File picker with the five stem files of one song selected at once
Add song dialog naming the new song, with the library list and stem counters behind it
Step 3

Host: open a room, share the code

Connect › Host Downbeat Server…: enter your name, choose Public or Private (with a password) — that's it. The app connects outbound to the relay and hands you a 6-letter room code (with a Copy code button) to give to your musicians. No port forwarding, no router settings, and your IP address is never shown to anyone. Prefer full control? Connect › Host on this PC… hosts directly on your machine instead, and Test Port checks from the internet whether your port is open.

Host view with an open room on the Downbeat Server showing the room code, song library and hidden IPs
Step 4

Musicians: type the code & grab the songs

Connect › Connect via Downbeat Server…: pick a name, type the room code (a private room asks for its password) — the end-to-end encrypted link is established automatically. Songs from host shows the host's library right away; the badge tells you how many stems of each song you already have. Download asks the host for a song ahead of time — the host approves, a Transfers window shows the progress, and the song is checked against the host's copy on arrival. If you already have an identical stem in My Songs, it is reused instead of downloaded again. Keep as My song copies a download into your own library for good. (Joining a self-hosted server by IP and port works the same way via Connect › Connect to Host….)

Downloads stay with you. Songs from a different host are shown greyed out while you are in another band's room, and are fully usable again when you are offline — or you can delete them from the list whenever you like.

Musician view connected to the host via room code, showing the host's song list and band chat
Step 5

Make the backing track yours

Double-click a song to load it and every stem gets its own strip: fader, M (mute), S (solo) and FX. Playing drums? Mute the drum stem and take its place. Practice with the player as long as you like — this is exactly what you'll hear during the jam, with zero latency. Your mix is saved with the song and comes back next time.

Musician stem mixer with five stems, each with volume fader and mute
Step 6

Set up your sound — once

Settings › Audio has three parts. First the Output: your headphones or audio interface (pick its ASIO driver — the app then knows your exact latency and compensates it automatically). Then your Inputs: Create new Input asks what is plugged in — a microphone, a guitar (clean or through an amp-sim VST), keys or a drum module, a MIDI controller or e-drums playing a VST3 instrument, other line-in gear — and shows only the options that type needs. Find your jack by playing: the channel whose meter moves is yours.

Every input becomes a strip in your mixer with ON, MON (you hear it), MIX (the band gets it), a fader and an FX chain. A headset mic is just another input — there is no separate talk mic anymore. Sound Activation keeps room noise out while you are quiet. All of this is remembered per song.

Diagram: an input runs through gain, an optional VST3, an FX chain and Sound Activation, then MON, MIX and VRChat switches decide where it goes; below, the input types of the Create Input window
Step 7

Soundcheck with “Prepare Jam” and “Send Input”

The host picks a song and hits Prepare Jam. Musicians who already have the song load it straight away. For the others the host is asked who may receive it, each of them is asked whether they want it, and the Transfers window shows every download with progress and remaining time. The status counts “2/3 ready”; a musician who declines sits this jam out (their strip goes dark, the host is told before starting). Now the soundcheck: next to every person in the band list sits a Send Input switch (red = off, green = on). The host switches it on for one musician at a time, hears their inputs live and sets their level on their strip. When everybody is ready, one click on Start Jam runs the 3-2-1 countdown for the whole band.

Send Input is also how you talk: whoever you switch it on for hears your MIX-enabled inputs as voice while no jam is running — the host for the soundcheck, a bandmate for a quick word. It starts off for everyone, so nobody hears you until you say so.

Diagram: in the band list every person has a Send Input switch; the ones switched on hear your inputs as voice between songs, the others do not
Host view during Prepare Jam with stem faders, musician fader and ready status
Step 8

Play — even on a slow connection

On “GO!”, the backing track starts on every machine at once and you simply play along. In the background, your take is latency-compensated, timestamped and streamed to the host, where the whole band is reassembled sample-accurately — the host hears the full mix about three seconds later.

Slow upload? Pick your own Stream to host quality in the quick bar under your mixer: PCM (lossless, ~0.8 Mbit/s), FLAC (lossless, ~0.4) or Opus down to ~0.09 Mbit/s — about what a voice call needs. Everyone chooses for themselves, and the host's musician list shows what each person is sending. Set it before the jam; it is locked while a jam runs.

Diagram: three musicians each pick their own stream quality — PCM, FLAC or Opus — and the host places all of them on the same beat
Step 9

Send the band into VRChat

Settings › Audio Stream is your broadcast desk: pick the virtual cable as the VRChat mic (no cable yet? Get VB-CABLE opens the free driver's official download page, and the app picks the cable up as soon as it is installed). The routing board has two columns — You (what you hear) and VRChat mic (what goes into the cable) — with a row and a meter for the backing track, every musician, your own inputs and every Windows program. The columns are independent, so Spotify can reach the cable while you stay silent. In VRChat you select CABLE Output as your microphone and switch its noise suppression off — it is made for speech and chews up music.

Audio Stream settings with VRChat mic cable selection and per-source You / VRChat mic switches
Step 10

Relive every jam

Switch on Settings › General › Record during a Jam and every jam lands in the Recordings window as separate, sample-aligned tracks — one per musician, plus the host's inputs and the backing stems. The window has its own player and mixer: fader, mute, solo and FX per track, without touching the main screen. Export renders the whole mix or single stems as WAV, FLAC, MP3, AAC or OGG; Send to… hands the recording to your bandmates through the app — as an MP3 mix or as FLAC stems, and they confirm before anything is received.

Diagram of the Recordings window: its own mixer and player on the left, the Export window with Whole mix or Single stems, formats WAV, FLAC, MP3, AAC and OGG and Send to on the right

Under the hood

For the technically curious

No magic — just careful audio engineering. This is what the app really does.

Native C++ audio engine

Built with the JUCE framework — the same technology behind many professional audio plugins. All processing runs in real-time audio threads with lock-free FIFOs between them; the network can never glitch your sound.

A simple, direct protocol

Host and musicians talk over one TCP connection: JSON messages for control (song lists, jam state, chat) and compact binary blocks for audio, each stamped with its absolute sample position. That stamp is the whole secret — audio can arrive late, but never in the wrong place. Every block also says which codec it carries (PCM, FLAC or Opus), so the host decodes a mixed band without any setup.

The Downbeat relay server

A deliberately dumb pipe: it understands three greetings — open a room, join a room, pair up — and after that it forwards bytes without interpreting them. Rooms die when the host leaves. Because it never decodes audio, one small server can carry many bands at once, and the whole jam protocol travels through it unchanged.

End-to-end encryption

After pairing, musician and host exchange ephemeral X25519 public keys and derive per-connection session keys; every frame is then sealed with XChaCha20-Poly1305 authenticated encryption (via the audited Monocypher library, keyed by the OS random generator). Keys live only in memory, tampering is detected, and the relay never holds any key material.

Positioned ring buffers

On the host, every musician gets a ring buffer indexed by absolute sample position. Incoming audio is written to exactly the right cell; the mixer reads and re-zeroes cells as it plays. Unwritten regions are silence — a dropout can never cause a permanent offset.

Everything in “song time”

The song's own sample rate is the master clock. Musician recordings are resampled from their device rate into song rate before sending; the host only converts to its device rate at the very last output stage. One shared timebase, zero drift.

The virtual mic, demystified

Windows only lets kernel drivers create virtual audio devices, so VRC Downbeat uses the free VB-CABLE driver (a separate download from VB-Audio — the app links to it and detects it once installed) and does all mixing and routing itself: band mix, your own inputs and per-app captures are resampled and merged into the cable, which VRChat sees as a normal microphone.

Independent audio devices

Your Output's driver (ASIO or Windows Audio) is the app's clock and carries your instrument. USB mics and headsets run as their own Windows devices, and the virtual-mic cable is another — all in shared mode, so they coexist cleanly and your precious low-latency ASIO driver is used by exactly one thing: your music.

FAQ

Frequently asked questions

Do the musicians hear each other while playing?

Not live — that's the point. Each musician hears only the backing track and themselves, with zero latency. The host (and the VRChat audience) hears everyone together, about 3 seconds later, perfectly in sync. Between songs you talk over Send Input: switch it on for the people who should hear you, and your normal inputs become your voice.

Do I need an ASIO interface?

Recommended, not required. ASIO drivers report exact latency values, so compensation is automatic and precise. USB mics, headsets and other Windows Audio devices work too — the app estimates their delay, and if a take lands audibly early or late you fine-tune a manual offset once.

What does the host need to set up?

Nothing, in the default mode: Connect › Host Downbeat Server…, get a room code, share it with the band — done. No port forwarding, no router settings, no IP sharing. Make the room private if you like; then it needs a password. If you prefer to host entirely on your own hardware, Connect › Host on this PC… gives you classic direct hosting (then you forward one port; Test Port checks from the internet whether it is open).

Is my session private?

Yes. All traffic — chat, songs, audio, recordings — is end-to-end encrypted between musician and host with modern authenticated cryptography (X25519 key exchange, XChaCha20-Poly1305). The relay server only forwards encrypted data and can't read any of it. In the app, you appear under a name you choose, IP addresses are hidden by default, and the relay logs no addresses and no contents.

How does the VRChat integration work?

VRC Downbeat plays its mix into a virtual audio cable (the free VB-CABLE driver). The app does not bundle or install the driver itself — it links you to the official download at vb-audio.com and picks the cable up automatically once it is installed. In VRChat you simply select the cable as your microphone. You can also route other apps — Spotify, a browser tab — into the same cable with the built-in app mixer.

Can I play e-drums or other MIDI instruments?

Yes. Create an input of the type E-Drums (MIDI) → VST drums or MIDI keyboard → VST instrument, browse to your VST3 (Superior Drummer 3, a synth, a piano …) and pick the MIDI port. The instrument's sound is rendered inside VRC Downbeat, so it's captured digitally with perfect separation — the backing track can never bleed into your take.

Can I use my amp sim or a vocal plugin?

Yes. The input types Guitar / Bass → amp sim VST and Microphone through a VST run your clean signal through a VST3 effect — an amp sim, a vocal chain, pitch correction — before it reaches the band. The host receives the finished sound, and you hear it with the latency of your interface only.

Do I need a separate microphone for talking?

No. Any input with MIX on is your voice as soon as you switch Send Input on for someone in the band list. A headset mic is simply another input; a singer talks through the same mic they sing into. During a jam nobody hears anyone live — Send Input is for the moments in between.

My internet is slow. Can I still join?

Very likely. Under Stream to host in the quick bar you pick how your audio travels: uncompressed PCM (~0.8 Mbit/s upload), lossless FLAC (~0.4) or Opus down to ~0.09 Mbit/s — roughly what a voice call needs. Every musician chooses for themselves, and the 3-second buffer absorbs Wi-Fi hiccups. What you hear locally is never affected.

Is a jam recorded?

Only if the host switches on Settings › General › Record during a Jam. Then every jam is saved on the host's PC as separate, sample-aligned tracks — one per musician, plus the host's inputs and the backing stems. Any recording can be remixed in the Recordings window, exported as WAV, FLAC, MP3, AAC or OGG (the whole mix or single stems), or sent to band members through the app — they confirm before anything is received.

Which platforms are supported?

Windows 10/11 for now. The virtual-mic features use Windows-specific audio APIs (per-app capture and routing), which is where VRChat runs for most performers anyway.

Ready to get the band back together?

Grab VRC Downbeat, send your bandmates a room code and start playing — no accounts, no subscriptions, no router setup.

Get it from Microsoft

Contact

Questions, feedback

Email

[email protected]

Discord

marokugrim