OBS NDI Voice Changer: Two-PC Streaming Setup Guide

An OBS NDI voice changer setup is the cleanest way to run real-time voice effects in a dual-PC streaming rig — no capture card audio tap, no analog loopback, no driver hacks. The gaming PC runs the voice changer and outputs processed audio to OBS NDI Output; the streaming PC picks it up as a LAN audio source in OBS Studio. This guide walks through the full topology, explains where latency actually comes from, and gives you the exact configuration steps for NDI 5/6 with Windows 10/11.

TL;DR

NDI carries processed voice audio from the gaming PC to the streaming PC over gigabit LAN at under 3ms additional latency.
No capture card needed — NDI handles both game video and voice changer audio in one LAN connection.
The voice changer runs entirely on the gaming PC; the streaming PC only receives and mixes the output.
VoxBooster outputs to a virtual WASAPI mic that OBS NDI Output captures automatically — no extra routing config.
Use a direct Ethernet cable between both PCs; Wi-Fi causes glitches that ruin the audio sync.
NDI 6 is more CPU-efficient than NDI 5; install the obs-ndi plugin version that matches your NDI SDK.

Why a Two-PC Setup Changes the Voice Changer Equation

Single-PC streaming is the norm, but it has a fundamental tension: the gaming PC must simultaneously run the game, encode the stream, and process voice audio. High-fidelity AI voice processing — the kind that reshapes your voice convincingly rather than just pitch-shifting it — requires consistent CPU/GPU headroom. When a game is already using 80% of your GPU, adding real-time neural voice conversion on top creates either frame drops in the game or stutter in the voice output.

A two-PC setup resolves this by splitting responsibilities:

Gaming PC: runs the game, runs the voice changer, sends audio and video over NDI.
Streaming PC: receives NDI sources, mixes audio, encodes the stream to Twitch/YouTube.

The voice changer stays on the gaming PC because that is where the microphone is physically connected. The streaming PC never touches raw audio — it only receives the finished, processed voice signal.

This separation means you can run AI voice cloning at full quality without worrying about encoder competition. The streaming PC’s CPU is free to handle NVENC or x264 encoding without interference.

Understanding the Signal Path: Gaming PC to Streaming PC

Before touching any settings, it helps to understand exactly what travels where:

Microphone (physical)
  └── Voice Changer (gaming PC — VoxBooster)
        └── Virtual Microphone Output (WASAPI)
              └── OBS Studio (gaming PC)
                    └── OBS NDI Output plugin
                          └── LAN (Ethernet / gigabit)
                                └── OBS Studio (streaming PC)
                                      └── NDI Source plugin (audio + video)
                                            └── Stream encoder → Twitch / YouTube

Every arrow in that chain happens in software — the only physical interface is the Ethernet cable between the two machines. The voice changer’s output virtual microphone is what OBS captures on the gaming PC, not the physical microphone. This is the critical distinction: if you select the wrong audio source in OBS NDI Output, you send raw unprocessed voice to the streaming PC.

Latency Budget Across the Chain

Understanding where latency accumulates helps you tune the setup correctly:

Stage	Typical Latency	Notes
Microphone capsule → ADC	1–3 ms	Hardware fixed
Voice changer DSP effects	8–15 ms	Pitch shift, EQ, modulation
Voice changer AI processing	80–200 ms	Neural voice model
WASAPI audio graph	2–5 ms	Windows audio stack
OBS audio capture	1–3 ms	Per-frame audio buffering
NDI network transmission	1–3 ms	Gigabit LAN
OBS NDI receive buffer	5–15 ms	Jitter buffer, adjustable
DSP total (end-to-end)	~18–40 ms	Real-time threshold is 100 ms
AI total (end-to-end)	~90–250 ms	Invisible to viewers at broadcast delay

The viewer never hears your raw voice — they hear the encoded stream which already has 5–10 seconds of Twitch/YouTube buffer. The 90–250ms you care about is what you hear in your own headphone monitoring mix. If you monitor your own voice, add a matching delay to your game audio so everything lines up.

What You Need Before Starting

Hardware:

Two Windows 10/11 PCs on the same local network
Gigabit Ethernet switch or direct Ethernet cable between both PCs (avoid Wi-Fi)
Microphone connected to the gaming PC only

Software on the gaming PC:

OBS Studio 30.x or later
obs-ndi plugin (version 4.x for NDI 5 SDK, version 5.x for NDI 6 SDK)
NDI Runtime (installed alongside obs-ndi)
VoxBooster (or any voice changer that creates a Windows virtual microphone)

Software on the streaming PC:

OBS Studio 30.x or later
obs-ndi plugin (same version as gaming PC)
NDI Runtime

Both PCs must use the same major NDI SDK version (both NDI 5 or both NDI 6). Mixing versions causes connection failures.

Step 1 — Install NDI Runtime and obs-ndi on Both PCs

Go to ndi.video and download NDI Tools for Windows. The installer includes the NDI Runtime that obs-ndi requires.
Install NDI Tools on both gaming PC and streaming PC.
Download obs-ndi from obs-ndi.io — choose the release that matches your NDI SDK version.
Run the obs-ndi installer on both PCs. It places the plugin files in OBS Studio’s plugin directory automatically.
Restart OBS Studio on both PCs after installation.

To verify installation: in OBS Studio, go to Tools in the menu bar. If you see NDI Output Settings and NDI Source, the plugin is active.

Step 2 — Configure VoxBooster on the Gaming PC

Open VoxBooster and set up your voice effect. The key output setting is the virtual microphone — this is what OBS will capture.

In VoxBooster:

Select your physical microphone as the input device.
Enable your chosen voice effect (real-time DSP preset, or load an AI voice model).
Confirm the Virtual Microphone output is enabled. VoxBooster creates a Windows virtual audio device named something like “VoxBooster Virtual Mic” — verify it appears in Windows Sound settings under Playback and Recording devices.

Quick sanity check: open Windows Sound settings, go to Recording, and speak into your microphone. The VoxBooster virtual mic input meter should move, showing the processed voice is reaching the virtual device. If only the physical mic moves, VoxBooster is not routing correctly.

Step 3 — Configure OBS NDI Output on the Gaming PC

The gaming PC’s OBS Studio sends its audio (and optionally video) to the streaming PC via NDI Output.

In OBS Studio on the gaming PC, go to Settings → Audio.
Under Mic/Auxiliary Audio, select VoxBooster Virtual Mic (or whatever your voice changer’s virtual device is named). This ensures OBS captures the processed voice, not the raw microphone.
Go to Tools → NDI Output Settings.
Enable Main Output (sends the full OBS scene mix including your processed voice audio over NDI).
Optionally, enable a dedicated Audio Only NDI output if you want the streaming PC to receive the voice as a separate audio source independent of the video NDI stream.
Note the Source Name — this is what the streaming PC will look for (default is usually your PC hostname).

Choosing Between Main Output and Dedicated Audio Source

Option	Best For	Limitation
Main NDI Output (audio + video)	Sending game capture and voice together	Video encoding adds CPU load on gaming PC
Dedicated audio NDI source	Sending voice only, video via separate path	Requires two NDI sources on streaming PC
Audio Monitor to NDI virtual device	Complex routing via virtual audio cable	Extra latency from audio monitor path

For most setups, Main NDI Output with audio included is the simplest. The streaming PC receives one source containing everything.

Step 4 — Add the NDI Source on the Streaming PC

On the streaming PC’s OBS Studio:

In a scene, click + under Sources → select NDI Source.
In the NDI Source properties, set Source name to the gaming PC’s NDI source (it appears in the dropdown once the gaming PC’s NDI Output is active and both PCs are on the same LAN subnet).
Under Audio, ensure Sync audio to video is checked if you’re receiving both audio and video in one NDI source.
Set the Bandwidth to Highest for the best audio fidelity (NDI lossless audio).
The Buffer compensation (latency) setting controls how much buffering OBS applies to the incoming stream. Start at 150 ms — this absorbs LAN jitter. You can lower it to 50 ms on a direct Ethernet connection.

After adding the source, you should see the gaming PC’s OBS output appear in the preview. Click on the NDI Source in the mixing board to confirm audio is also coming through — the audio meter should move when you speak.

Step 5 — Verify Audio Routing End-to-End

This is the step most guides skip, and it is where setups break:

On the gaming PC, speak into your microphone. Confirm VoxBooster’s virtual mic meter moves with the processed voice.
In OBS on the gaming PC, confirm the Mic/Aux channel meter moves (showing OBS captures the virtual mic, not silence).
On the streaming PC, open OBS audio mixer. The NDI Source audio channel should show activity when you speak.
Do a test recording on the streaming PC. Play it back and verify the voice is processed (not raw) and in sync with any game audio you included.

If the streaming PC receives raw voice (unprocessed), go back to Step 3 and confirm that OBS on the gaming PC is set to the virtual mic, not the physical mic, in its audio settings.

Step 6 — Latency Compensation and Audio Sync

In a two-PC NDI setup, slight audio-to-video sync issues are common on first setup. The NDI receive buffer on the streaming PC introduces a small delay that may not match the video delay.

Fix audio sync in OBS on the streaming PC:

Right-click the NDI Source in the mixer and select Advanced Audio Settings.
Adjust the Sync Offset in milliseconds. If audio arrives slightly before video, add a positive offset (delay the audio). If audio is behind, use a negative offset.
Start with small adjustments (±50ms) and do test recordings to verify.

For voice-only NDI audio sources (audio-only NDI stream), the sync offset needs to match the delay of your game capture source — this is why many streamers prefer the single Main NDI Output that carries both audio and video together.

Firewall and Network Configuration

NDI uses multicast UDP and TCP on your LAN. If your gaming PC and streaming PC cannot discover each other, a firewall is blocking NDI traffic.

On both PCs:

Open Windows Defender Firewall → Allow an app through firewall.
Add OBS Studio and NDI Runtime to the allowed list for Private networks.
Alternatively, temporarily disable the firewall to test connectivity — if NDI works without the firewall, you know it is the issue.

NDI uses port 5960 for TCP connection management and dynamic UDP ports for data. If you use a third-party firewall (Malwarebytes, Bitdefender, etc.), add the same exceptions there.

Subnet check: Both PCs must be on the same LAN subnet (e.g., 192.168.1.x). NDI multicast discovery does not cross router boundaries by default. If your gaming PC is on a wired connection (192.168.1.x) and your streaming PC is on Wi-Fi (192.168.2.x via a different router), NDI sources will not appear in the dropdown — use NDI Access Manager to add the gaming PC’s IP address manually.

NDI 5 vs NDI 6: What Actually Changed for Audio

Both versions carry uncompressed or lightly compressed audio perfectly fine for streaming. The practical differences:

Feature	NDI 5	NDI 6
CPU usage (video)	Moderate	Lower (~15% reduction)
Audio quality	Uncompressed PCM	Uncompressed PCM
HX3 compression	Not available	Available
OBS plugin support	obs-ndi 4.x	obs-ndi 5.x
Backward compatibility	NDI 5 tools	NDI 6 tools break NDI 5

For audio-only NDI voice changer use, NDI 5 and NDI 6 are effectively identical in quality. The upgrade to NDI 6 matters if your setup also sends 4K or high-framerate video and CPU overhead is a concern.

Important: You cannot mix NDI 5 tools on one PC with NDI 6 tools on the other. Upgrade both PCs simultaneously or stay on NDI 5.

Integrating Voice Changer Effects for Streaming Entertainment

Once NDI audio is flowing cleanly, the streaming PC’s OBS receives your processed voice exactly as if it were a local microphone. All the entertainment techniques that work in a single-PC setup work identically here:

Hotkey-bound voice switching: Assign different VoxBooster presets to keyboard shortcuts. When you switch effects on the gaming PC, the change propagates to the streaming PC’s NDI audio source instantly — the streaming audience hears the switch in real time with no additional latency.

Soundboard integration: VoxBooster’s integrated soundboard routes through the same virtual mic output that NDI carries. Fire a soundboard clip on the gaming PC; the streaming PC receives it mixed with your voice in the NDI stream. No separate audio cable needed.

AI voice cloning during streams: Heavier AI voice processing (250–500ms latency) is completely viable in this setup because you’re already isolating the gaming PC. Viewers hear it through Twitch’s 5–10 second buffer, making the latency invisible. For more on what AI voice cloning can do in a professional context, see our dedicated guide.

Live captions via speech-to-text: VoxBooster includes Whisper Large-v3 speech-to-text. In a two-PC setup, captions generated on the gaming PC can be sent to the streaming PC via a text overlay tool or browser source over LAN, keeping the streaming PC’s workload minimal.

For a deeper look at general voice changer for streaming concepts independent of the NDI topology, see our full streaming guide. If you also use Discord on the gaming PC while streaming, the same virtual mic setup covered here applies — read our voice changer for Discord guide for the Discord-specific configuration steps.

Troubleshooting Common NDI Audio Issues

Problem: Streaming PC does not see gaming PC’s NDI sources

Verify both PCs are on the same subnet. Open NDI Access Manager on the streaming PC and manually add the gaming PC’s IP address.
Check that OBS NDI Output is enabled on the gaming PC (Tools → NDI Output Settings → Main Output checkbox).
Temporarily disable Windows Firewall on both PCs to test. If NDI works, add permanent exceptions.

Problem: Audio arrives on streaming PC but is the raw unprocessed voice

In OBS on gaming PC, confirm Mic/Aux is set to the VoxBooster virtual mic, not the physical microphone.
Open Windows Sound → Recording. Verify VoxBooster Virtual Mic device shows audio activity when you speak.

Problem: Audio glitches or dropouts in NDI stream

Increase the NDI buffer compensation on the streaming PC’s NDI Source settings (try 200–300ms).
Check network: run a continuous ping from streaming PC to gaming PC (ping -t 192.168.1.x). Packet loss or latency spikes above 5ms indicate a network problem. Switch to a direct Ethernet cable if on Wi-Fi.

Problem: Audio is in sync but has a consistent offset from video

Use OBS Advanced Audio Settings on the streaming PC to add a Sync Offset to the NDI Source. Adjust in 50ms increments until audio and video match in test recordings.

Problem: NDI causes OBS to crash on gaming PC

NDI plugin version mismatch with OBS Studio version. Download the obs-ndi release that matches your OBS Studio build. Check the obs-ndi GitHub releases for the correct version matrix.

Performance Notes: Gaming PC CPU/GPU Impact

Running OBS NDI Output on the gaming PC adds a small encoding overhead:

NDI Main Output (full quality): adds approximately 5–10% CPU overhead for NDI encoding, depending on resolution and frame rate.
NDI audio-only: negligible CPU overhead — essentially free.

For gaming PCs where CPU headroom is tight, consider:

Lower the NDI Output resolution (720p at 30fps is enough for monitoring/audio purposes when the streaming PC does the final encode).
Use the dedicated audio-only NDI source instead of full video NDI Output.
Lower VoxBooster’s AI processing thread count if the voice changer also competes for CPU.

If you use hardware like the ATEM Mini Pro for video switching in your production, NDI audio from the gaming PC integrates cleanly — the streaming PC can receive both the ATEM HDMI feed and the NDI audio source from the gaming PC and mix them in OBS.

Frequently Asked Questions

What is NDI and why use it for a two-PC voice changer setup?

NDI (Network Device Interface) is a royalty-free LAN protocol by NewTek that carries high-quality audio and video between devices on the same network. In a two-PC stream setup, NDI lets the streaming PC receive the processed audio from the gaming PC without a capture card — the voice changer output travels over Ethernet at under 1ms additional network latency.

Does NDI add noticeable latency to the voice changer output?

On a gigabit LAN, NDI 5 and NDI 6 add approximately 1–3ms of network latency on top of your voice changer’s own processing delay. A DSP voice effect running at 8–12ms total on the gaming PC arrives at the streaming PC in under 15ms combined — well inside the real-time threshold.

Can I use a wireless connection for NDI multi-machine audio?

Technically yes, but avoid it for production. Wi-Fi introduces variable latency and occasional packet drops that cause audio glitches in OBS. Use a gigabit Ethernet cable between gaming PC and streaming PC. If cabling is impossible, a dedicated 5 GHz Wi-Fi 6 router with both PCs on the same band is the minimum acceptable fallback.

Do I need a capture card for a two-PC OBS NDI setup?

No. NDI replaces the capture card for both video and audio transmission between PCs over LAN. The gaming PC sends its full desktop or game capture as an NDI source; the streaming PC receives it in OBS as an NDI Source plugin input. The audio from the voice changer travels embedded in the same NDI stream or as a dedicated NDI audio-only source.

Which OBS NDI plugin version should I install?

Install the obs-ndi plugin from obs-ndi.io — the version matching your OBS Studio release. As of mid-2026, obs-ndi 4.x supports NDI 5 SDK and obs-ndi 5.x supports NDI 6 SDK. Both work for audio and video. Check the GitHub releases page for the latest build; avoid unofficial forks.

Does a voice changer conflict with NDI audio capture in OBS?

No, as long as the voice changer outputs to a virtual microphone device or WASAPI loopback that OBS NDI Output captures on the gaming PC side. The key is selecting the correct audio source in OBS NDI Output settings — pick the virtual mic (processed voice), not the physical mic (raw voice).

What is the difference between NDI 5 and NDI 6 for streaming audio?

NDI 6 introduced improved codec efficiency and lower CPU overhead compared to NDI 5. For audio-only transmission the practical difference is minimal. NDI 6 also added better support for low-bandwidth NDI|HX3 encoding. Unless your streaming PC struggles with CPU load, either version works well; upgrade to NDI 6 if your plugins support it.

Conclusion

An OBS NDI voice changer configuration across two PCs is more straightforward than most documentation makes it seem. The core principle is simple: voice changer runs on the gaming PC, outputs to a virtual microphone, OBS captures that virtual mic and sends it over NDI to the streaming PC. NDI 5 and NDI 6 both carry the audio cleanly over a gigabit LAN connection with under 3ms of added network latency — a non-issue for any real-time voice application.

The parts that trip people up are almost always the same: selecting the virtual mic (not the physical mic) in OBS audio settings on the gaming PC, and firewall rules blocking NDI discovery. Get those two right and the rest falls into place.

VoxBooster handles the gaming PC side of this setup out of the box. It creates a WASAPI virtual microphone that OBS NDI Output captures automatically, supports both DSP effects (under 15ms) and AI voice models, and includes a built-in soundboard that routes through the same NDI audio path — no extra virtual audio cable routing needed. Try it free for 3 days with no credit card required.