Voice Changer for OBS Studio 31: Full Setup Guide

The OBS 31 voice changer setup is cleaner than ever, and the new features in this release — WHIP/WHEP low-latency streaming, AV1 hardware encoding, and an improved noise suppression filter — each interact with your audio chain in ways worth understanding. This guide covers every integration point: how to route a real-time voice changer into OBS 31, how to use Audio Monitoring Devices correctly, which new OBS 31 audio filters to stack and which to skip, and how the new streaming protocols affect your transformed voice delivery.

TL;DR

OBS Studio 31 has no built-in voice changer — you route one in via a virtual microphone source.
The cleanest integration: VoxBooster virtual mic → OBS Mic/Aux source → filters chain → stream.
OBS 31’s Audio Monitoring Devices let you hear your transformed voice in real time without feedback.
AV1 hardware encoding reduces CPU load, giving your voice changer more headroom.
WHIP/WHEP (WebRTC) in OBS 31 cuts stream latency to sub-second — voice processing runs upstream and is unaffected.
OBS 31’s improved noise suppression is good but redundant if your voice changer already handles noise.

What Changed in OBS Studio 31 for Audio

OBS Studio 31, released in late 2025, shipped a set of updates that meaningfully improve the streaming audio experience. The headline video feature — AV1 hardware encoding support for AMD, Intel, and NVIDIA — gets most of the press, but several audio-relevant changes matter more for voice changer users.

WHIP/WHEP support (WebRTC): OBS 31 added native WHIP (WebRTC HTTP Ingestion Protocol) for sending streams and WHEP (WebRTC HTTP Egress Protocol) for receiving. WHIP targets platforms like Cloudflare Stream, Dolby.io, and custom WebRTC ingest servers. The practical effect for streamers: end-to-end latency can drop below one second, compared to 5-30 seconds with standard RTMP. Your voice changer integration is unaffected by the output protocol — it processes audio before it ever reaches OBS — but lower latency matters when your audience is interacting with you live.

Improved noise suppression filter: OBS 31 overhauled the built-in noise suppression filter, with better RNNoise and Speex modes and a new DeepFilter-based option that removes background noise more aggressively without the “metallic” artifacts of earlier versions. This is relevant because many streamers used to stack Krisp or RTX Voice as a separate VST — OBS 31’s native filter is now competitive.

Audio Monitoring Devices improvements: OBS 31 made it easier to assign per-source monitoring output devices. This is the most directly useful change for voice changer setups, covered in detail below.

AV1 hardware encoding: Available on RTX 40-series (NVIDIA), RX 7000-series (AMD), and Intel Arc. This offloads video encoding to dedicated hardware, freeing CPU cores. If you run a CPU-intensive voice changer or AI voice cloning model alongside OBS, this headroom matters.

How OBS Studio Handles Audio Sources

Before setting up a voice changer, it helps to understand OBS 31’s audio graph. OBS treats audio sources as independent tracks that feed into the stream mix and optionally into recordings.

The key source types:

Mic/Aux — a direct device input (your microphone, or in this case, a virtual microphone device)
Desktop Audio — loopback capture of Windows audio output
Application Audio Capture — per-app audio (available since OBS 27)
Browser source audio — inline audio from browser widgets

For voice changer integration, you are working entirely within the Mic/Aux path. The voice changer sits upstream of OBS in the Windows audio graph, so OBS sees a virtual microphone device and treats it identically to a physical microphone.

The Signal Chain

Understanding the correct signal chain prevents most setup problems:

Physical mic
    ↓
Voice changer (VoxBooster)
    ↓
VoxBooster Virtual Microphone (Windows audio device)
    ↓
OBS Studio 31 — Mic/Aux source
    ↓
OBS filter chain (optional: noise gate, compressor)
    ↓
Stream mix → RTMP / WHIP / SRT output

The voice changer processes audio in Windows before it reaches OBS. OBS never sees your unmodified voice — it sees the already-transformed output. This is important because it means OBS filters applied to the source (noise gate, EQ, compressor) operate on the processed voice, not the raw signal.

Installing VoxBooster and Creating the Virtual Mic Source

VoxBooster runs on Windows 10 and Windows 11. After installation, it registers a virtual microphone device in the Windows audio system without installing a kernel-level driver. This keeps it compatible with anti-cheat systems in games you might be streaming.

Step 1 — Install and launch VoxBooster. Download from voxbooster.com/download. The 3-day trial requires no credit card. Launch the application; it will appear in the system tray.

Step 2 — Select a voice or effect. In VoxBooster, choose your voice preset from the Voice Effects or AI Voice Clone tab. Enable the “Real-time” toggle to start processing your microphone input.

Step 3 — Open OBS Studio 31. Go to Settings > Audio. Set Sample Rate to 48000 Hz and Channels to Stereo. These match VoxBooster’s output format and prevent sample-rate conversion artifacts.

Step 4 — Add the virtual mic as a source. In the OBS Sources panel, click the + button and select Audio Input Capture. In the device dropdown, select VoxBooster Virtual Microphone. Name the source something descriptive — “Voice Changer” works.

Step 5 — Test the audio. Speak into your microphone. You should see the audio meter for the Voice Changer source moving in OBS. In the Audio Mixer, confirm the source is active and at an appropriate level (-12 to -6 dBFS peaks are ideal for streaming).

Step 6 — Optional: mute the default Mic/Aux. If OBS has your physical microphone set as the default Mic/Aux source, mute it in the Audio Mixer so you don’t send both the raw and transformed signal. Or remove it from the scene entirely.

Configuring Audio Monitoring Devices in OBS 31

One of the most useful OBS 31 improvements is the refined Audio Monitoring system. This lets you route individual audio sources to a separate output device — so you can hear your transformed voice in your headphones while your audience hears it through the stream, without creating a feedback loop.

Setting Up a Monitoring Output

In OBS, go to Settings > Audio.
Find the Monitoring Device dropdown at the top. Select your headphones or a dedicated monitoring output device. Do not select your speakers if your microphone can hear them — that creates a feedback loop.
Click OK to save.

Enabling Monitoring on Your Voice Changer Source

Open the Advanced Audio Properties panel: Edit > Advanced Audio Properties or right-click the Audio Mixer and select Advanced Audio Properties.
Find your Voice Changer source row.
In the Audio Monitoring column, change from “Monitor Off” to “Monitor and Output”.
Now speak. You hear your transformed voice through your monitoring headphones, and the stream receives it normally.

“Monitor Only (mute output)” is useful during testing — it mutes the stream while you listen to the effect. Switch back to “Monitor and Output” before going live.

Why This Matters for Voice Changer Setups

Without monitoring, you are performing blind — you cannot hear whether your voice effect sounds as intended during a live stream. The OBS 31 improvements make it easier to assign dedicated monitoring devices per source, so you can monitor your voice change, your game audio, and your music sources on separate outputs if needed.

OBS 31 Noise Suppression: Stack or Skip?

OBS Studio 31’s improved noise suppression filter uses DeepFilter, a deep learning model that runs efficiently on CPU. It is genuinely better than the previous RNNoise implementation for removing mechanical keyboard noise, air conditioning hum, and mouse clicks.

The question for voice changer users: should you use OBS 31’s noise suppression in addition to your voice changer’s built-in noise suppression?

The answer depends on your setup:

Scenario	Recommendation
Voice changer with noise suppression enabled	Skip OBS 31 noise suppression — redundant, adds CPU load
Voice changer without noise suppression	Add OBS 31’s noise suppression as a Mic/Aux filter
Noisy environment, maximum quality priority	Enable voice changer noise suppression first, then add OBS 31 as a light second pass
CPU-limited system (streaming + gaming)	Choose one: either voice changer’s or OBS 31’s noise suppression

VoxBooster runs noise suppression before voice transformation in the processing chain, which is the optimal order — clean audio produces better voice effects. If you add OBS 31’s filter on top, you are applying it to already-transformed audio, which is less effective and may introduce phase artifacts.

To add OBS 31’s noise suppression to your Voice Changer source:

Right-click the source in the Sources panel → Filters.
Click the + in the Audio Filters section → Noise Suppression.
Select DeepFilter method. Set the suppression level to 20-30 for subtle use alongside another tool, or 50+ if using it solo.

AV1 Hardware Encoding: What It Means for Your Audio

OBS Studio 31’s AV1 hardware encoding is primarily a video story, but it has a real impact on CPU-intensive voice changer setups.

Why it matters: Software x264 encoding — the traditional OBS default — can consume 20-40% of your CPU on a mid-range processor, even at medium presets. If you run an AI-powered voice changer simultaneously, CPU contention causes audio buffer underruns, which appear as audio dropouts, glitches, or latency spikes.

AV1 hardware encoding on a compatible GPU offloads encoding entirely from the CPU. On an NVIDIA RTX 4060 or AMD RX 7700, hardware AV1 encoding uses near-zero CPU. The freed cores are available for your voice changer’s real-time processing.

How to enable AV1 in OBS 31:

Go to Settings > Output.
Set Output Mode to Advanced.
In the Streaming tab, select your encoder:
- NVIDIA: NVENC AV1
- AMD: AMD HW H.265 or AV1 (GPU-dependent)
- Intel: Intel AV1 (QSV)
Set Quality Preset to Quality or Balanced. Use Rate Control CBR at 4000-8000 kbps depending on your upload bandwidth.

Note: AV1 is only useful if the platform you stream to accepts it. Twitch and YouTube both support AV1 as of 2025. RTMP to most other platforms may require H.264 fallback.

With hardware encoding active, you will likely notice more stable audio performance from your voice changer, especially during CPU spikes (loading screens, shader compilation, complex game moments).

WHIP/WHEP and Low-Latency Streaming with a Voice Changer

WHIP (WebRTC HTTP Ingestion Protocol) is the new streaming standard supported in OBS 31. It enables sub-second end-to-end latency compared to RTMP’s 5-30 second glass-to-glass delay.

For voice changer users, the practical implications:

What changes: Your transformed voice reaches your audience nearly instantly. If you stream interactive content — live Q&A, collaborative games, audience-participation streams — the sub-second latency eliminates the awkward “they heard me but I can’t tell” delay.

What does not change: Your voice changer integration. VoxBooster processes audio before it reaches OBS, and the stream output protocol has no effect on that upstream processing. Whether you stream via WHIP, RTMP, or SRT, the voice changer works identically.

How to set up WHIP output in OBS 31:

Go to Settings > Stream.
Set Service to Custom.
In the Server field, enter your platform’s WHIP ingest URL. Cloudflare Stream example: https://live.cloudflare.com/webrtc/[your-stream-id]/whip
Enter your Bearer Token in the Stream Key field.
Click OK. Start streaming.

WHIP requires a platform that exposes a WHIP endpoint. Cloudflare Stream, Dolby.io, and self-hosted Janus/Mediasoup servers support it. Twitch and YouTube still use RTMP as their primary ingest — use their standard settings for those platforms.

Comparing Voice Changers for OBS 31

Not all voice changers integrate equally well with OBS Studio 31. The critical requirement is that the tool presents a standard Windows audio input device — OBS reads this natively, no plugins needed.

Tool	Integration Method	Real-Time	Kernel Driver	OBS 31 Compatible
VoxBooster	Virtual Mic device	Yes	No	Yes
Voicemod	Virtual Mic device	Yes	Yes (kernel)	Yes
Clownfish Voice Changer	System-level hook	Yes	Yes	Limited
MorphVOX Pro	Virtual Mic device	Yes	No	Yes
Voice.ai	Virtual Mic device	Yes	No	Yes
NVIDIA RTX Voice	Virtual Mic + APO	Yes (noise)	No	Yes (noise only)

Voicemod’s kernel driver installation is the most common point of concern among gamers who stream — some anti-cheat systems flag it. VoxBooster uses WASAPI without a kernel driver, which avoids this entirely.

For a deeper comparison of real-time voice changers for streaming, see our voice changer for streaming guide. If you stream games competitively and need to avoid driver conflicts, the best voice changer for gaming guide covers anti-cheat compatibility in detail.

Advanced OBS 31 Audio Filter Chain for Streamers

Once your voice changer source is in OBS 31, you can add a filter chain to polish the output further. Here is a recommended chain for live streaming:

Filter order (applied top to bottom in OBS):

Noise Gate — cuts audio below a threshold, prevents background noise between sentences. Set Close Threshold to -40 dB, Open Threshold to -30 dB. This prevents the transformed voice from picking up ambient noise during silences.
Compressor — evens out volume dynamics. Set Ratio 3:1, Threshold -18 dB, Attack 1ms, Release 60ms, Output Gain +2 dB. Streaming benefits from consistent loudness levels.
Limiter — hard ceiling at -1 dBFS to prevent any clipping. Essential when doing loud reactions or excited commentary.

What to skip:

Noise Suppression — redundant if your voice changer handles it upstream
Pitch Shift (OBS plugin) — if you are already using a voice changer, adding pitch shift in OBS doubles the processing and degrades quality
VST plugins — unless you have a specific need (reverb room, specific EQ), they add latency and CPU overhead without clear benefit over built-in filters

The goal is a minimal, clean filter chain. Your voice changer does the heavy lifting upstream; OBS just needs to gate, compress, and limit.

Troubleshooting Common OBS 31 Voice Changer Issues

Voice changer source shows no audio in OBS

Check that VoxBooster is running and real-time mode is enabled. In Windows Sound settings (right-click the speaker icon → Open Sound settings), confirm the VoxBooster Virtual Microphone appears as an enabled recording device. If it is disabled, right-click it and choose Enable.

In OBS, right-click the Voice Changer source and confirm it is not muted. Check Advanced Audio Properties to confirm the source is not set to “Monitor Only (mute output).”

Audio drops out during streams

This is usually a buffer underrun caused by CPU contention. Solutions in order of preference:

Enable AV1 or HEVC hardware encoding in OBS 31 to free CPU
Increase the audio buffer size in VoxBooster settings to 256 or 512 frames (adds 5-10ms latency but prevents dropouts)
Close background applications — browsers with many tabs, update services, background game launchers
In OBS, go to Settings > Advanced and check if Audio Monitoring Device is on a USB device that disconnects — switch to a stable internal device

Transformed voice sounds doubled or has an echo

Your physical microphone is likely active as a second audio source. Check the OBS Audio Mixer for a default Mic/Aux source — mute it or remove it from the scene. Also check Windows Sound settings to confirm your physical microphone is not set as the system default for communication devices and being picked up separately.

OBS 31 noise suppression interferes with voice effect

The DeepFilter model in OBS 31 can misidentify artificial voice-changer output as noise and attempt to suppress it. If you hear the voice effect being degraded by OBS’s filter, remove the OBS noise suppression filter from the Voice Changer source. Your voice changer’s upstream noise suppression is the better place to handle it.

Voice Changer Presets That Work Well on Stream

Not all voice effects translate well through a streaming audio codec. AAC and Opus — the two most common streaming audio codecs — compress high-frequency content aggressively above 16 kHz. Effects that rely on extreme high-frequency shimmer or very narrow formant peaks may sound degraded to viewers even if they sound clear in your monitoring headphones.

Effects that stream well:

Pitch-shifted voices (deep, high, robotic) — the fundamental frequencies are in the 80-2000 Hz range, which codecs handle cleanly
Narrator / announcer effects with added reverb — the reverb tail compresses well
Character voices with formant shifts — the primary information is in the 300-3500 Hz range

Effects that can degrade:

Very high-pitched voices shifted above 1000 Hz fundamental — codec artifacting becomes noticeable
Heavy metallic/ring-modulation effects — the intermodulation products are at frequencies that get compressed
Extreme reverb with long tails — codec struggles with diffuse high-frequency decay

If you are using VoxBooster’s AI voice cloning mode, the neural conversion targets the 80-8000 Hz range where codecs perform well, so those voices typically stream cleanly.

For production-level streaming setups, see our voice changer for vMix professional guide and voice changer for Wirecast live guide if you use those platforms alongside OBS. If you are choosing between real-time voice changers and NVIDIA Broadcast specifically, voice changer vs NVIDIA Broadcast covers that comparison in depth.

Frequently Asked Questions

Does OBS Studio 31 have a built-in voice changer?

No. OBS Studio 31 does not include a voice changer. It added improved noise suppression, better audio monitoring, and WHIP/WHEP streaming — but voice modulation still requires an external tool like VoxBooster that presents a virtual microphone OBS can select as its audio source.

What is the best way to add a voice changer to OBS Studio 31?

Install VoxBooster, enable real-time processing, then open OBS Studio 31 and set your Mic/Aux source to the VoxBooster Virtual Microphone device. No additional plugins or routing cables required — OBS sees it as a standard audio input device.

Does OBS Studio 31’s AV1 encoding affect voice quality?

AV1 is a video codec — it handles video compression only. Your voice audio is encoded separately, usually as AAC or Opus. AV1 hardware encoding in OBS 31 improves video quality and reduces CPU load, which indirectly helps audio by freeing resources for real-time voice processing.

How does WHIP/WHEP in OBS 31 work with a voice changer?

WHIP (WebRTC HTTP Ingestion Protocol) is an ultra-low-latency streaming protocol in OBS 31. Your voice changer runs upstream of OBS — it processes your mic input and feeds a virtual mic into OBS. The output protocol (WHIP, RTMP, SRT) doesn’t affect how the voice changer integrates.

Can I use OBS Audio Monitoring with a voice changer in OBS 31?

Yes. Set your voice-changed source to Monitor and Output in OBS 31’s advanced audio settings. This lets you hear your transformed voice in headphones while streaming it to your audience. Use a separate monitoring output device to avoid feedback loops.

Does OBS Studio 31’s noise suppression work with a voice changer?

It can, but stacking both is often unnecessary. VoxBooster’s built-in noise suppression runs before voice transformation, which is the correct signal chain order. If you add OBS 31’s noise suppression filter on top, apply it after the voice source enters OBS — not as a pre-filter on the microphone.

What OBS audio settings give the best voice changer quality?

In OBS 31, go to Settings > Audio and set Sample Rate to 48 kHz, Channels to Stereo. For the Mic/Aux filter chain, keep it minimal — your voice changer handles processing upstream. Avoid stacking multiple noise suppression filters, which can cause phasing artifacts.

Conclusion

OBS Studio 31’s new features — WHIP/WHEP low-latency output, AV1 hardware encoding, and improved noise suppression — each give voice changer setups a meaningful upgrade. AV1 frees CPU for your real-time voice processing. Audio Monitoring Devices let you hear your transformed voice while streaming. WHIP brings your voice to your audience in under a second.

The integration itself remains straightforward: voice changer → virtual mic device → OBS audio source → filter chain → stream. Understanding the signal chain order, keeping the OBS filter chain minimal, and using hardware encoding when available are the three things that most improve voice changer quality in OBS 31.

If you want a voice changer that slides cleanly into this pipeline — no kernel driver, no routing cables, 3-day free trial — VoxBooster is built specifically for this use case on Windows 10 and 11. It runs at sub-20ms latency, handles noise suppression upstream of OBS, and works with every streaming platform OBS 31 supports.

Download VoxBooster — free 3-day trial, no credit card required.