Voicemeeter Banana actually works as a voice changer if you stop treating it like magic and start treating it like routing software.
The way people talk about this thing makes it sound like you install it and suddenly sound like a robot or a demon. That's not how it works. Voicemeeter Banana is primarily an audio routing and mixing application. The voice changing capability comes from effects you route through it, usually via a second program or its built-in pitch modifier. I've been running this setup for about four years across three different studios and a bunch of home rigs. The official download lives at biquad.com/voicemeeter. You want Banana, which is free. The regular version lacks the extra aux buses you need for proper voice routing. Download the installer, run it, and during setup pay attention to the driver selection screen. It will try to install two virtual ASIO drivers and two WDM drivers. Check all of them. If you skip the ASIO drivers your latency numbers will be garbage later. Windows usually asks for a reboot after. Do the reboot. There is a persistent myth that you need to buy a license. You do not. The free version has every feature relevant to voice changing. The paid versions remove watermark reminders and add some hardware integration that most people do not need.
Unlock The Magic Of Voicemeeter Banana Your Ultimate Voice Changer Guide
The actual voice changing pipeline looks like this: microphone into Voicemeeter, Voicemeeter routes that signal to a processing tool, the processed signal comes back through an auxiliary input, and you send it out to your streaming or call software. The trick is getting the loopback path right without introducing feedback or catastrophic latency. My typical chain runs the mic into Voicemeeter's input strip, through the built-in pitch shifter for basic shifts, then out to a physical output or into ReaVoice inside Reaper for anything more complex, then back into Voicemeeter's Auxiliary Input A1 so the destination application picks it up clean.
Basic Configuration Steps
Open Voicemeeter Banana and go to the Hardware tab at the top left. Set your output device under Out A1 to whatever your headphones or speakers are. This is non-negotiable. If you route the output to your speakers instead of headphones you will get an immediate feedback loop that will make your room sound like a broken intercom system. Nobody wants that. Set Out B1 to Voicemeeter Input. This is the virtual output that other programs will listen to. Navigate to the Input strip for your microphone. Make sure the STR button is lit so the signal gets sent to the hardware output. Turn the GAIN knob up to about 12 o'clock as a starting point. Below that you will see the PITCH knob. Turning this clockwise raises your voice pitch. Counter-clockwise lowers it. This is the simplest voice effect Voicemeeter offers and it works adequately for basic cartoon or chipmunk styles. The range is roughly plus or minus twelve semitones before it starts sounding noticeably artificial. Now open whatever application you are using for voice output. Discord, Zoom, OBS, Streamlabs, whatever. Go into that program's audio settings and set the input device to Voicemeeter Input. That is B1. Your microphone is no longer feeding the application directly. Voicemeeter is now the middleman, which means you control the signal before it ever reaches the listener.
Get the Full Details
Advanced Processing With REAPER And VST Plugins
For anything beyond a simple pitch shift you need a proper voice processing chain. I use REAPER as a backend and route Voicemeeter's output into it through the ASIO driver. Within REAPER I run a chain of plugins: a pitch correction module like GSnap or Wavelet for autotune-style effects, a compressor to even out dynamics, a noise gate to cut background sound, and sometimes a formant shifter for deeper character changes without just cranking the pitch knob. The ASIO buffer size determines your latency. At 128 samples you might get around ten milliseconds of delay, which feels instant. At 256 samples you are looking at twenty milliseconds, which is still acceptable for most streaming. At 512 and above people will notice the lag and it becomes annoying during conversation. Set your buffer based on how complex your plugin chain is. More plugins means you need a larger buffer or your audio will crackle. It is a constant balancing act between quality and responsiveness. I once spent three hours debugging why my voice would randomly drop out during a stream. The issue was that my noise gate threshold was set so aggressively that any breath below a certain volume got cut completely, and the gate was closing faster than my speaking rhythm. I lowered the threshold by about six decibels and adjusted the release time to 200 milliseconds. Problem solved. Gate settings are the first thing to check when your voice cuts in and out unpredictably.
Common Pitfalls That Will Waste Your Time
Sample rate mismatch is the most common problem I see. If your microphone runs at 48kHz and Voicemeeter is set to 44.1kHz the playback will sound slower and deeper, which ruins any pitch shifting you attempted. Check both your hardware device settings and Voicemeeter's sample rate under the Hardware tab. They should match exactly. Same goes for your DAW if you are running one. Another issue is mono versus stereo routing confusion. Your microphone is mono. Voicemeeter's input strips handle mono sources fine, but if you accidentally send a stereo signal through a pitch algorithm that expects mono the results sound hollow and phasey. Keep your mic inputs in mono mode unless you have a specific stereo recording setup, which most voice changers do not. Feedback loops are also worth mentioning. If you have your headphones unplugged and your speakers are on, Voicemeeter's output can feed back into your mic input through the room. The result is a piercing whine that damages your ears and your neighbors' patience. Always use headphones when testing voice change routing. It is the only safe way to verify the signal path before turning speakers up.
Limitations You Should Know About
Voicemeeter Banana alone does not contain advanced voice change algorithms. The built-in pitch shifter is functional but limited. It changes fundamental frequency without adjusting formants, which means lower pitches sound like a slowed recording rather than a naturally deeper voice, and higher pitches sound like you played a tape backward at double speed. For realistic voice transformation you need formant-aware processing, which requires a separate tool or plugin. There is also a hard ceiling on how much pitch shifting you can do before artifacts become unacceptable. Push more than four or five semitones in either direction and the audio starts sounding robotic in an unpleasant way. Formant shifting helps with the naturalness but introduces its own latency and CPU cost. If you need broadcast-quality voice effects you are better off using a dedicated voice changer like MorphVOX Pro or AV Voice Changer Software Diamond alongside Voicemeeter for routing. Voicemeeter handles the mixing and distribution side; it does not replace specialized voice processing engines. CPU usage scales with plugin count and buffer size decisions. A heavy REAPER chain with multiple VSTs can draw five to ten percent of a single core on modern hardware, which is manageable but adds up if you are also encoding video or running other CPU-intensive applications simultaneously. Monitor your CPU load before going live. A throttling CPU causes audio dropouts that are far worse than any voice effect sounding slightly off.
