Voicemeeter soundboard: play voice clips into Discord
Voicemeeter Banana is a free (donationware) virtual audio mixer for Windows. It lets you mix your microphone with sounds from any app, so friends in Discord or a game hear both your voice and your soundboard. Fill the soundboard with spoken lines, such as a game announcer, a cartoon character or your own trained voice, and it becomes a voice board.
This guide follows the VB-Audio manual for the routing and shows how to make the clips with the free SuperClonica engine.
How Voicemeeter’s virtual devices work
According to the VB-Audio manual, Voicemeeter installs virtual audio devices: the playback devices, such as Voicemeeter Input and Voicemeeter Aux Input, are its virtual inputs, and the recording devices are fed by the B buses; the one fed by bus B1 is called Voicemeeter Out B1. Banana has three hardware inputs, two virtual inputs and buses A1 to A3, B1 and B2, and every input strip can be sent to any bus.
Route the soundboard and your mic
Set your soundboard app to play into Voicemeeter Input (or Voicemeeter Aux Input). In Voicemeeter, select your microphone on a hardware input strip. Send both the mic strip and the virtual input strip to B1, and to A1 if you want to hear the sounds yourself on your headphones. Then in Discord, open the voice settings and choose Voicemeeter Out B1 as the input device. The manual’s installation steps end with exactly that: reboot, then check the device selection in apps such as Discord, Teams or Zoom.
No soundboard app? Use the tape deck
Banana has an integrated player and recorder, the tape deck, which can play WAV, MP3, M4A and other files with one click. It is handy for a single clip or a quick test; for many clips on hotkeys, a soundboard app or a Stream Deck is easier.
Make the spoken clips
Hear voices free on their pages with a line of 30 to 100 characters, no account needed, and download the recording. For a full board, buy a voice once, download its file and install the free engine with python -m pip install superclonica-cpu, then voice each line: superclonica --text "Objective secured!" --voice voice.json --lang en --normalize --output objective.wav. The engine writes WAV, which Voicemeeter and soundboard apps open directly, and --normalize keeps every clip at the same loudness so none blasts your friends.
Keep it fair
Turn down a clip on its strip if it is louder than your voice, and keep Discord’s noise suppression in mind: it is made to keep your voice and can cut other sounds. Character voices in the catalog are fan-made, for fan works, parody and personal use; never use a voice to pretend to be a real person.