For quite some time now, I have been streaming and recording our sessions and activities with OBS Studio. It makes it very easy to consolidate and pre-configure a great many things, and it is very compatible with my Zoom H6 audio recorder I use to capture all the audio. However, sometimes it is necessary to have some players remote, and for that we utilize Discord. When that happens, I routinely receive complaints about poor audio and difficulty hearing, despite numerous attempts to fix it. This latest Extinction Curse session was a good example, much frustration during the storyline, which made an already stressful combat worse.
Today I think I finally solved this issue. Fundamentally, the issue is a compatibility concern – I use the H6 to record the audio in independent tracks. That allows me to adjust the volume independently, but also do post-processing and editing if needed after the game. This works great with OBS, which accepts an ASIO (multi-track) input fairly easily. Discord however doesn’t like that format, and barely tolerates sharing the audio with another application.
Originally, I tried Voicemeter as a way to either re-direct the audio, or otherwise mix it down to the stereo Discord needed. All attempts ended up with garbled, staticky and otherwise messy audio. I simply had no way to isolate the audio to two different destinations.
Finally, I figured out a solution though – if I could not redirect the stream at the PC level, let me redirect it at the Zoom H6 level. Primary audio now goes out thru the USB connection as multi-track audio, strong and clear. I then activated Direct Monitor and connected a basic audio cable from the Zoom output to the laptop input, selecting it as a microphone. At that point, I can tell Discord to use the onboard microphone, while OBS uses the multi-track input. Two apparent sources, split at the zoom. End result is strong and clear audio on both sides, without distortion or crackling.
Fingers crossed this solves the problem long term, but if anyone else has a similar issue, this may be a workable solution for you instead.
If anyone knows of a more direct solution, potentially using Voicemeter or something else, let me know.
