VB-Audio's Forums

Voicemeeter latency is pending on different parameters and the overall configuration of your PC. Since this optimal settings is not applicable to all PC Setup, you may have to experiment to find an optimal configuration with the best compromise between latency and audio stream stability.

------------------------------------------------------------------------------------
RECOMMENDATIONS TO REDUCE LATENCY
------------------------------------------------------------------------------------

FIRST Point to optimize latency in Voicemeeter is to select your best audio device as output A1 (giving the main audio stream). optimal latency is usually made by selecting an ASIO driver configured with a buffer size around 256 samples. if no ASIO Driver present, WDM/KS device can be used instead. Some of them could work with 256 sample buffer (but some not).

Use ASIO device in optimal way to get an optimal audio path from Microphone to output through Voicemeeter connected to an ASIO device, thanks to follow this guide: https://youtu.be/35N8hkOy8h4

VBAN latency optimization is also pending on Voicemeeter main stream buffer size (given by the output device A1) and the VBAN net quality parameter. Outgoing Stream are set in Optimal per default, but incoming stream are usually set in FAST or Medium Net Quality. if you are on local wire network, incoming stream can work with the NEt Quality set to Optimal. That will improve the latency on your VBAN link.

------------------------------------------------------------------------------------
NOT RECOMMENDED / AUDIO EXPERT ONLY
------------------------------------------------------------------------------------

For audio expert only, This video shows how to get smallest live latency (MIC -> Voicemeeter -> Speaker).
https://youtu.be/LAYKcIC5iFY
However the WDM-Exclusive mode is not available anymore by the Voicemeeter user interface. because this option is usually not stable in the time (due to bug in WASAPI it can generate crash or BSOD within a 48h session) and SWIFT Engine mode is experimental. In case of problem, we recommend to go back to default settings.

Finally, Virtual Audio I/O Latency is pending on specific internal buffer and also on Voicemeeter Main Stream Latency (output A1 device). For Voicemeeter VAIO, we recommend to keep internal buffer size around 4096 to 7168 samples. Reducing latency here is for expert and cannot be done without having read this technical note:
https://www.vb-audio.com/Cable/VBCABLE_ ... ttings.pdf

However the latency on virtual I/O is also pending on the application connected to it, and the audio interface used (WASAPI, Direct-X, WaveRT, Ks, MME). The Virtual ASIO interface add one buffer latency while MME interface can add 150ms.

I looked at the referenced post. I'm sorry, but the English in many of the instructions given in these forums is very difficult to understand. I'm doing the best I can to try to guess at the meanings sometimes. The YouTube video was easier to understand, but I had already matched the sample rates in my devices, and my Voicemeeter application does not allow me to change the engine to Exclusive or Swift mode. You need to release these functions for use if you are going to instruct people to use them. However I'm not sure that the WDM Exclusive mode would help me at the moment since I am trying to reduce latency in MME mode. The Swift mode might be worth a try. I do no know how to directly change the buffer size on the inputs.

NOTE: After experimenting, I did find that lowering the MME buffer size from 1024 to 576 helped quite a bit.
Update: Reducing the buffer worked at the time, but the first place I noticed problems was in Discord. Infact, there are a lot of problems with Discord and this software - they don't play nice together at all. I tried the suggestions in another thread about fixing issues with Discord dealing with changing the Windows audio process but found that this did not help me. I think this is mainly for PCs with slower processors or certain kinds of processors that have certain kinds of multi-threading issues.

So, I'm back to not having a solution to the delay issue.

xRYANx555 wrote: ↑Wed Jan 05, 2022 8:59 pm I tried the suggestions in another thread about fixing issues with Discord dealing with changing the Windows audio process but found that this did not help me. I think this is mainly for PCs with slower processors or certain kinds of processors that have certain kinds of multi-threading issues.

So, I'm back to not having a solution to the delay issue.

i still have to change the audiodg file to higher priority and select a cpu core, though i dont select the core based on those other posts, i look at which core has the least usage which generally is the higher cores then i select the primary of that logical core, otherwise discord is nasty, im running a ryzen 3900x, should be more than fast enough with plenty of cores to handle it, never been an issue on older versions, getting very annoying to still be doing this many versions on
this is only with discord and my mic, the rest of the sounds are fine

i also find from time to time i have to switch from wdm to mme due to latency or i think buffering issues, then i seem able to switch them back when that stops working, honestly no clue why, windows 11 did add in an annoying issue regarding bitrate and hz, do i now have to run all my audio devices at default allowing windows to take exclusive control, or i end up with painful quality on my speakers with lots of artifacts and latency issues, hoping this gets resolved soon

i still have to change the audiodg file to higher priority and select a cpu core, though i dont select the core based on those other posts, i look at which core has the least usage which generally is the higher cores then i select the primary of that logical core, otherwise discord is nasty, im running a ryzen 3900x, should be more than fast enough with plenty of cores to handle it, never been an issue on older versions, getting very annoying to still be doing this many versions on
this is only with discord and my mic, the rest of the sounds are fine

i also find from time to time i have to switch from wdm to mme due to latency or i think buffering issues, then i seem able to switch them back when that stops working, honestly no clue why, windows 11 did add in an annoying issue regarding bitrate and hz, do i now have to run all my audio devices at default allowing windows to take exclusive control, or i end up with painful quality on my speakers with lots of artifacts and latency issues, hoping this gets resolved soon

I found this very interesting. I had been just picking core #2 like it said in the video, since I figured the core number really didn't matter as long as it was just assigned to one single core.
MME has been a complete no-go for me with Discord. Anytime I've ever tried to use MME at all with Discord, people tell me they can't understand me at all on it. And I always go in Discord settings and switch off all the bells and whistles it tries to throw at your audio (like noise codecs, gain control, etc.) so that none of that interferes with it because all that is supposed to be handled by Voicemeeter before it ever gets sent to Discord anyway. I've been able to use WDM fine with it.

My issue is the lag I get in general between talking into my mic and when I hear it loop back to me, say by directly routing it to A1 or monitoring my B1 output. If I try to do it, the lag will actually make me slur up my speech from the effect of hearing my own lagged speech come back to me. And forget being able to sing along to music.
MME is worse than WDM with this. I will say that MME has a nicer sound to it. Too bad it doesn't work well at all.
I tried an XLR mic setup and routed it into my motherboard's Line In input via the Realtech driver instead of using my normal USB mic. I was hoping it would show up as an ASIO input, but it still shows up under WDM or MME.