Menu Close

Rejecting Background Noise

I recently upgraded to a new headset microphone. This was to reject background noise, specifically keyboard noise. There are a few interesting bits of this that I wanted to breakdown the pros and cons for anyone else who might be shopping for better audio recording.

Note: I am going to break some technical details down and will no doubt gloss over and simplify things. I encourage you to do some additional research and correct me in the comments.

My Requirements

  • Live conferencing with software like:
    • Zoom, GoToMeeting, GoToWebinar, Fuze, Skype, etc.
  • Audio recording with software like:
    • Audacity, Camtasia, Adobe Audition, etc.
  • Music listening (not a microphone requirement specifically, but related)
    • Mostly Spotify and YouTube

Over the years I’ve had a few different microphones. Including a number of USB Blue microphones like the Snowflake, Snowball, and Yeti. They are all pretty good microphones, and very popular. I’ve heard from other people they’ve had trouble with them, but I found them fine. From talking to audio professionals they tell me that Blue is focused on making “professional grade” consumer devices – in other words, they aren’t used by “professionals.” (I don’t want to be pedantic about what defines a “professional” here.)

Audio Interfaces

A little detail about the four different ways microphones interface. As you move down this list, audio quality improves.

  • Bluetooth – This is purely for convenience, and has the lowest sound quality (same is true for headphones). I’ve tried a few of these, but whenever you can use the same microphone or headphone via both Bluetooth and a wire, the wire sound quality is significant better. Bluetooth has a slight lag too.
  • 3.5 mm or 1/8″ jack (Unbalanced, aka the headphone jack) – This is the most common consumer grade audio interface for both headphones and microphones. Significantly better sound quality than Bluetooth. There are some subtle variations that are less common in consumer grade equipment, including slightly smaller (2.5 mm especially common with microphones) or larger plugs (1/4″), as well as RCA/phono. What they all have in common is that they are analog and unbalanced. Unbalanced means that they have two connectors per channel. When you are connecting to your computer you use the jack that goes into the sound chip or card on your computer.
  • USB Audio – The universal serial bus is truly universal. When you connect a microphone via USB it appears to the computer as a whole new sound device. This is because the USB microphone does its own analog to digital conversion. So it doesn’t matter what kind of sound hardware you have in your computer, you are completely bypassing it. You are however dependent on the quality of the converter in the microphone, so it may be better or worse than the build in audio chip in your computer.
  • XLR, TS, or TRS (Balanced) – This is where you enter professional territory. TS and TRS may look a lot like larger 1/4″ headphone jacks, but they are different in that they are balanced. This means they use three wires per channel: positive, negative, and ground. Balanced audio cables resist interference and impedance mismatch (that hum when you have two devices connected via unbalanced audio cables). They also support phantom power microphones. I don’t know that I’ve ever seen a computer that supported these directly, so you will need a mixer to interface with them. This can mean running your mixer in through the 3.5 mm jack, or there are USB Mixers.

There is more, for example you can combine the headphone and microphone into a single plug, and most XLR are XLR3 (3 pins) but there are versions with more pins! Additionally, there are lavalier microphones (aka lapel or body mic) with wireless transmitters. These are a type of XLR microphone – essentially another layer on top of an XLR balanced audio system to hide the microphone.

In my opinion, if you want professional quality audio, then you want balanced cables and this means you want a USB Mixer and XLR connectors on your microphone.

But there are still a few more details. I’ve used two different XLR microphones. Both are Audio-Technica and both are good, but I’ve found they serve two different purposes.

My first was an AT2050 Multi-pattern Condenser Microphone. Condenser microphones are more sensitive. It also requires phantom power, which I’ve heard described as producing a “warmer” sound. Either way, it is a great microphone, but because it is more sensitive I found it picked up too much background noise, like keyboard, computer fan noise, or the air conditioner. If I had a dedicated, sound proofed, audio studio, then this would be a great solution.

The new microphone I just picked up is the BPHS1 Broadcast Stereo Headset. While technically it is a “headset,” it is better to think of it as a dedicated microphone attached to headphones. Usually headsets lack the quality of a dedicated microphone, and I’ve used a lot of headsets that this was the case with. These however exceeded my expectations. It is a dynamic microphone instead of a condenser microphone, which is typically less sensitive. It also does not require a phantom power mixer. It does a fantastic job of only recording my voice and mostly ignoring the other noises. Both microphones are cardioid, which allows them to reject background noise, but the BPHS1 does a better job rejecting background noise. You can still hear the keyboard (despite being a “quiet” model with Cherry MX Brown switches), but now it doesn’t overpower my voice. Oh, and the headphone audio quality is fantastic too.

I created a video where you can hear both keyboards with the same settings on my mixer while I type.

AT2050 vs BPHS1

Both are great microphones, but without a dedicated, sound proof, recording studio I don’t think the AT2050 is the right solution for me. The BPHS1 still has great sound quality, but it sounds better without a studio.

Software Background Noise Rejection

Right after I got the BPHS1 I discovered a pure software solution for background noise rejection called Krisp by 2Hz. It uses a neural network to filter background noise – and it is really, really effective!

  • 2Hz overview of the technology and their SDK
  • Krisp is the consumer software solution ($20/mo)
  • Nvidia has a blog post on their technology
  • AppSumo has a limited time, lifetime subscription deal for $39 (but only for new users that don’t have a trial account)

I discovered Krisp via AppSumo and decided to go ahead and sign up for it (AppSumo has a really, really good refund policy – I’ve used it.) And despite it looking like magic, it works! Combining it with my microphone setup doesn’t improve it significantly, but it does include a feature where I can background noise filter incoming audio too (when I’m on a call with someone who has a lot of background noise.)

So if you already have a good microphone, I’d really suggest checking out Krisp, especially while AppSumo has this lifetime deal going on. After that you can get Krisp for $20/mo (or $10/mo billed annually). You can get 2 months free trial with this affiliate link (since I got the lifetime deal from AppSumo I don’t think I get anything if you use the link. Here is a direct link if you prefer.)

(I’ve been using AppSumo for a while now. I’ve got a few of their lifetime deals, and only one deal. I’ve recommended some specific deals to others and they got it . They offer a Pro subscription that gives you a discount and other benefits, I’ve considered it. If you use the affiliate link in this post I get a credit if you buy. If you don’t like affiliate links, you can access it directly here.)

Krisp works by setting up a virtual audio device, and you can route your software through it (works with pretty much any any software) on both Windows and macOS. You can turn the “filter” on or off for incoming our outgoing audio.

I might create a demo recording, but haven’t yet. Their demos totally appear legit to me though. Since it shows up as an audio device you can use it for recording audio for Audacity or Camtasia.

I really recommend checking out Krisp via AppSumo if you want to reject background noise.

Virtual Audio Mixer

A piece of software I’ve recently started using is Voicemeeter by VB-Audio software. They offer a range of low level audio software as donationware (similar to Shareware, but you decide how much to donate if you like it.) Originally I was using their VB-Audio Cable driver which lets you create a virtual audio device you can pipe sound, as if it were attached to its own individual sound system. When you combine the two, it gives you the ability mix and control the sound from six different sound inputs and outputs.

VoiceMeeter Potato Ultimate Virtual Mixing Console
Voicemeeter Potato (aka Voicemeeter 3)

There is a bit of a learning curve, but if you really want to take full control of your sound system it is a my recommendation. You can also combine it with the Krisp system since it provides its own input and output virtual devices.

Audio Interface Protocols

I don’t know much about this, but wanted to point out that there are different audio protocols on computers. Depending on the hardware and software you have different options.

  • MME – Multimedia Events – Legacy system with higher latency
  • WASAPI – Windows Audio Session API – Lower latency and more recent interface
  • DirectSound – Direct X wrapper or abstraction – it is more universal, but has more latency than WASAPI, and more simplified interface.
  • ASIO – Extremely low latency, but supported less often then the above. Optionally use the free ASIO4ALL driver to add support.

I’m using WASAPI when given the choice. I’m looking at using ASIO, but haven’t committed to it yet since it requires installing more software and drivers.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.