TechTalk: Measurement microphones
With a DAC+ DSP or the Beocreate 4CA, it is relatively easy to implement room acoustics correction filters (and it will be easier in the future, but we’ll come to this when it’s the time). However, there is one thing you need: a measurement microphone. You might have some microphones laying around somewhere and you might think about using these. However, “normal” microphones designed to record audio are usually not designed for measurements. A measurement microphone is designed with only one goal: linear frequency response.
There are lots of different measurement on the market. The more expensive ones usually come with individual correction files that allow to trim these to perfect linearity, while the more affordable ones do not have these individual correction files. No two microphones are perfectly equal due to variances in manufacturing, therefore it makes sense to have the microphone corrected individually – correct? Let’s have a look.
For this test, we use 4 different measurement microphones, all of which connect to the USB port. Why USB? Measurement microphones are usually so-called condenser microphones that require not only a separate pre-amplifier, but also a power supply. That requires quite a lot of stuff for a simple measurement. Luckily, in the last years, USB measurement microphones became more popular. They basically combine a power supply, pre-amplifier and USB-ADC into one device. This is a lot smaller and more affordable than “classic” measurement microphones. In our test, 3 of the 4 microphones are USB microphones, while the ECM8000 uses an external pre-amplifier and USB sound card. We tested these three because all of these are very affordable for the DIY market. These microphones cost between €60 and €150. The Behringer ECM8000 is even below this price range, but it requires an additional pre-amplifier which will add costs again. All microphones are relatively new, except for the ECM8000 that was our workhorse here for the last 15+ years.
On the right side you also see a dynamic microphone that’s designed for sound recordings. We’ll talk about that later.
All these are on the lower end of the price range of measurement microphones. If you wish, you can spend a lot more.
The setup is quite simple. The microphone is positioned about 2 meters in front of a single speaker. We won’t measure the frequency response of the speaker, but of the speaker within an existing listening environment. For this type of measurements it is important that the microphone is positioned vertically as you see it on the following picture.
We did not use any correction files, even if they were available. You might think that this isn’t a good idea, but have a look at the results first.
The software used here is our own application. It uses sine sweeps to measure the frequency response of not just the speaker, but the speaker within the room. Therefore, you will not see a real flat frequency response – this just isn’t realistic in a normal listening room at home. We’re working on an integration of the software into HiFiBerryOS, but there is still a lot of work to do for this.
Ok, have a look at these measurements:
Let’s begin with some assumptions that have nothing to do with the microphones itself:
- The speaker used here is far from perfect. You will notice that the volume below 1kHz is quite high compared to the volume at frequencies > 1kHz. This is really caused by the characteristics of this speaker. We won’t tell what speaker it is, but we can say it is an affordable speaker from a well known manufacturer.
- The room is far from perfect. The extreme peaks at about 105Hz and 150Hz are effects of the room, not the speaker. We’ll look into room acoustic optimizations in a future post.
Having said all that, what else do you see? The different measurements from the different microphones match almost perfectly. Yes, there are differences of 1-2dB in some frequency ranges. However, this is a level of precision that just isn’t required for normal room acoustics optimisation.
The dynamic microphone
Now have a look at this:
You see that these frequency responses do not match at all. Was it just a bad microphone? No. A microphone designed to record voice must have completely different characteristics than a measurement microphone. You don’t want this to record frequencies below 200Hz – and that’s one thing you see here. It is also not designed to be omni-directional, but optimised to record sound coming from the front of the microphone. It also doesn’t need to be perfectly linear. A microphone like this should not record very low frequencies as they would only create problems for this use case. There are other factors that are important for this use case, but you won’t see these in these plots. We only added this graph to show you that you really need a measurement microphone if you want to optimise room acoustics.
- Even affordable measurement microphones perform reasonably well to measure frequency responses. While there are differences in frequency response, they are quite small and you might not even need individual correction files.
- You should not try to use “normal” microphones for this kind of measurement – they are just not designed for this purpose.
November 13, 2019