Microphone/Audio Recorder Tests

A recording to test the noise performance of the Olympus LS-P4 versus the Zoom F4. Two sets of a stereo pair of FEL Clippy microphones (3.5mm into the Olympus, XLR into the Zoom) were mounted on a 20cm stereo bar. Gain was set at the maximum of 30 on the Olympus and +72db (out of a max of +75db) on the Zoom, recording on both at 24-bit/48khz. Both recorders were hovering around -6db on the meter with each tick of the clock. The Zoom is first, followed by the Olympus. No processing was done other than normalisation of the two files to 0db in Reaper (although the level of both recordings was near identical already).

To my ears the amount of noise is comparable between the two recorders, but the character of the Olympus noise is slightly broader spectrum. The gain is really cranked up on both recorders so this is really, really impressive for the Olympus considering you'd normally have the gain much lower with sensitive mics like the FEL Clippys. Also impressive is that the Olympus seems to be capable of adding almost as much gain as the Zoom F4.

Olympus LS-P4 Noise Test High Gain

Another recording to test the noise performance of the Olympus LS-P4 versus the Zoom F4. This time at a more 'normal' gain setting. Two sets of a stereo pair of FEL Clippy microphones (3.5mm into the Olympus, XLR into the Zoom) were mounted on a 20cm stereo bar. Gain was set at 15 out of a maximum of 30 on the Olympus (a level which often seems about right for typical ambience recordings) and +38db (out of a max of +75db) on the Zoom, recording on both at 24-bit/48khz. Both recorders were hovering around -12db on the meter with each tick of the clock. The Olympus is first, followed by the Zoom. No processing was done other than increasing the gain of the Zoom recording by 0.5db to match the Olympus level.

I cannot hear any difference between the two recorders in either the amount of, or the character of the noise. The Avisoft website says the Olympus LS-P1 (which should have the same preamps as the LS-P4) EIN is only -114dBu compared to -127dBu for the Zoom F4. This sounds bad on paper, but if you're using Primo EM172-based microphones there would appear to be no audible difference, at least to my ears.

Olympus LS-P4 Noise Test Medium Gain

The Zoom F6 occasionally displays the error 'exceeding maximum input levels' when recording high SPL sound sources in 32-bit mode. Other users have reported the same problem but on analysis afterwards, the sound isn't clipped or distorted in any way. On this test I'm using a pair of Clippy (EM172) microphones right on top of a ride cymbal. The first couple of hits are quiet but then for all the subsequent ones the error came up on screen so I was expecting either clipped or distorted sound afterwards. However, I was able to normalise the audio to get rid of the clipping, and there is no audible distortion. This looks like maybe a software problem, where the unit is displaying the error at the wrong time.

Zoom F6 Exceeding Maximum Input Level

2 x AT4022, 1 x BP4025 (stereo microphone) and 2 x Clippy recorded at unity gain on the Zoom F6 and normalised in reaper (+22db for AT4022, +24db for BP4025, +16db for FEL Clippy). 

AT4022 sound the quietest to me, with BP4025 very slightly more noisy, and the Clippy microphones quite significantly more noisy,

Audio Technica AT4022, Audio Technica BP4025, FEL Clippy (EM172) Noise Test

Another test of the BP4025 vs Primo EM172 (FEL Clippy XLR). This time in a more real-world setting. 

I have noticed how bright the EM172s are in comparison to the BP4025 (and AT4022 which I tested earlier). This led me to think they can stand a little EQ adjustment to try to reduce the noise, which is naturally higher than the BP4025.

1. Primo EM172 stereo pair
2. BP4025
3. Primo EM172 with EQ applied (-4db reduction in the 5-15khz range)

1 and 2 were recorded simultaneously to a Zoom F6 and later normalised. 3 is a copy of 1 with EQ applied.

Conclusion 1: BP4025 is naturally more quiet than the EM172 but has a more narrow stereo image. However, applying some EQ to the EM172 brings the noise level even lower, and without any negative impact to the overall sound. To my ears, it sounds even more natural now. 

Conclusion 2: I didn't need to buy the BP4025 to achieve lower noise recordings. Damn!

EM172 VS BP4025 VS EM172 With EQ

One clip is Sennheiser MHK 8020, the other clip is Primo EM172 with EQ applied. Can you tell the difference?

As of August 2020, a pair of Sennheiser MKH 8020 costs more than £2,000 in the UK. A pair of Primo EM172 microphones can be purchased ready-made for less than £100, or built for even less. 

Original recording by Vicki Powys used with permission. You can read more about the setup of the original recording here: https://caperteebirder.com/?page_id=333

Sennheiser MKH 8020 VS Primo EM172 Blind Test

Quick test of two Primo EM172 stereo setups. One is AB 40cm. The other involves two ends of a foam roller to make two boundaries, with the distance between the microphones approximately head-spaced. 

Can you tell which is which? Is the stereo imaging better with one than the other?

Boundary Array Test

2 x Clippy microphones in a cupboard recording to an Olympus LS-P4. Same source recorded at 7 different gain settings (1, 5, 10, 15, 20, 25, 30) and normalised in post (adding 70db to the lowest gain setting, 5db to the highest). Recorded at 24/48.

This shows a very clear linear progression in the reduction of noise. I often record at level 15 in quiet ambiences but this shows there's something to be gained (in terms of HF noise reduction) by cranking it up to 20 so long as there's no likely to be any loud sound sources that will clip the recorder.

Olympus LS - P4 Self - Noise Test

Test of the baffled omni array I have been working on with two Primo EM172 capsules (FEL Clippy). Microphones were pointing out from a first floor window up towards the sky (at some birds clattering about in the gutters).

Baffled Omni Array Test

This test is comparing the self-noise of the Sony PCM-A10 and the Olympus LS-P4 at various gain settings using external microphones (Clippy EM172). Both recorders allow gain settings from 1-30. I recorded the same sound source (quiet ticking clock) at levels 1, 5, 10, 15, 20, 25 and then 30 simultaneously on both recorders and then normalised the levels in Reaper. I then had to do quite a bit of manual correction of the levels because there were some low frequency sounds that were louder than the ticking clock and which were messing with the normalisation process. Recordings were made at 24/48 WAV.

In this recording you first hear the Sony at gain level 1, followed by the Olympus at gain level 1 and so on up to level 30.

As shown in an earlier test the Olympus exhibits much higher levels of self noise at lower gain settings, with a noticeable improvement coming at level 20. This means that when recording quiet scenes, it is worth taking the risk of setting the gain high to minimise the recorder self noise.

The Sony behaves quite differently. At gain level 1 the noise is quite terrible  but at level 5 there is a sudden improvement and in fact there is almost no difference between the noise at level 5 and level 30. This seems to suggest that you can be much more conservative with setting the gain on the Sony to avoid clipping, and increase the levels later in post with no apparent self-noise penalty.

Sony PCM - A10 VS Olympus LS - P4

Another test of the Sony PCM-A10 at different gain settings using external microphones (Clippy EM172 stereo pair). Recordings of two different subjects at low and high gain to compare the recorder self-noise. Tracks normalised in post afterwards. Recorded at 24/48.

Clip 1: Indoors with windows shut. Birds audible through the windows. Recorded at input level 5.
Clip 2: Same as above recorded at input level 25
Clip 3: Outdoor ambience recorded at input level 5.
Clip 4: Same as above at input level 20 (highest I could go without the sparrows clipping the recording!) 

A gain setting of 5 on the PCM-A10 is very, very low in case it isn't clear. There was virtually nothing registering on the meters while I was recording.

I hear zero difference in noise for the recordings at very low and very high gain settings. This is incredible as it means you can set your levels very conservatively to avoid clipping and increase levels in post later with no fear of losing the signal in the noise.

Sony PCM - A10 Gain Test

More tests of the Sony PCM-A10 at extremely low gain levels. I recorded a few seconds of test tone followed by 20+ seconds of a quiet room at gain levels 1, 2, 3, 4 and 5, and then level 25 as a high gain level for comparison (highest gain should have the best noise performance).  The recordings were then normalised in Reaper, resulting in gain being added to each clip of: +40.8dB, +39.1dB, +37.7dB, +35.7dB, +34.6dB and finally +2.87dB (gain level 25).   

Hopefully what you can hear in this is that levels 1 and 2 have a high level of noise, but then it suddenly drops off at level 3, and levels 3, 4, 5 and 25 all have nearly identical levels of audible noise.

My conclusion is that when recording a dynamic source such as a thunderstorm, I can safely go all the way down to gain level 3 to minimise the chance of clipping, while not adding significant noise when I later raise the levels in post.

Sony PCM - A10 Gain Test (Extremely Low Levels)

Quick test of a quasi-binaural* setup in the local park, intended for use in public spaces where I don't want it to be obvious that I'm recording. Clippy EM172 microphones (with foam covers) wedged inside a 'Windfree' (https://www.windfree.se/en/home). The wind protection isn't great, but it holds the microphones in place against the sides of my head, and makes it look a bit more like I'm just wearing some sort of headphones and listening to music. I might try using Bubblebee Windbubbles if they'll fit inside the Windfree.

The road is to the rear, and kids at the sides and in front. 

The Sony PCM-A10 has a nice Android app which I can use to start and stop recordings, monitor (visual only) and adjust levels so (hopefully) it really isn't obvious that I'm making recordings.

*Because the capsules are against the sides of my head, rather than right in my ears, this isn't true binaural.

Hinksey Park Binaural Test

In this test I compare the self-noise of Clippy EM172 into a Sony PCM-A10 recording at extremely low gain (level 3 out of 30) versus the same microphones into a Zoom F6 recording in 32-bit float mode.

Clip 1: Sony recording of outdoor scene at original volume
Clip 2: Same Sony recording normalised (+24.2dB)
Clip 3: Zoom F6 recording of the same scene normalised to the same level
Clip 4: Sony recording of indoor scene normalised (+24.2dB)
Clip 5: Zoom F6 recording of the same scene normalised to the same level

My conclusion is that when using Clippy EM172 microphones (14dBA self noise, -28dB sensitivity), low noise preamps such as those on Zoom F series recorders or Sound Devices are overkill. In addition, if you want to avoid clipping (a reason for using 32-bit float) you can simply set the gain level on the PCM-A10 very low, and when you bring the noise levels back up there is no noise penalty (you should avoid going below level 3 however as I have demonstrated in a previous test that the noise increases considerably below level 3).

Sony PCM-A10 Low Gain Vs Zoom F6 32-Bit Float

Quick self-noise comparison of Primo EM258 (aka Lom mikroUsi) vs Primo EM172/272 (aka Clippy, aka Lom Usi).

EM258 is rated at 20dBA and EM172/272 is rated at 14dBA. The difference is noticeable indoors, but not so much outdoors in the situations I have tried so far (not super quiet).

EM258 EM172/EM272 Self-Noise Comparison

First clip is RX8 Spectral De-Noise just using the hiss control preset at 4db of reduction. Second clip is using EQ to reduce the hiss (5db band reduction around 8khz).

Noise Reduction Test

Recording of a quiet suburban scene. EM172 followed by EM258. Simultaneous recording to 2 x Sony PCM-A10 recorders.

EM172 Vs EM258 Self-Noise Ambience

Another test of a spaced omni array (54cm spacing) versus a quasi-binaural (baffled, roughly head-spaced) array, both using Clippy EM172 microphones.

Spaced clip is first followed by the quasi-binaural. High frequencies on the quasi-binaural setup are being attenuated a bit by the big windjammer.

Recorded simultaneously to 2 x Sony PCM-A10 recorders at the same gain setting.

AB54cm Vs Quasi-Binaural

Two simultaneous recordings made a while back when I was temporarily in possession of a pair of Audio Technica AT4022 microphones and a Zoom F6 recorder. The AT4022s have a similar performance in terms of self-noise to the Sennheiser MKH 8020s but come in at about a third of the price.

I record everything with Primo EM172-based microphones at the moment but keep getting tempted by lower self-noise (but more expensive, larger and heavier) microphones. However, a test like this makes me question how worthwhile that would be...

One clip is the AT4022s, the other is the EM172s with some noise reduction applied (Izotope RX8 Spectral De-Noise). The difference is very subtle and I doubt it would be noticeable at all outside of an AB comparison like this. Also, in 80-90% of recording situations I find myself either the subject or the environmental noise is much louder than in this test (the gain is really cranked here), and so you wouldn't be hearing any self-noise from the Primo capsules anyway.  Combine that with the fact that the Primos are tiny, weigh almost nothing and are cheap to replace, I can see why so many people love them!   

Let me know in the comments which you think is which.

Audio Technica AT4022 Vs Primo EM172 Blind Test

I wanted to test whether the Primo EM258s (same capsules used for the Lom mikroUsi) have low enough self-noise for recording in a quiet, natural environment. I recorded the following clips in the middle of winter in a relatively quiet local wood. I set the gain so that when played back at maximum volume on a computer or phone, the clips should be approximately the same volume as the sound was in reality. That meant setting the gain lower for the EM172s, which are a bit more sensitive than the EM258s.

The first section presents first the EM258s and then the EM172s without any gain added. This was a very quiet environment, so even at this very low playback volume, the noise of the EM258s (clip 1) is pretty noticeable, whereas the the EM172s sound pristine.

The next section presents the same two clips again with 12dB of gain added, which is probably a bit of a more 'typical' playback volume. The EM258s are very noisy now, with the EM172s still sounding great. There is a small amount of noise, but really very little.

The third section presents the same two clips again still with the 12dB of gain, but also with some noise reduction (Izotope RX8 spectral de-noise). Despite the noise reduction, the EM258 self-noise is still pretty noticeable. There has also been quite a loss of high frequency content meaning the birds are harder to hear now. The EM172s however fare much better. The self-noise is completely inaudible, and the high frequencies although attenuated, are still sufficiently present. In fact, you could say it makes them a bit more flat and natural sounding, in line with higher-end mics like the Audio Technica AT4022s, or Sennheiser MKH 8020s. 

Probably a bit of an obvious outcome based on the specs of the microphones, but it has confirmed for me that I should always have my EM172s with me for recording in quieter environments. They are more sensitive and have lower self-noise than the EM258s. In addition, they naturally have a boost in the high frequencies compared to other microphones, meaning you can apply significant noise reduction but without drastic loss of high frequency content. 

Apologies for my swallow, and the odd little crackling sounds throughout. I was wearing the microphones on the straps of my backpack and picking up some human self-noise!

EM258 And EM172 Self - Noise Test In A Forest

Stereo soundscape recordings. Sometime nature, sometimes people, sometimes machinery. Sometimes quiet things like a single branch creaking in the wind on Dartmoor, sometimes odd moments in the city. Mostly local, but sometimes travelling. 

Microphone/Audio Recorder Tests

Douglas Langley