How do I improve the Signal to Noise Ratio of low level signals?



I'm recording breathing sounds (inside anechoic chamber) with a MEMS microphone in a custom housing. The sound level is low; and near the noise floor of the mic, so we have a lot of input gain and the result is quite noisy. How can the noise be reduced?

Here is what I'm working on:

  1. Removed all ground loops
  2. Reducing the microphone noise floor with electronics. (Battery power, bypass capacitors)

  3. Low pass filtering at the microphone's output.

  4. Using a quiet mic. preamp (RDL RU-MLA2)

  5. Impedance matching with series and shunt loads

Noise gating is a nice idea but I understand this gates the quiet bits, where there is no signal anyway. If you suggest this please explain in detail how you think it could reduce the SNR of my signal.


Adam Ratcliff

Posted 2018-02-01T23:20:48.733

Reputation: 51

Just a silly question. Since you have an access to an anechoic chamber, doesn’t the laboratory have any proper microphones with low noise? – jojek – 2018-02-01T23:30:13.967

@jojek What "proper" microphones are you suggesting, and how low do you think is low enough? – Adam Ratcliff – 2018-02-02T00:51:11.410

1I’ve been using GRAS 40HF which goes as low as 0dBA. It was possible to record soap bubbles bursting in anechoic chamber. Again, I am simply asking if your lab has this sort of thing or it can get a grant for it. – jojek – 2018-02-02T00:55:44.267

@jojek yes a lower noise microphone is a good suggestion. Before taking that option I want to show my peers that I've done all I can to lower the noise floor. – Adam Ratcliff – 2018-02-02T01:06:52.357

What is the total SNR of your system, all items included? You need at least 32 bit (140dB) or 24 bit (120dB) worst case net dynamic range, so you can be sure the noise source is the microphone or a mic per-amp. – None – 2018-02-07T02:38:56.287

You can't improve on the inherent S/N of the microphone by adding anything after it, anything you add to amplify the signal will also amplify the noise. Using a better preamp decreases the noise added by the preamp it does not do anything for the noise of the mic. so if the mic is the problem then the only thing you can do is try to get more SPL into the mic (more signal). Or a different mic – little_birdie – 2018-02-09T16:55:46.777

@little_birdie Sorry for the delay in responding. This was a long time ago but I want to rebut your assertion about the inherent S/N of the microphone: Filtering can improve the SNR, where the noise is out of band. – Adam Ratcliff – 2020-07-03T05:03:27.873

That is a fair point but you said you were already filtering. I was referring to noise at the frequencies of interest. – little_birdie – 2020-07-04T16:02:38.910

How did it all work out in the end? – little_birdie – 2020-07-04T16:05:14.777

OK I understand. It was ultimately an effective design. Thanks for your input! – Adam Ratcliff – 2020-07-06T05:28:06.930



The point of an MEMS microphone is to deliver good directionality and SNR at minimal size and weight, like on a mobile phone. When recording in an anechoic chamber, that seems like a silly optimization target.

A good small membrane condenser microphone will likely already work better here, and even the inherent colorization of a large membrane condenser will likely not make much of a difference for breathing sounds.


Posted 2018-02-01T23:20:48.733


@user24052 the MEMS microphone is being used to pick up vibrations that come from a Gel-like substrate into the air via a custom interface medium. Directionality and SNR at minimal size is what we need, but that's all off topic. – Adam Ratcliff – 2018-02-07T02:08:18.863


Based on what I know about MEMS-microphones, I'd say your one option - not counting getting the best preamp you can get which is always a good idea - is to get a proper microphone for this. I record many far quieter stuff than that, and can always rely upon especially my Sennheiser MKH-mics for things like that.

Christian van Caine

Posted 2018-02-01T23:20:48.733

Reputation: 3 112

1Thanks, changing to a microphone with a lower noise floor is a good suggestion. Unfortunately specs for that sort of thing don't seem to be published.

My noise floor is about -60dB below full scale.

@Christian Can you measure the noise floor with yours?

I would really like to do everything I can to lower the noise floor first before changing microphones so I'm open to other answers. – Adam Ratcliff – 2018-02-02T00:52:05.897