| Author |
Topic: Measuring Noise Floor |
Chris Bauer
From:
Nashville, TN USA
|
Posted 31 May 2020 9:29 am |
|
Okay, I know this is a pre-kindergarten level question but here goes...
If I want to measure the noise floor of a room...
If, say, I want the maximum noise floor to be -60db, does that simply mean that no ambient noise peaks above 60db or, as I suspect, does it mean something else entirely?
If, indeed, it means something else, what’s the best way to measure it? |
|
|
 |
Georg Sørtun
From:
Mandal, Agder, Norway
|
Posted 31 May 2020 4:04 pm
|
|
As they say; this is complex…
(well above pre-kindergarden level.)
Reason is that any "db" value is, and has to be, relative to something that is pre-defined, or universally understood.
A quick look through this description: https://en.wikipedia.org/wiki/Decibel will give you an idea about how many ways "db" can be used.
To take your values first…
-60db is (in most cases) read as: 1/1000 of a known level. But since no reference level is given, in this case -60db is 1/1000 of an unknown level.
Now, in a living-room the acoustic noise-floor can be measured with a regular but quite sensitive db-meter with a mic, and can be expected to be in the range of 30db to 45db above the lowest sound any human can be universally understood to hear.
In your case that noise-floor can also be expressed as being -60db to -45db lower than (for instance) music played at around 90db above the lowest sound any human can hear. |
|
|
|
Dave Mudgett
From:
Central Pennsylvania and Gallatin, Tennessee
|
Posted 31 May 2020 4:50 pm
|
|
When I think of noise floor, I'm thinking of how many decibels (dB) of RMS (Root Mean Squared) noise signal level lower than the RMS level of whatever is the 0 dB level of whatever signal it is being compared to.
So yes, it's a bit arbitrary, but there are standards for what to consider the 0 dB level of, let's say, music being recorded.
I would not think of ambient noise "peaks", but RMS signal, which is obtained by squaring a signal f(t), integrating it over a sample time interval, dividing by the time interval, and then taking the square root - i.e., for the mathematically inclined:
The RMS calculation is a type of averaging over a time period. But in this case the signal is first squared (to make all signal points non-negative and accentuate large signal values over small ones), and then averaged over the time period. |
|
|
|
Mike Auman
From:
North Texas, USA
|
Posted 31 May 2020 6:57 pm
|
|
Chris, to reply to the second part of your question, you can measure the ambient sound pressure level (noise floor) with a smart phone app. I use "deciBel" and "Audizr" on my Android phone (both have free and paid versions) and there are similar ones for iPhones. The results of these are usually good enough, but if you want a higher level of accuracy, you can calibrate the Android apps against a known level or compare them side by side with a calibrated SPL meter. Most iPhone apps are already pretty accurate without calibration. It has to do with the inconsistencies in Android hardware from brand to brand, vs. the sameness of Apple hardware. Best regards, Mike
_________________
Long-time guitar player, now wrestling with lap steel. |
|
|
|
Dave Mudgett
From:
Central Pennsylvania and Gallatin, Tennessee
|
Posted 1 Jun 2020 7:43 am
|
|
There are a myriad of details on actually computing practical RMS sound pressure levels - wikipedia actually has a pretty good writeup on that: https://en.wikipedia.org/wiki/Sound_level_meter
Dedicated sound pressure meters can be purchased pretty reasonably. But I think most low-end stuff is pretty limited, mostly oriented to safety aspects of sound levels, not low levels like a noise floor. And if you look at the dominant "A" frequency weighting, it heavily de-emphasizes low and low-midrange frequencies. I'd want a meter which gave other options, including unweighted (Z-weighting) as well as both "fast" and "slow" time-weighting.
Just some examples, not pushing amazon - https://www.amazon.com/s?k=Sound+Level+Meter |
|
|
|
Georg Sørtun
From:
Mandal, Agder, Norway
|
Posted 1 Jun 2020 11:36 am
|
|
I'm with Dave on this, and how good/sensitive a meter needs to be depends on the job. For "quiet living-room" a reasonably good instrument that goes down to 30db will be plenty low enough – regular rooms are always more noise than one thinks.
For "rooms" designed to be quiet, 30db will barely do, and instruments for qualified – not to mention "unweighted" – measurements down in that range tend to be pricey.
So, what are those measurements for? |
|
|
|
Chris Bauer
From:
Nashville, TN USA
|
Posted 1 Jun 2020 12:37 pm
|
|
| Georg - I'm recording for a variety of sites who have varying requirements as far as the noise floor. I'm working in a currently poorly treated room and want to be sure I'm meeting or, hopefully, exceeding the expected standards once the room is appropriately treated. |
|
|
|
Georg Sørtun
From:
Mandal, Agder, Norway
|
Posted 1 Jun 2020 1:56 pm
|
|
Got it.
Start by establishing the reference-level – full volume during recordings – which with amplified instruments and singing voices can be expected to average out around 100db (VU or approx RMS).
A quite acceptable noise-floor will then be -60db (40db measured under quiet conditions with any reasonably good db-meter).
FWIW; I've rigged a few studios – mainly for radio but also for recording, and those were in that range and more than acceptable. What I focused more on than the noise-floor as such, was what noise-frequencies that were dominating. Always most difficult to keep sub-freqs out of a room/building, but depending on what you're recording and type of sub-frequencies, those lows may also be easiest to filter out during recording. |
|
|
|
