In this article, I’d like to talk a little about noise floor, specifically, what it is, and how to measure and augment it.
Noise floor, or sometimes called room tone, is the cumulative sum of sounds in a room when no speaking is taking place. These sounds come from a variety of sources. Things like refrigerators, computer fans, open windows, HVAC systems and of course environmental stuff like birds, wind and traffic. Sound also emanates from things using electricity. Things like light bulbs, power sockets and even TV’s and monitors. All of these sources gel together to create an ambient sound in your studio we call noise floor and for all professional audio creation purposes, a lower number is always better.
We can measure noise floor in several ways. Let’s cover the digital, or dBFS scale first as it’s what we use to speak about volume in terms of recorded audio. dBFS, that’s Decibels, Full Scale, is a digital scale measured in negative numbers starting at the loudest possible sound that can be captured (sampled) by it’s scale – that’s 0 dBFS.
Then, as sound drops in volume, or rather, signal presence, we also drop in the number scale. So again, 0 dBFS is full scale (loud) and as sound attenuates (lowers, gets quieter) the numbers drop (-1, -2, -10 etc…). It’s a bit confusing at first and to further complicate matters, while 0 dBFS can be loud, it can also be quiet.
The dBFS scale measures signal input or output and is often measured at the converter stage, that is, the digital audio converter – your soundcard. You could, say, send a lot of input level from your mic through your preamp to your soundcard and this level could measure an average of -15 dBFS. this in fact is a good level to target for most commercial voice over recordings. However, you could then turn the volume output knob down and listen to this fairly loud signal at a reasonable level in your headphones or on your speakers, something comfortable to your ears. Keep in mind that the level getting converted isn’t necessarily the level you hear on playback – what is good however, is the standard which dBFS gives us. From studio to studio, should the monitoring (volume level) be calibrated, we can speak of program (audio) in relatable terms.
SPL, or sound pressure level, is another scale used to measure sound. This scale is often used to measure ambient noise – like your studio. Sound is in fact pressure variations in air and is a mechanical type of energy. We can measure this energy with a sound pressure level meter, or SPL meter.
To confuse you a little further, we measure sound in this realm in a similar dB scale, though we flip the scale and call it dB SPL. 0 dB in this scale is the threshold of hearing for humans (the quietest possible sound). These numbers increase in a positive fashion until around 130 dB SPL, which is the threshold of pain – ouch!.
If I was to speak to you with an average volume indoors, my voice might measure around 65 dB SPL. OSHA, the Occupational Safety and Health Organization actually measures and sets levels for work environments around the USA. In it’s documentation, the level of 85 dB SPL has been set for the upper limit of sound pressure for a human 8 hour work day. As pressure levels increase the amount of time one can be exposed to such levels, goes down.
Now, back to noise floor.
Before we discuss the finer points of your studio, we first need to grab a tool to measure with. I’m hoping you have an iPhone or Android handset. If you do, head to your app store and look for an SPL METER by a company called Studio Six DIgital – it should be free. Download the app. Below is a picture of the meter.
Let’s discuss this meter. First, a large dial to the right makes up most of the interface. This dial sets the range of the meter readings above. You’d need to set the dial for a reasonable measurement you think you’ll be taking. For this first measurement, let’s take a reading of your voice – in a standard room. Let’s set the dial to 70. On the left of the meter we have two scales, A and C and below, the response time, FAST and SLOW. When measuring more chaotic material, like voice or music or louder sources, C weighting in preferred, in a SLOW response time. So, having set the dial to 70, let’s set the weighting to C and response time to SLOW. Now, hold your phone out just a bit and read the measurements.There’s a max level on the left and a dB level on the right. Let’s focus on the right level. Holding the phone out a bit, read the level. It’s likely around 65 when talking.
For fun, try measuring other stuff. For an idea on times of exposure and SPL levels, I’ve included a chart below.
Notice the music concert levels and the amount of time you’re allowed to listen without hearing protection. Yikes!
So, back to the practical matter at hand. Let’s measure your studio. When measuring rooms, like studios and businesses and work places, quiet places, we measure with an A weighted scale and a FAST response time.
From Bruel & Kjaer, sound measurement experts…
Measuring dB(A) The most common measurement in environmental noise is the dB(A) level. It can be measured with a simple Sound Level Meter having an A-weighting filter to simulate the subjective response of the human ear. The dB(A) level is used to report ambient noise and noise intrusions, it is also used in computing LAeq,T and LAN,T.
Here’s the original article. [Environmental Noise]
Set your main dial on the meter to about 40 and set the scale to A and response to FAST. Now put the device down on your music stand or chair and close up the studio just like you’d do for a voice over session. Now, at some distance, monitor the level readout and change the dial according to the reading your getting. This is how we measure you studio. In decent studios you’re likely to come up with dB SPL readings in the 20 to 30 range. My professional studio comes up at 25. In fairly good studios you might come up with 30-40 dB SPL.
Ok, so now having discussed dB SPL, let’s talk about another effective way to measure noise floor. This method coincides with Audible and ACX standards and requires an RMS meter.
Here’s a link to a great video explainer from ACX on the subject – All About Noise Floor
First, set your recording level, the signal to computer via your preamp or soundcard’s input dial, to a comfortable recording level. If you’re not fairly dynamic and are recording commercial type VO copy, a level average around -12 should be ok – but watch for overs. For audiobooks, a level around -25 is more inline with the dynamic nature of this medium. Even when targeting these above mentioned levels, it’s likely that a measured final output would yield a lower value. Often when I target -12 for average VO, I end up measuring in the mid 20’s. This is ok, setting levels is not an exact science.
Having set your level, go ahead and record some voiceover, maybe a 30 second spot or so and when finished speaking, continue to record but be as absolute quiet as possible, in fact, if it’s possible to step out of the room for a bit, do so. Record about a minute of room tone at this preamp level.
Now, back at your computer, bring up a RMS meter. If you have Pro Tools, head to the Audiosuite menu and down to Other and finally to the GAIN plugin.
Using this plugin you can highlight a larger section of the quiet recorded room tone and measure using the RMS option by hitting analyze. The result level indicated is your noise floor. Make sure when measuring that you’re only measuring the very quiet section with no extra noises, this includes the door open and close if you left your studio and your voice. Also, out of that minute of recorded room tone, maybe use the best 45 seconds or so. By setting this plugin to RMS, you hit analyze and the unit measures using this standard. ACX requests a noise floor of -60 dbFS after mastering. Although this can be difficult to achieve, as mastering raises the overall level, at least it’s a target. If you measured around -60 with this meter using the RMS button, then I’d say you studio is fairly quiet and good for recording standard VO. Anything higher than this is ok, but getting your room to this level will suit you well. Obvious things to consider are shutting all windows, doors and turning off machines in the vicinity. You can also begin to soften the room with pillows and blankets and if possible, wrap the area in sheets. The softer and more hefty an object, the more sound absorption it can handle.
Another way to verify a decent noise floor is to measure the difference between signal (voice) and floor. If you take that recording we just finished, using the Pro Tools plugin mentioned above, measure and take note of the speaking only section (highlight the VO and measure that). Once you’ve jotted down that number, take a measurement of the noise floor (if you haven’t already). A good signal to noise level would be a difference of 40-50 dB, excellent would be 50-60 dB.
There’s also a way to measure with RMS using Twisted Wave on a mac. Like mentioned above, please set your levels to a tasteful spot, making sure not to peak over 0 dBfs. A good level is achieved by sending a hot signal to the soundcard, that is, a loud signal that shows well on the meter, but leaves some headroom in case of overs.
Once you’ve set your level, record 30 seconds of some copy and when finished, another minute of room tone (noise floor). If you can step out of the room when doing this, please do so.
Now, back in Twisted Wave, highlight the voice section of your audio and head up to FILE and the ANALYZE option.
Note the AVERAGE RMS level in this readout – jot this down.
Now, highlight a clean section of your room tone, maybe 3/4 of the actual recorded material and analyze that. Your RMS average for the quiet room tone is your average noise floor level and the difference between that and the RMS level of your voice is the signal to noise ratio – again, 40-50 dB here is a good value, 50-60 dB is great, less is ok, but striving for a larger number here is best.
WRAPPING IT UP
To summarize, noise floor is what some call room tone. It’s a variable beast and it changes with the input level you send to your preamp. Noise floor is the culmination of all sounds sans voice in a studio. Although there are ways to measure using a static method, like the SPL meter, it’s often best to use the industry standards set out by companies like Audible. To accurately measure a noise floor an acoustician would often reach for a balanced omni-directional mic and a fancy RTA meter, but those tools can be expensive. For the average person, checking the RMS levels of pre-recorded material is best.
Although the input level to your system will change, measuring the signal to noise level is a great approximation of the quality of your studio. Also, just measuring the RMS level of you noise floor during a standard voiceover session, will give you a number which you can feel comfortable with.
A low noise floor is is the mark of a good studio. The lower you can get the number, the more dynamic you can record and during stages of mastering or compression, you can be assured that raising levels won’t unnecessarily bring up louder levels of tone. The quieter the base, the easier it is to augment the audio. Professional studios operate in the -60 to -80 range of dBFS. A signal to noise level of 40-60 dB is also considered professional and if measuring the ambient dB SPL level with a meter, A weighting FAST, an ambient SPL of around 20 would be considered great, 30 is ok.
For further reading, here is a great document on studio acoustics from Aurelex.