Signal to Noise Ratio and Mic Self Noise - How does it work?

[normington]

I'm struggling to get my head around the two different figures that microphone specifications seem to give.

The first is Signal to Noise Ratio, given in dB. Is a high rating good here?

The other is the self noise, given in dBA - what does dBA mean? And it seems that the reverse is true here, is a low rating good?

Why do some specifications give a Signal to Noise Ratio, while others give a self noise rating? This makes it difficult to compare one mic to another, because I don't know how these two ratings link together?

Thanks,
Andy

[tarnationsauce2]

Originally Posted by normington

I'm struggling to get my head around the two different figures that microphone specifications seem to give.

The first is Signal to Noise Ratio, given in dB. Is a high rating good here?

Yes

Originally Posted by normington

The other is the self noise, given in dBA - what does dBA mean? And it seems that the reverse is true here, is a low rating good?

Yes
dBA is a noise measurement that takes into account the way the human ear perceives the audio frequency range. For example if a mic has 20dBA, all on it's own it will create a noise that is 20dB over the theoretical lower threshold of human hearing. Basically dBA is a useful way to measure noise because it discredits frequencies we don't hear as well (or at all). But using a different weighting it might be have more or less dB of noise.

Think of the self noise as "hiss". The less hiss the better.

Originally Posted by normington

Why do some specifications give a Signal to Noise Ratio, while others give a self noise rating? This makes it difficult to compare one mic to another, because I don't know how these two ratings link together?

I'm not sure why one manufacturer would give certain specifications and not others. It's lame. To confuse you even more there are dynamic range and max SPL too.

Let me try and break it all down for you:

• Max SPL = The point where the mic distorts, or clips the waveform. [More=better]
• Self noise = The amount of noise the mic creates all on it's own. (i.e. hiss). [Less=better]
• Dynamic range = The range between self noise and Max SPL. [More=better]
• Signal to Noise ratio = The range between self noise and a reference signal. [More=better]

One caveat is if you compare specifications of mic A to Mic B, be sure they use the same weighting, references, or THD.
For example Mic A has Max SPL = 125dB @ 0.5% THD versus Mic B that has Max SPL 127dB @ 1.0% THD.
They use different "THD" reference points. In actuality mic A could have a better Max SPL than mic B @ 1.0% THD.

[normington]

Thanks, that's helpful.

Is there a way to convert a signal to noise ratio into a self noise rating, or is this impossible because of the different nature of the two readings? If not, is there at least a way you can estimate the value of one from the other?

What would you say is a 'average' and 'good' figure for each of the ratings to be?

[mshilarious]

Originally Posted by normington

Thanks, that's helpful.

Is there a way to convert a signal to noise ratio into a self noise rating, or is this impossible because of the different nature of the two readings? If not, is there at least a way you can estimate the value of one from the other?

What would you say is a 'average' and 'good' figure for each of the ratings to be?

SNR = 94 - self-noise. Both measurements are usually A-weighted, but not always, and some manufacturers state a few different noise weighting measurements.

Average self-noise ratings depend on capsule size, which is the primary factor. 1/2" mics are generally 12dBA - 16dBA self-noise; 1" mics, 6dBA - 10dBA. Very small diaphragm mics (1/4" / 6mm or less) will be more like 20dBA - 24dBA. Tube mics may have higher ratings; very old designs tend to be higher as well.

Any rating that is too much higher than those is an indication of a suspect design--you will see a lot of cheap mics advertised as LDCs with self-noise ratings like 17dBA - 20dBA. There should be no reason for that, if those noise ratings (and capsule size) are accurate.

It's also possible to lower a noise rating by increasing polarization voltage, to some extent (you need to double voltage for a 6dB decrease). That also changes the capsule's response; in fact if you crank the voltage too high, you can suck the diaphragm right onto the backplate! But if you see a manufacturer with super-low ratings on their mics (there are a couple of suspects I am thinking of), that's probably why.

[antichef]

Originally Posted by tarnationsauce2

[*]Signal to Noise ratio = The range between self noise and a reference signal. [More=better][/LIST]

Originally Posted by mshilarious

SNR = 94 - self-noise.

So the reference signal is 94 (and all the manufacturers use it (albeit with sometimes different weighting)?)?

[mshilarious]

Originally Posted by antichef

So the reference signal is 94 (and all the manufacturers use it (albeit with sometimes different weighting)?)?

Yes, the standard reference signal for microphones is 1kHz at 94dBSPL. The weighting factor for 1kHz is 0dB in every scale I'm familiar with--stated another way, all those weights are referenced to 1kHz. Such that if you have an A-weighted self-noise figure of say 14dBA, then your dynamic range is always 80dB (A, but often not stated as such).

A-weighting is pretty close to a first-order bandpass filter (6dB/octave slope) with corner frequencies of 10kHz and 500Hz. The rolloff accelerates to higher orders below 100Hz, but basically that gives you an idea of the range being measured--you can't easily hear noise about 10kHz, and below 500Hz, it's usually drowned out by program material, and you don't notice it as much anyway, even though it is technically audible.

[dgatwood]

Originally Posted by tarnationsauce2

One caveat is if you compare specifications of mic A to Mic B, be sure they use the same weighting, references, or THD.
For example Mic A has Max SPL = 125dB @ 0.5% THD versus Mic B that has Max SPL 127dB @ 1.0% THD.
They use different "THD" reference points. In actuality mic A could have a better Max SPL than mic B @ 1.0% THD.

In fact, it is pretty much certain that mic A has a better Max SPL. What those specs are saying are that at 125 dB of sound, mic A produces a signal that differs from the ideal signal by no more than 0.5%. (I'm being exceptionally sloppy here; it's actually a fairly complex set of mathematical sums to get that 0.5% number, but that's good enough for here.) Mic B produces nearly twice as much distortion at very nearly the same volume level.

[fcabanski]

Question on this.

The MXL 990 listed here: http://pro-audio.musiciansfriend.com...one?sku=271009

shows S/N at 80dB but an equivalent noise level of 20dB. By the calculation described (S/N = 94 - self noise) the self noise would be 14dB rather than 20.

Does this mic have more self noise? http://www.zzounds.com/a--2676837/item--AUTAT892CT4TH

It shows S/N at 60, which by the formula (60=94 - self noise) self noise would be 34dB.

And then this microphone: http://img3.musiciansfriend.com/dbas...pec/271593.pdf

Lists self noise at 5dB, but by the formula (it's S/N is 88dB) it would be 6dB. Even though there's the difference between what they list and the calculation, it's about 2/3 quieter than the second one (the Shure)?

[Shadow_7]

I think the numbers are fudged a bit between manufacturers. Since not all measure from the same distance from the mic and other factors. It's really only useful for comparing mics of the same manufacturer. If they use a consistent method across the board. And/or publish specs to start with.

You must also consider the output level of a given mic. If it doesn't put out that strong of a signal you've got to boost the gain on your preamp which has it's own noise specs. So all things are not created equal. Yes it's a reference where there might otherwise be none. But trust your ears more than the specs. I just glance at SNR, then look at frequency response and max SPL. Since the last two are more apt to determine if you can even use the mic in first place.

[fcabanski]

Of the three two are 20 - 20KHZ while the MXL is 30-20KHZ

SPL:

MMX is 103dB at .5%
The NTI is 137 at 1%

What's the difference between those specs - .5% as opposed to 1%.

Still, it looks like the NTI has a slightly larger frequency range, less noise, and a better SPL. Yes or no?

[mshilarious]

Originally Posted by fcabanski

Question on this.

The MXL 990 listed here: http://pro-audio.musiciansfriend.com...one?sku=271009

shows S/N at 80dB but an equivalent noise level of 20dB. By the calculation described (S/N = 94 - self noise) the self noise would be 14dB rather than 20.

Yes, since they state both figures are A-weighted, one of them is wrong.

Does this mic have more self noise? http://www.zzounds.com/a--2676837/item--AUTAT892CT4TH

Yes, but that's probably unweighted. Figure about 4-6dB less for A-weighted.

And then this microphone: http://img3.musiciansfriend.com/dbas...pec/271593.pdf

Lists self noise at 5dB, but by the formula (it's S/N is 88dB) it would be 6dB. Even though there's the difference between what they list and the calculation, it's about 2/3 quieter than the second one (the Shure)?

The spec is >88dB SNR. 89 is greater than 88

The Rode is orders of magnitude quieter than the Shure - dB is a logarithmic scale. 30dB (assumed A-weight) less 5dBA is 25dB, which is 1/18 the noise of the little Shure. Practically, the noise of the Shure would be very noticeable, and the Rode would seem to be noise-free (compared with ambient noise in any practical recording environment).

Of course, you have to consider the added noise of the preamp--nearly all condenser mics will have a signal hot enough to be above the preamp's noise, but not all. Just something else to be aware of.

[fcabanski]

I'm thinking of buying the Rode, but wonder how to ensure its noise is above the USB interface I use - TASCAM US 122.

[mshilarious]

Originally Posted by fcabanski

I'm thinking of buying the Rode, but wonder how to ensure its noise is above the USB interface I use - TASCAM US 122.

More math:

The Rode NT1A sensitivity of -32dBV/Pa, less its SNR of 88dB is -120dBV noise floor.

Compare to the Tascam--well, here we go, ambiguous specs. Preamp manufacturers do that all the time. To me, a proper noise spec clearly states equivalent input noise. Tascam states "dynamic range and signal to noise ratio" for the mic input to insert send as ">100dB". OK . . . as we know dynamic range is supposed to be maximum level to noise floor, and signal to noise ratio is nominal level to noise floor. The insert send max level is stated as +6dBV, and the nominal level as -10dBV. If we take the max level as the reference, that's an appalling -94dBV noise floor. I don't believe that, so let's assume they really meant 100dB SNR and 116dB dynamic range (yes, 116dB is more than 100dB, but that's still a lazy way to quote a spec). -10dBV nominal level less 100dB SNR is -110dBV, or 10dB noisier than the Rode.

This isn't terribly surprising as Rode prides themselves on extreme low noise, and a low-end USB interface is bound to be somewhat noisy. Note however that Tascam spec'ed noise at minimum gain, which is bold because noise is usually relatively worse at low gain (contrary to popular belief). It's just that no one really cares about noise at low gain, because low gain means you have a really hot signal. So maybe the higher gain settings are quieter.

One thing that Tascam is kinda trying to bury under the rug is the noise performance of its converter, which they state as 93dB. That's a 16 bit dynamic range--or is that the SNR, with some digital "headroom"? I have no idea, thanks to their somewhat obtuse specs. Maybe it's really 109dB dynamic range, who knows?

Does it matter to you? If you are doing nature recordings, probably. You will compromise the noise performance of the Rode when recording bird chirps or something. For rock music, it won't matter much. Acoustic music . . . you may or may not notice, it's hard to say.

[fcabanski]

I'm doing broadcast quality voice over recording, so I need as little noise as possible.

[mshilarious]

Originally Posted by fcabanski

I'm doing broadcast quality voice over recording, so I need as little noise as possible.

I wouldn't really consider that a noise-critical application. First, a medium like FM radio, the dynamic range is limited. Second, the traditional mics used for VO are low-output dynamics like the Shure SM7B. At nearly -60dBV/Pa, it's impossible for the SM7B to have much more than about 70dB of SNR to a very quiet preamp's input noise. So your Rode-Tascam combination should about the same as what the majority of radio stations are using.

Third, for VO your face is generally quite close to the mics. That means the mics see much higher than 94dBSPL, which is the reference level for a mic's sensitivity rating. So if your VO clocks in at say 104dBSPL at the mic's capsule, you just pushed your noise down by 10dB compared with the specs.

With the Rode-Tascam combo though, you might have the opposite problem: lack of headroom. The maximum mic input level is listed as 0dBu (-2dBV). That means if your Rode sees 124dBSPL at its capsule, it will clip the Tascam's input even at its minimum setting. That might sound unlikely, but human voice at 2 inches can easily hit that level with a good loud yell, scream, sung note, etc. That problem can be fixed with a pad, but it's something else to be aware of.

[fcabanski]

Voice over work is generally done at 6 inches distance. 2 inches if a whisper or an "inner voice" such as voicing a person thinking to himself, and 1 foot with head turned away from the mic if yelling.

Local market stuff is different, but even when I'm doing local market I stick to those standards.

What would you recommend as a good USB interface/Mic combo?

[Shadow_7]

Originally Posted by fcabanski

What would you recommend as a good USB interface/Mic combo?

Is there such a thing? USB Interfaces are limited. And the connection points can be interrupted fairly easily. Is it really something you want to trust a mission critical tool to? On location.

I used one live on location once. First time out and someone tripped on the mic cable and jerked the interface, and while everything remained physically connected, the recording application crashed as a result of the interruption. And since it was set it and forget it mode, an hour plus of the recording session never happened as a result. It was just a rehearsal that I was recording to get a feel for my gear, so it didn't matter at that time. But there was a lot of good stuff that didn't get recorded as a result. And at the end of a 2 hour rehearsal, I was left with about 15 minutes of recordings and only one song all the way through. That and my levels were too high, not realizing that one person blatting as loud as possible is still quieter than 20 people playing at the same time.

[KineticSound]

Originally Posted by fcabanski

Voice over work is generally done at 6 inches distance. 2 inches if a whisper or an "inner voice" such as voicing a person thinking to himself, and 1 foot with head turned away from the mic if yelling.

Local market stuff is different, but even when I'm doing local market I stick to those standards.

What would you recommend as a good USB interface/Mic combo?

Is that you, Frank??? Don't see many V123 users over here at the HR BBS!

For what we're doing, there are sometimes as many opinions on which is the best combo as there are combo's to begin with! I think it depends greatly on your room, your voice, your budget, and your technique. It also probably deserves it's own thread...

Many swear by their RE-20's. I like my Rode NTG-1 (poor man's MKH-416) and my Apogee Duet. I've also got a KSM-27 and a Focusrite TrakMaster, but it's far more sensitive to the room. While the SM-7b is still considered a "standard," I'm seeing far more in our business opt for the Electro-Voice's lately.

[fcabanski]

David,

Yes, it's me from voice123.

I currently have a Tascam US122 and MXL 990 mic. I'm thinking of an upgrade to lower noise from the mic and/or interface.

I notice the SM-7B has a response from 50Hz to 20KHZ, whereas mics like the MXL 990 have 20Hz - 20KHZ. Is the floor at 50 better than at 20HZ?

[KineticSound]

Well, think about what sort of sound is going into it and what the response will actually sound like. Technically speaking, 20Htz-20kHtz is a wider range than 50Htz-20kHtz, but unless you sound like Barry White or Don LaFontaine, I doubt you're getting down to that end of the spectrum. In fact, I typically use a roll-off for everything below 70Htz to eliminate any mechanical rumble from the building I'm in (a converted garment factory in a downtown area).

There's also a difference in moving from a condenser mic to a dynamic. I don't subscribe to the "you must use a condenser if you're a professional VO" school, mostly because it's a load of BS. A more accurate statement should be "a professional VO should use a high-quality microphone that correctly matches his/her voice." Notice I said "high-quality," not "high-price." There are plenty of folks out there who use MXL's, Audio-Technica's, Shure's, AKG's, EV's, RODE's, Sennheiser's, and Neumann's. There's no one definitive step up; you just have to find the one that's the right fit for your voice.

That said, if I were in your shoes and looking to move up from the 990 to another mic well-suited for VO work, my test list might look like:

* Shure KSM-27
* RODE NT-1, RODE NTG-1
* Audio-Technica AT-4033
* AKG C-3000
* Electro-Voice RE-20, RE-27
* Heil PR-40

The bigger steps up, if they're in your budget, might be:
* RODE NTG-3
* Sennheiser MKH-416
* AKG C-414
* Neumann TLM-103
* Neumann TLM-67

I have some other VO-specific resources I'd be happy to share as well offline. I'll PM you.

[mshilarious]

Originally Posted by fcabanski

I notice the SM-7B has a response from 50Hz to 20KHZ, whereas mics like the MXL 990 have 20Hz - 20KHZ. Is the floor at 50 better than at 20HZ?

OK, sometimes you have to take specs with a grain of salt. Shure specs, take those to the bank. MXL . . . I ain't saying it's a bad mic or a bad company, but consistent specs don't see to be foremost in their mind. Compare their website specs with the specs in the owner's manual. Is the noise 18dBA, 20dBA, or 14dBA? Why are the frequency response charts so different? Why do they call a mic that appears to be more hypercardioid cardioid?

I dunno. Like I said, none of that makes it a bad mic, but it does make it hard to select on the basis of specs. Specs can only tell you if something is really wrong, or if a mic would be really bad in a particular application. They can't tell you if a mic really sounds good.

Back to low frequency response: these are cardioid mics, and you are using them at 6 inches (I bet closer for the SM7B, but maybe I'm wrong). That means proximity effect, which is an enhanced bass response. I doubt that lack of low end will be a problem for you. It certainly isn't in most of the VO on the radio; your average VO artist seems to think that radio should be a big tub of proximity goo.

[fcabanski]

I only move in closer than 6 inches for a whisper. I'm not a deep announcer voice or proximity guy, but rather was trained to use a regular voice at very low volume - more of an average Joe. My voice isn't deep, but at the same time I can't afford to lose the low end because I don't have much there to start.