Q. Can you explain audio interface input sensitivity?

By Hugh Robjohns & Mike Senior

My interface allows me to switch line‑input sensitivity between +4dBu and ‑10dBV. My calculations suggest that should be a difference of 14dB, but it looks more like 12dB on my DAW's meters. What's going on?
Via SOS web site
SOS Technical Editor Hugh Robjohns replies: As anyone with a GCSE Maths qualification knows, the difference between +4 and ‑10 is, indeed, 14. However, as any audio engineer knows, the difference between +4dBu and ‑10dBV is actually 12dB. So your meters are telling the truth!
This is a very common, but very basic, misunderstanding for a lot people (and even some manufacturers), but it is well worth understanding the facts.
The reason for the apparent discrepancy is that the two standard operating levels your interface allows you to select are quoted in respect of different reference signal voltages. Those little letters after the dB values are there for a very important reason!

  
It may seem obvious that the difference between these two input sensitivity settings is 14dB, so you may be surprised when your DAW measures it as 12.

The professional operating level of +4dBu is measured with respect to a reference signal level (denoted by the little 'u') of 0.775Vrms, and works out at a signal voltage of 1.228Vrms. (The 'rms' appendage basically means that we are assessing the average signal level — and we're talking about sine-wave test‑tones here.)

The semi‑pro operating level of ‑10dBV is with reference to 1.000Vrms (denoted by the big 'V') and works out to 0.316Vrms. The difference between the two is 11.790dB, although, unless you wear anoraks or have a PhD, it's probably much easier and more convenient to think of it as a 12dB difference. If you want the maths (which is still only GCSE level, thankfully), here it is:

Decibels = 20 x log (signal voltage/reference voltage)

So:
20 x log (1.228/0.775) = 4dBu (Note the term dBu to denote the 0.775Vrms reference.)

And:
20 x log (0.316/1.000) = ‑10dBV (Again, the use of dBV denotes the 1.000Vrms reference.)
To calculate the difference between the two standard operating‑level voltages:
20 x log (1.228/0.316) = 11.79dB ~ 12dB

 Note that in this case, where we are simply calculating the ratio of two signal voltages, no reference is involved, so the letters 'dB' are used on their own.


Published February 2011

Korg microLAB in action at Las Lomas Elementary

Q. Can I get rid of string buzz?

By Hugh Robjohns & Mike Senior

I've got a recording of an acoustic guitar that I'm loath to re‑record, but there are several sections in which string buzz is clearly audible. Can I remove this with a bit of clever processing?

Mike Fenton, via email

SOS contributor Mike Senior replies: As far as after‑the‑fact mix processing is concerned, I'm not sure I can think of any decent way to remove string buzz, I'm afraid. The problem is that, unlike a lot of other mechanical noises the guitar makes, there's not really any way to get independent control over it with normal plug‑in processing. (I suspect that even high‑end off‑line salvage tools such as CEDAR's Retouch might struggle to make much of an impact with this, in fact.) In the case of pick noise, for example, the transient nature of the noise means that it can be effectively targeted with transient‑selective processors such as SPL's Transient Designer or Waves' TransX Wide. For fret squeaks you can use high‑frequency limiting, or simply an automated high‑frequency shelving EQ to duck the high end of the spectrum briefly whenever a squeak occurs, because such noises are usually brief and occur as the previously played notes are decaying (therefore having less high‑frequency content to damage). String buzz, on the other hand, isn't transient by nature and usually happens most obviously at the beginnings of notes, where the noise spectrum is thoroughly interspersed with the wanted note spectrum.

 
It's relatively difficult to fix fret noises with processsing, due to the very specific nature of the transients produced. For this reason, it's always advisable to record several takes of an important guitar part.

All is not lost, however, because you still may be able to conjure up a fix using audio editing if your recording includes any repeated sections and the string buzz isn't common to all sections; you may be able to just paste clean chords or notes over the buzzy ones. The main thing to remember is to try to put your edits just before picking transients if possible, to disguise them, but you should also be careful that all notes sustain properly across each edit point too, because you may not have played exactly the same thing every time. If you know that string buzz is a problem for you, I'd recommend doing several takes of guitar parts, as this will increase your editing options. If the guitar part is important enough that a bit of string buzz really matters, you should probably be comping it anyway, to be honest, if you're after commercial‑sounding results.


 

Published February 2011

*Korg Havian30: Digital Ensemble Piano - Listen & Believe

Q. Should I be mixing in mono?

By Hugh Robjohns & Mike Senior

I've read a lot of articles about the benefits of mixing in mono. So is pressing the mono button on your DAW's stereo output and turning off one of your stereo monitors the way to go? I've had a quick go from one speaker and mono switch, but it was a bit of a mess, to be honest! I've also read a lot on panning in mono, but I didn't think this would work.

Is it better to record in mono, pan, add effects, and mix to a single stereo master? The reason I ask is that when we recorded everything in mono it seemed to sit better than many stereo files fighting in the mix.

Via SOS web site

SOS Technical Editor Hugh Robjohns replies: There certainly are some advantages to mixing in mono. The main reason for checking the derived mono from a stereo mix is to make sure it still works for mono listeners, and there can be a lot of them. For example, a lot of portable radios are mono and all car FM radios automatically switch to mono whenever the signal suffers multi‑path problems or receives weak signals, which is surprisingly often in most places. Many clubs also play music in mono, and sometimes Internet files are converted to mono to reduce data bandwidth, or become near‑mono just because of the chosen codec.

So checking that the mix works in mono is a very sensible thing to do, and if you're going to do that, it is infinitely better to check mono on a single speaker, rather than as a phantom image across a stereo pair of speakers, because the latter over‑emphasises the bass end and stimulates more room reflections, which can be distracting and affect the perceived mix.

But what about mixing in mono? Well, it's generally much harder than mixing in stereo, but you'll get much better results for your effort. The fact is that when you mix in mono you can really only separate different instruments by using differences in their relative levels and spectral content. So achieving the right balance and applying the right EQ to separate sources, becomes a lot more critical: that's why it feels harder to do. But when it's right, it is very obviously right.
  
Even now that we have hi‑tech digital radio, a lot of listeners are still hearing a mono mix, as is the case for users of the Pure Evoke radio, currently one of the best‑selling DAB radios.

Conversely, when mixing in stereo you have the same level and tonal differences to help make the mix work, but you also have spatial position (panning). By panning sounds across the stereo image, you can make the mix sound great very easily, even if you have several near‑identical‑sounding sources. Yet, when that great‑sounding stereo mix is collapsed to mono, you will often find it no longer works, because those sources occupy the same spectrum and end up trampling all over one another.

However, if you can get the mix to sound good in mono first, it will definitely sound great in stereo too. I find it a lot easier and more satisfying to work in that way, although that's possibly partly to do with my formative BBC days working in mono. If you create the stereo mix first, it can be very frustrating afterwards to have to make it work in mono too.

Of course, the only problem with mixing in mono is what happens when you come to pan sources to create the stereo image. Panning a source will inevitably change its relative level in the two output channels. Depending on the panning law in use, this may also, therefore, affect the mono mix balance slightly. Since the mono balance is inherently more critical than the stereo balance, the result is that you end up having to work around the loop a few times. For example, you set up the initial mix in mono by adjusting the fader levels and, possibly, also using EQ to ensure each source occupies its own spectrum and doesn't trample over anything else. You then switch to stereo and pan the instruments to create a pleasing stereo image. This will usually modify the balance slightly, although you are unlikely to notice anything significant while listening to the stereo mix; it will still sound great. You then switch back to mono and, if you notice the mix has gone 'off' slightly, you can fine‑tune the fader positions to get the mix balance perfect once more.

Finally, check once again in stereo and print to master tape (or whatever!). In some descriptions of mono mixing, you'll come across the idea of finding the spatial 'sweet spot' for a source by adjusting the pan pot, while listening in mono. However, this is, quite obviously, completely bonkers! What you're doing in this case is fine‑tuning the mono mix balance by using the pan pot as an ultra‑fine fader, trimming the signal level by very small amounts. Sure, it may well make it easier to fine‑tune the mono mix, but there probably won't be much sense in the stereo image positioning when you finally do come to check the stereo mix. It's obviously far better to pan the sources while listening in stereo, so you can position them precisely where you want them, then revert to mono and fine‑tune the fader positions, if necessary, to make the mono mix work as well as it can.

Reverbs and some stereo effects can be tricky when you're trying to find a perfect balance in both mono and stereo. Almost all reverbs will sound much drier in mono compared to stereo, and so, usually, some compromise will be needed. If you adjust the reverb for a good sense of space or perspective in mono, it will often end up sounding a little bit too wet in stereo (although some people like it that way), and if you get the reverb sounding right in stereo, it will often end up a little too dry in mono. There's nothing you can really do about this; it's a fundamental issue with the way most reverbs are created and the way stereo works.
Narrowing the reverb width can make the differences less obvious — and some reverbs have a parameter to enable you to do this — but it also makes the reverb less spacious‑sounding in stereo. Some mix engineers like to pan mono reverbs with each individual instrument to try to maintain a better stereo‑mono balance, but it's a lot of extra work and I'm not convinced it sounds that much better anyway.

Published February 2011

Radiophonic Workshop on the ARP Odyssey with Matt Berry

Q. How can I link outboard to prevent degradation in quality?

By Matt Houghton

I have a few bits of outboard gear that I want to set up as external plug‑ins in Cubase. Should I be linking each bit of gear to different inputs and outputs of my soundcard (a Focusrite Saffire Pro 40), or should I just use a patchbay so that I can link multiple processors together in series? Presumably, doing it the latter way, I get less degradation of the audio signal as it's not passing through the Saffire's D‑A/A‑D each time?

 
Adding external effects with hardware can really open up your options in terms of adding character to your music. With good‑quality gear, you'd have to go through several stages of conversion to notice any degradation in sound quality.

 

John Corrigan via email

SOS Reviews Editor Matt Houghton replies: You are perfectly right in theory: yes, there is some distortion each time audio passes through your interface's A‑D or D‑A converter stages. So, if you're chaining multiple processors in series (say, an EQ and a compressor), then it's better to only pass through one stage of D‑A and A‑D conversion. But that's the theory and (as in all matters audio), in practice, it comes down to what you can hear.

With a good modern interface, like those in Focusrite's Saffire series, you have to go through many stages of conversion before you'll notice any audible degradation. This is especially true if you're using outboard to impart a bit of 'character' or 'flavour'; it's extremely unlikely that a couple of extra stages of conversion will be at all noticeable. If you're a mastering engineer then maybe you have good reason for worrying about this, but then you'd already know enough from listening to the difference that you wouldn't be asking this question! In my opinion, the benefits, in terms of saving time and being able to go with the creative flow of patching in your external effects as if they are DAW plug‑ins, far outweigh any theoretical disadvantage. Just remember to use and trust your ears!

Published March 2011