Gain Staging - Episode 1

Introduction Hi, I'm David Mellor, course director of Audio Masterclass. In this Sound On Sound podcast, I'll be talking about gain staging. This is a big topic and I won't be able to cover it all in one go. But this podcast episode will get you started and I'll take you all the way through in further episodes in this series. Are you worried about gain staging? Don't be. There's hardly anything to worry about. Welcome to my podcast series on gain staging, where I'll tell you how simple gain staging can be. Yes, it can be complicated if that's what you want, but it doesn't have to be. What I'm going to present here is based on my own experience going back to the late 1970s with analogue equipment through to the latest digital audio workstation systems used in home recording studios. If you're working at a professional level, then of course you know all this already and you'll probably be combining the power of the DAW with a rack full of hardware equipment. You'd never make a gain or level mistake anyway. But gain staging helps you achieve an efficient workflow. But I'll be mostly considering someone who's working in the box with maybe the occasional excursion in and out to a cherished piece of physical equipment, maybe a fully featured compressor, or maybe a guitar pedal. I'll mostly be telling you how simple gain staging can be and how it's mostly really difficult to go wrong. There's a lot of what I see as misinformation on gain staging about, mainly through over-complicating the issue. DAW developers make it easy for us never to go wrong - almost never to go wrong. Let's start. Using The Correct Terminology Firstly, I need to start with decibels to make sure you know as much about decibels as you need. As you probably already know, we use decibels to relate sound levels across any type of media. You can compare the level of one sound or signal with another, or the same signal before and after a level change. When we relate one sound to another or describe a level change, we use decibels in the form of dB. Simple dB is a ratio between levels. It doesn't describe any absolute level. To describe an absolute level, then, for sound travelling in air, we use dB SPL, Sound Pressure Level, which relates the sound you're hearing or measuring to zero dB SPL, which we take to be the quietest sound the human ear can hear. In the Digital Audio Workstation, we use dBFS, decibels relative to full scale, which compares the level of the signal to the highest level that can be contained in a regular 16 or 24 bit signal. This highest level is 0 dBFS and everything else is a minus figure. Internally, your DAW can go way higher than 0 dBFS. But I'll come back to this later. In analogue audio and I have to talk about analogue audio because it's still relevant in the digital age, in gain staging, we often talk about zero VU meter. VU stands for Volume Units and is the zero measurement on a standard VU meter. Electrically, this is plus 4dBu, there's another type of decibel for you, which in volts is 1.228 volts. One more thing is dBTP, decibels relative to true peak. I'll come on to this later in another episode in this series, but we don't need it for now. Now, two words that I need to pin down are level and gain. These are two different things and even professionally these terms are often used incorrectly and casually, I often even do so myself. Here I'm going to try and use Level and Gain correctly, consistently throughout this podcast. There's volume as well, but let's handle Level and Gain first. Okay, let's imagine a sine wave. No, let's hear a sine wave. 220Hz at minus 20 dBFS. It may come out a little different in the podcast, but all of my examples will be in proportion. Here it is. Why have I chosen 220Hz? Well, I could say it's my favourite frequency and I don't see why I shouldn't have one. It's also the note A below middle C. It's also, which is why I've chosen it, not as hard on the ears than if I'd chosen the usual reference frequency of 1kHz. This signal has a level, as I said, minus 20 dBFS.  Now, I'm going to apply 6dB of gain to it. It's now at a level of minus 14 dBFS. So you can see, level refers to the existing level of a signal in dBFS. Gain is applied to change the level, this gain being measured in decibels, dB, then the signal reaches a different level, in this case, -14 dBFS. Gain, therefore, is a change in level measured in dB. It can be, and often is, negative. Negative gain can also be called attenuation. Volume, well, it's often taken to mean the same as level. But I see volume as real sound travelling in air. So, an orchestra has a volume, your TV has a volume, your electric guitar through your Marshall stack has a volume, but signals travelling in electrical wires, or the digits of your DAW, have a level. Indeed, according to the good old internet, the first use of the word volume was in an 18th century review of a performance of Handel's Messiah, that praised the immense volume and torrent of sound. So, for me, volume refers to actual sound. There's another word, loudness. I'll come on to that too in due course. Well, I hope I got through that quickly enough. But without this understanding, a lot of what I'm going to say wouldn't have entirely made sense. But it will now. Why Use Gain Staging? So, if you ask me, in a nutshell, what is gain staging, I'd say it's managing levels. As simple as that. You use gain, or attenuation, to make sure that your levels are where they need to be and conversely, don't set your levels too high or too low. Saying where they need to be, I mean where they need to be technically so things don't go wrong and whatever needs to be optimised is optimised. When all that's ok, you can adjust your levels for the best artistic result. But what can go wrong if you don't get your gain staging right? Well, you'll probably find it hard to go so wrong that it'll affect your music, but it might and these will be the potential pitfall areas. Firstly, clipping. This is where you're trying to make the signal go higher in level than the system can cope with. So, in purely electrical equipment, if the highest level the equipment can handle is +24dBu, and you try to make the signal go higher than that, the peaks will stick at +24dBu, because the system can't go any higher. So we talk about the peaks being clipped off, because that's a very good analogy. In a regular 16 bit or 24 bit WAV file, the highest possible level is 0 dBFS. If you try to push the level beyond that, again, you'll get clipping.  Clipping causes distortion, really unpleasant distortion. It isn't anything like harmonic enhancement or warmth, it’s an ugly sound that's unpleasant to the ear. Okay, it does have some artistic applications but for now, just don't. At the other end of the scale, there's noise. Everything we use in audio generates a certain amount of random noise, which mostly sounds like a hiss. There's hum as well, but that's a different problem and I would always refer to it as hum, rather than just a type of noise. I have to say that noise is not the problem that it used to be, but we do need perspective here. Going back to the old and pre digital days of analogue tape, which was the mainstay of recording until around 2000, when digital audio workstations became powerful enough and reliable enough for multitrack recording. Analogue tape is noisy. Very noisy. When I say very, I mean that you could record on a professional machine that cost 10,000 Great British Pounds or more and hear the noise. You didn't have to listen out for the noise, you could just hear it. Dolby noise reduction systems reduced the level of the noise to near imperceptibility, but that was at the cost of more thousands of pounds for a professional type A or type SR system. By the way, type B on cassette decks did actually work, but that's another story for another day. So on analogue tape it was important to record to a high enough level that the sound you wanted to hear obscured the noise. To be fair, when the level of the signal is high enough, the ear doesn't notice any noise. But, for instance, classical music is full of quiet bits. Beethoven did not write a noise part into his symphonies. But too high a recording level leads to distortion. In analogue tape, this isn't so much clipping, but a gradual increase into more and more distortion, until it gets too unpleasant to listen to. So, in the olden days, it was always a battle between noise and distortion, distortion and noise. This persisted into the digital era when we recorded using 16 bits. 16 bits, in theory, gives us something like 96 decibels of signal to noise ratio between the highest level before clipping and the noise floor. But it's just good enough, so it was still important to record to a high enough level. But now, we almost exclusively use 24 bits when recording and mixing. The noise floor is way, way down at a theoretical -144 dBFS. OK, that's theory, but practice is close enough. You'd have to record way, way down in level for noise to be a problem. But there are plenty of old-school guys and girls around who are so ingrained in the idea of pushing the level high that it is still a thing to record to high levels. Not so high that you get clipping, but as high as you think you can get away with and these old-school types pass on their old-school ways to young people who've never used even 16 bit recording, let alone analogue. And I include myself. To me, any unused bits at the top end of the meter scale seem like a waste. Don't worry, I'll deal with this as we go along but, to recap, we gain stage because we don't want to hear the problems of distortion or clipping at high signal levels and we don't want to hear any noise. That's it. Gain staging will eliminate any problems due to distortion and noise. The Benefits Of Gain Staging Let's not dwell on the negatives. Gain staging can eliminate problems, but can it bring us any benefits? Yes, it can. Here's the good news. Firstly, gain staging can give you a lot of convenience. It will ease your workflow so that you don't have to worry about levels and you can enjoy your music making and recording. Secondly, while there really is very little to go wrong in modern DAW recording, processing and mixing, it doesn't hurt to work to a method so that you know your levels are always in the right place within a reasonable margin. It's something you'll benefit from being confident about while you put all of your creative effort into your music and sound. Thirdly, you don't live in a bubble. Sooner or later, your work is going to get out into the big, wide world, either to a client or the market. And, guess what? The level you put it out at is important. It's the final stage of gain staging and you really do have to get this right. I've mentioned analogue tape already, because in the olden days, it was vital to set a good compromise between not too much distortion and not too much noise. The Goldilocks zone, if you like. But because we so often relate DAW recording back to analogue, particularly with analogue emulation plug-ins, it's useful to have an understanding of how gain staging works in the analogue world. Here's the thing, we never used the term gain staging. In fact, we hardly even thought about it because you just had to do things right and if you didn't, it would be obvious because of distortion or noise. Or distortion and noise if things got really bad. In a traditional studio, we have a mixing console which has its own microphone preamplifiers. This feeds a multitrack tape recorder while recording basic tracks and overdubs. When the recording stage is finished, the multitrack tape recorder feeds back into the mixing console, the engineer balances the channels and records the output to stereo tape. We'd call this the master, but of course there would be another mastering stage for the vinyl record. So the first thing to do when recording a vocal or instrument, like we do now, is set the preamp gain. Typically you'd do this by soloing the channel you were using, get the musician to play as loud as they intend to, then increase the gain until you get a strong reading on the meter, but not hitting the end stop. That's a bit of a summary and different equipment can be different, but it's a good enough generalisation. If you do this correctly, then you can set the channel fader to zero, the output fader to zero, then record that vocal or instrument to a track on the multitrack and the level will be fine. Not too much distortion, not too much noise. You might consider pushing the level a bit and tolerate some distortion for a better signal to noise ratio. Again, it's a bit of a summary and different equipment can be different, but it's the gist. Where could you go wrong? Well, you could set the preamp gain too high. You would then lower the channel fader or the output fader to compensate. All would be well. You've done something a bit not quite right, but nothing has actually gone wrong. Aha, you say, what if I set a really high gain and pull the fader down a long way to compensate? Whoops, you've clipped the console. Normally a mixing console has bags of headroom so this is unlikely but it isn't impossible. Having said that, you're going to have to do something crazy to get a bad result. Let's think about something else. Again, in the analogue domain you're recording a lead vocal. It's a virtual certainty that you're going to want to compress it when you mix, so you record it at a nice, healthy level, finish up all the other tracks, then mix. You insert a compressor into the vocal channel and compress. But what's that you hear? Noise. Tape hiss. What's happened is that you've compressed the signal coming off tape. Because compression lowers the levels of peaks, you've used make-up gain to compensate and now the tape hiss is at a higher level, and you can hear it.  An alternative way of working, note that I didn't say better because there are always pros and cons with everything, is to compress the signal while you're recording, before it gets on to tape. This way the signal is kept well above the noise floor of the tape and you probably won't need to compress it anymore when you mix. Or even if you compress mildly when recording and you compress again when mixing, you've improved things and noise should not be a problem. Working In Digital Audio Okay, that's enough analogue. This isn't a lesson on how to time travel to the 1970s, but the perspective is important. We use digital recording now and what we've gained is freedom. These problems that we used to have every day with analogue, so much so that we didn't have to think about them, nor even call the solution gain staging, we have massive freedom now to do all the crazy things we want and get away with it. OK, setting the preamp gain too high will still cause clipping and you can't fix that later. But set it too low? 24 bit digital audio has such a low noise floor, it won't matter. You could set the gain 48 decibels too low, and that's a lot, and your recording would still be 16 bit compact disc standard. Well, okay, there's a little bit of theory versus practice there, but there's a ton of leeway, there really is. So as we shall see, effective gain staging is a simple matter of applying a few simple rules and with the wonder of modern digital technology, you can hardly go wrong. So that's it for the old analogue ways, I don't have to mention them anymore. Hold your horses. What about the VU meter? That thing with a moving needle, a scale and a lovely warm glow from the back light - an analogue filament bulb, of course. If you're listening to this podcast, and of course you are, because you are, then you will have undoubtedly watched YouTube videos on the topic of gain staging. Now, let me say there's a lot of fantastic information on YouTube. I don't want that to sound sarcastic in any way because there really is. I saved loads being able to replace the belt on my tumble dryer thanks to a YouTube video and my abilities with a silicone sealant gun are now second to none, thanks to YouTube, seriously. But, I have to say that some videos make things out to be more complicated than they need to be. That's my opinion of course and you may consider that against what I have to say. In your YouTube explorations of gain staging, you'll come across a number of presenters who suggest that it's useful to use VU meters to  check your levels. When I say VU meters, of course I don't mean old fashioned hardware mechanical VU meters, I mean plug-in equivalents, which you might be able to find for free from a generous plug-in developer, which is not unappreciated. So the VU meter, developed in 1939 and unleashed on a grateful world in 1942, according to Wikipedia - it’s a bit before even my time. This is 80 year old technology and very nice to have in its day. The VU meter in physical form is a simple electromechanical gadget with no active devices, no tubes, nor transistors. Drive it with a sine wave signal of 1.2 to 8 volts RMS or plus 4dbu and you'll get a reading of 0 VU. What's good about this, and it is good, is that it provides a standard for audio equipment of all types to adhere to. A reading of 0VU is good. A standard VU meter goes up to +3VU full scale and a more modern one will have a red LED to indicate even higher peaks. So in a properly set up studio, you could set your gains and faders to get a reading of 0VU, record that, and your level on tape would be good. So why is it that BBC engineers are known to have called the VU meter the virtually useless meter? This is because it under-reads. Give it a sine wave tone and it will give you an accurate reading. Give it any real world vocal or instrument “My heart will melt if you hold me while the pieces set again” and it will under-read. Give it drums and it will under-read severely.  With old school analogue tape this matters less than you might think, because tape is fairly forgiving of fast transients. So if a drum hit is a bit distorted, well hey, that might even make it sound better and a vocal is normally not too transient rich that the VU meter is too far out. BBC  engineers by the way, much preferred the Peak Programme Meter, or PPM, because it has active electronics that drive the needle up the scale very much faster. It still isn't instantaneous though, just fast enough for analogue. Did I mention that although I'm not BBC, I'm very much PPM? In fact, I've never used a VU meter for anything more significant than whether there's a signal present or not. In my old days I always metered on PPMs. If you're hearing a bit of negativity here towards VU meters, then I suggest that you use that to balance out the other opinions that you come across. I'm going to come back to this later in another episode of this podcast, but the reasoning behind using VU meters in the DAW is to set the levels of individual instruments and vocals in the tracks, before the signal gets to the fader. So, you insert a VU as you would an EQ or compressor. It doesn't change the signal, it just measures it.  I mentioned the goodness value of 0 VU earlier, that equipment manufacturers found common ground around this level. If you're hitting 0 VU, then you're probably not going wrong. In the DAW, we have a relationship between VU levels and dBFS. If you remember, I said that the highest level in a regular 16 bit or 24 bit WAV file is 0 dBFS and this is the level normally at the top of your DAW's meters. What we do, therefore, is equate 0 VU, or 1.228 volts of electricity, to -18 dBFS inside the DAW. This is an accepted figure going back to the early days of practical digital recording, even before we recorded using computers. Now, whatever disparagement I might have of VU meters real world or digital, it is absolutely a good thing that 0VU equals -18 dBFS. It gives us firm ground to stand on and where 0VU in the analogue world is the right level, in air quotes, -18 dBFS is the right level, also in air quotes, in the DAW. Now here's a thing that often is not made sufficiently clear, -18 dBFS is not the peak level you should be aiming for, it’s an average level, using the word average loosely because it does have a more precise meaning, but we don't need it here. -18 dBFS is not the peak level. If you're recording and the meter is bouncing around -18 dBFS, hitting maybe -12 or -10 on peaks, then that is perfect. Perfect. -18 is just a roundabout figure to approximately aim for. The DAW really doesn't care all that much. -28? Okay, that's too low. But it wouldn't make that much of an audible difference to your work. -8? Yes, that's okay too. But you'd have to be careful about peaks. No clipping, remember. So relax and let it all hang out. Your signal is bouncing around -18 dBFS, more or less. You're good. You're golden. Oh, a complication. If you look up EBU and AES specifications on this, you will see A, something different to what I've just said, and B, that they are different to each other. Oh dear, where I've got the figure of 0 VU equals -18 dBFS from is real life, as in VU meter plugins from Klanghelm, TB Pro Audio and Waves, which in my opinion are all good. They all read 0 VU for a level of -18 dBFS on default settings. A VU meter plug-in from PreSonus, on the other hand, reads -18 VU. Consistency is the last refuge of the unimaginative, so said Oscar Wilde. For this podcast series, 0 VU  equals -18 dBFS.  We all love WAV files, invented by the wondrous Microsoft Corporation, with help from IBM, in 1991. 1991!  It seems longer ago than that and I'm starting to wonder what I used before WAV, because I was recording audio on my computer before that.  Anyway, WAV files. They're more than you think and I have to create some clarity here. The WAV file itself is a container and there's a stream of audio data inside. There may also be metadata, or data about the data, which shows us that the WAV file isn't in itself the audio, it’s the container. Now, we tend to think of 16 bit WAV files and 24 bit WAV files because they're everywhere and there's hardly anything else anywhere. But although we use these uncompressed pulse code modulated files every day, many people forget that a WAV file can contain lossy compressed audio data. Or lossily compressed, maybe that's the adverb I'm looking for. Anyway, it's not the lossy compressed audio that's the issue, because it's rare and you probably won't go wrong if you don't know anything about it or have forgotten everything you ever knew about it. It's the 32 bit floating point format that has the potential to confuse the issue and 32 bit float is most definitely becoming a thing and it's a whole ‘nother gain staging thing. With a regular 16 bit or 24 bit WAV file, hang on, I've used that North Americanism regular when I should have said normal in the great British manner. I've done that a few times already. Well, I don't quite like that, because it sounds like anything that isn't normal is an exception, that normal is good, and anything that doesn't conform is at least suspect, if not outright bad. But regular? Well, we use 16 bit or 24 bit files regularly. I use them daily, at ten o'clock precisely and at hourly intervals after that. Now, kidding there, but I use them throughout every working day and so do you almost certainly. 24 more than 16, but 16 bit hasn't gone away yet. The thing about these formats is that a 16 bit file can encode audio into 65, 536 levels. A 24 bit file can handle 16,777,216 levels and in both cases, there's a highest level, which we call 0 dBFS.  These formats can go no higher. The 32 bit float format goes way further than that and can store any level up to a Saturn V rocket at 20 paces and I'm not sure that's an exaggeration. This is a good thing. It means that you don't have to worry about level. Gain staging becomes almost irrelevant, at least as far as the audio files are concerned. I'll have more to say about 32 bit float in another episode in this series, but for now, I have to make it clear that when I'm talking about gain staging in relation to WAV files, I mean gain staging in relation to regular 16 bit and 24 bit WAV files. WAV files that contain 32 bit floating point audio are a very different matter. I think that's enough for now, enough to be getting on with, at least. What I've done here is set the scene and define a few parameters that I'm going to need for the next episode of this podcast series. Three Golden Rules Of Gain Staging For now though, I'll leave you with some food for thought in my three golden rules of gain staging. 1. Don't clip when you're recording. No red lights. 2. Don't clip when you're bouncing. No red lights. 3. Keep in mind that the level you're putting into your plug-ins may be important. Not so much with out-and-out digital plug-ins, but more so with analogue emulations. I'll be delving into and expanding on these points in my coming episodes. That's it for now. I'm David Mellor, course director of Audio Masterclass. See you in the next episode. Thank-you for listening and be sure to check out the show notes page for this episode where you'll find further information along with web links and details of all the other episodes. Oh, and just before you go, let me point you to the soundonsound.com/podcasts website page where you can explore what's playing on our other channels.