|
Mixing - and what the spectrum analyzer reads question
|
View this Thread in Original format
| Ray_Chappell |
Hey all - I'm finishing just my second song... so relatively new to mixing it properly. I've listened to it on various sound systems and each time find some more problems. In another thread, someone recommended Voxengo's spectrum analyzer - so I grabbed that, put it on the master channel, but don't know what the hell it's telling me. I've googled a couple things but keep getting information that doesn't help me in reading spectrum analyzers for the purposes of music production. This a snapshot of it on the master channel - can someone give me some knowledge on the best way to interpret it (I'm trying to determine if there is enough brightness to the highs and too much "low" for the bass)? Anything would be helpful... thanks.
Also, when I have one on the kick track and the bass track, they are sitting in the exact same frequency - but even relatively dramatic eq changes do very little to change that. ??
Thanks. |
|
|
| MrJiveBoJingles |
| quote: | Originally posted by Ray_Chappell
(I'm trying to determine if there is enough brightness to the highs and too much "low" for the bass)? |
You use your ears and a pair of good speakers for that. |
|
|
| derail |
Frequency analysers are good for comparing your overall mix to a mix you really like the sound of, as well as quickly identify an instrument that is causing a spike/ bump which you may not have focussed on. Ideally you'd train your ears and eq these as they occur, but until then an analyser can be a good safety net. I know professionals who still use the analyser as a safety net - it's a quick glance, couple of seconds to see "yep, nothing odd going on". For the piece of mind, it's a couple of seconds well spent.
A number of analysers let you save frequency graphs. So you can run one of your favourite tracks through it, capture the frequency graph and save it. Then you can have that displayed on the analyser while your mix is playing and you can see the two curves overlayed - you'll quickly see if your mix is way up or down somewhere compared to your reference track. (If the analyser doesn't provide this functionality, either get one that does, or make notes yourself on where the graph sits - which is more fiddly and time-consuming)
It's also good for finding instruments that are causing massive undesirable bumps - just have the analyser on the master channel and either solo each instrument to see where it's predominant frequencies are, or mute the instrument you think is the offending one and see if the bump disappears. Then you can fix the situation - try a different sound, lower the volume, use some eq - whatever your mixing experience tells you to do.
Yes, it's good to develop your ears so analysers aren't necessary, but in my experience your ears will develop just fine, even if you're using an analyser. You'll get a feel for what it's telling you and be able to adjust things before turning to the analyser, just so it doesn't tell you the same thing it always tells you.
There are some ways you can horribly misuse analysers - if your mix is very different to your reference mix, there may be a temptation to slap an eq on the master channel and just wrench the sound with massive eq changes to try and get it to look like the reference mix. This will in the vast majority of cases sound exceptionally bad. The first instinct should be to adjust the instrument levels - if the bass region is too hot, drop the level of the kick and bass. If the kick doesn't sound so good anymore at a softer level, consider using a different kick sample which sounds better and doesn't add so much bass. And so on...
It's not a problem if the kick and bass occupy the same frequency, as long as they are kept out of each other's way in the arrangement. A quick and easy way is to sidechain the bass, so whenever the kick hits, the bass level drops. They're both fine playing 65 Hz, as long as you're not trying to get both the 65 Hz signals playing at exactly the same time. If you want them playing at the same time, then yes, you'll have to get creative with eq (though generally it's better to use samples that don't fight in this manner. Which isn't always possible.) Another option is to automate the low end of the bass, just remove the low end whenever it's playing at the same time as the kick since the kick will provide enough low end on it's own. (though some producers have the bass provide the low low energy, beneath the kick, so you can take or leave this advice, depending on what your mixing preferences are). Or layer your basses and have them playing exactly the same pattern, but drop the low bass out on the notes which align with the kicks. That way the overall bass pattern will still be heard, but won't conflict with the kick. Get creative!
(just a quick note on that frequency chart you posted - it looks like it could sound quite horrible, there doesn't seem to be much midrange content in the track at all. Generally for trance, the graph will go up from 0 Hz to around 65 Hz, which is the loudest part, then gradually slope down to 20,000 Hz. There is a tendency to "scoop out" the middle to try and gain some clarity/ separation between the low and high elements, which results in some incredibly bassy and tinny mixes. Yeah, they're separated, but they sound horrible and on speakers which don't reproduce highs and lows well they're going to sound even worse. Just keep mixing, listening, learning and your ears will develop, so you'll know when something's too bassy or tinny or piercey) |
|
|
| Ray_Chappell |
| Thank you very much - that is tremendously helpful. Appreciate you taking the time to do that. |
|
|
| Dj Nacht |
| Thank you for that post! was looking for a good explination like that as well. |
|
|
| Derivative |
Its pretty simple really.
The y axis is amplitude (in decibels)
The x axis is frequency (in hertz)
If you run a 200hz sawtooth wave through a spectrum analyser (just to keep things simple) the resulting graph you get shows you at what frequencies there is the most power.
Its really easy to tell with a 200hz sawtooth wave because the biggest 'spike' will be at 200hz and there will be another smaller spike at the first even harmonic and a smaller one at the first odd harmonic etc etc all the way down the scale.
In another thread I mentioned the relationship between frequency and musical notation. a 440hz tone is (roughly) an A. 880hz is also (roughly) an A but an octave higher. 1.76hz is the next octave up. i.e. doubling the frequency is the same as going up an octave.
You can see this on the scale used on the x axis. You will also notice the grid lines are not evenly spaced. Take note of the spacings between 2khz and 5khz and the spacings between 1khz and 2khz. Although 4khz is not marked on the graph it does have a vertical grid line between 2khz and 5khz. The space between that line and 2khz is the same as between 1khz and 2khz - this is a span of 2 octaves.
Also take note that the decibel is a logarithmic unit. This unit is used because it allows you to represent a huge variation in volume using a small number of units.
Also, you will notice that in SPAN you can change the grid size and the update rate of the analyser. A spectrum analyser is just a realtime graph of amplitude over frequency. The faster the update rate the less it will 'hold peaks' giving you a more accurate representation of the varying power at a given frequency range over time. However, the faster you set the update rate, the harder it is to follow since it changes so fast.
You can slow this rate down to get a more general picture of the overall sound. Some spectrum analysers like the one in Izotope Ozone can hold peak indefinitely. Ozone can also infinitely average the spectrum whilst doing so. This is often used to follow the '6dB slope' which is very roughly a downward slope starting at like 4khz (indicating that there is a diminishing amount of power at higher frequencies). The 6dB slope is just a really really common observation - lots and lots of tracks roughly follow this line for whatever reason. However you shouldn't slavishly mix by it if it sounds crap to your ears. It is a useful guide to get your entire mix roughly in proportion with what everyone else is doing if that is what you wish to do.
SPAN also has a bias which is hard to explain (its the 'Slope' tab). The best way I found to understand how this works is to use a synth like Albino (which has a white noise and a pink noise generator). Pink noise as you may well know is a type of noise which has equal power in every octave band (every time you double up the frequency). Compared to white noise there is more low frequency energy but you can bias the curve in SPAN so a spectrum of white noise appears flat.
I use spectrum analysers all the time and I think you should to. However you must always keep the following in the forefront of your mind when you are mixing with a spectrum analyser:
What you see on the spectrum analyser doesn't always correspond to what you hear. You shouldn't trust an analyser to make mixing decisions for you. On the other hand, I know very well that I can't always trust my ears - especially when I've been mixing for a few hours and my ears feel like they are shot. This is when I make really bad mixing decisions that usually sound terrible the next day with fresh ears.
It is a tool. A visual aid much like dB meters. A reference point with which you can ground yourself if your mixing is starting to go all over the place. It is not something you should implicitly trust all the time and for the love of god, don't make your tunes so that the spectrum looks like some professionally released track. It will sound completely messed up if you do that. |
|
|
| derail |
"for the love of god, don't make your tunes so that the spectrum looks like some professionally released track. It will sound completely messed up if you do that"
Yes, this is an important point. For my overall mix, I use a program which contains frequencies of ten tracks which are really pleasing to my ears. Broadly, these ten tracks give me an overall "sonic ballpark", rather than using just a single track. It's good to know if my track is much louder or quieter anywhere than all ten of the tracks - then I focus on that and see if that's what I intended. Depending on the instrumentation I've chosen, I'm quite happy to have each of my tracks showing different frequency curves - it's just the way the instruments have been put together in that particular track.
Experience in mixing and using analysers will get you to a stage where you're comfortable using them and you're also comfortable with each song of yours looking different in a spectrum analyser.
A certain grouping of instruments playing in a certain key will have a certain sound and look a certain way in an analyser. It wouldn't sound good to push them into an exact frequency shape occupied by a different set of instruments playing a different key.
But yes, for overall mix decisions (like noticing 3 kHz to 8 kHz is all terribly quiet, for example) it can be quite handy. |
|
|
| Derivative |
Thats pretty much what I use it for on an entire mix. Look at broad trends and follow them loosely but not to the point where my mixes start to sound rubbish. Its good for helping you find out what 'ballpark' you are playing in as mentioned above.
I use spectrum analysers on individual channels too. This is where it gets more specific, especially on pitched synth sounds because they usually have very regular harmonics. You can see all the harmonic peaks on a spectrum analyser where the amplitude is highest around a given frequency.
Lets say you have a saw wave bassline with the fundamental at 60hz on channel A and a bass drum on channel B which has alot of power around 60hz during the decay phase. You can add a spectrum analyser to both channels and pick out the harmonics and fundamentals and see where stuff overlaps and may cause problems.
By the way, Voxengo GlissEQ has a better spectrum analyser built into it than SPAN. Well, I lie - it is exactly the same as SPAN but it has some extra bits. The best one is spectrum overlays.
You can overlay 4 different spectrums from 4 different channels using GlissEQ. So if you wanted to see a spectrum of a bass drum superimposed with the spectrum of your bassline you could do it. This is one of the best things about GlissEQ because you don't have to keep loads of windows open and keep flicking between channels to read a spectrum. It improved my workflow something obscene. |
|
|
| Ray_Chappell |
Great info - and very helpful. Appreciate it.
Now that I've switched to a Mac, have to find a new freebie plug in for this.... |
|
|
| derail |
You're probably aware already, but it's good to keep mentioning it - kvraudio is a good place for all your plugin needs, whether demo versions of commercial plugins or freeware plugins.
http://www.kvraudio.com/ |
|
|
|
|
