Digitally Imported Homepage

Go Back   Digitally Imported Forums > General Discussion > The Artist Lounge
Register FAQ Members List Calendar Search Today's Posts Mark Forums Read

Reply
LinkBack Thread Tools Display Modes
Unread Aug 30th, 2003, 09:05 AM   #1 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Audio Production Theory

Well in this thread I'll be giving terms, ideas and practices on how to create sound, also know as beats or music.

This may seem technical at times, and may seem fun at others but everything I'm going to tell you is *very* important. You should take to heart what I post and learn it. You should try to keep it in your mind while you try your hands at production.

It's going to take me some time, so please if you're going to post in here post relevant information. No random crap without backing it up. If you talk about gear please say why you like it and what makes it so good.

Thanks, and here we go.

-maz
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 09:49 AM   #2 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default psycho acoustic cues

When creating sounds you need to understand how the human ear percieves sound waves and the standard levels that are involved. You need to understand the difference between dBspl and dBu. You should know what a dB is, and the cues that humans have to understand directional sense when it comes to sound.

Psychoacoustics pretty much means how we hear sound, and how it affects us. There are a few things that determine our sense of direction and the size of a room. These are quite important as it can decide how you apply DSP (digital signal processors) to a specific track within a song.

First cue for psychoacoustics is intensity. This is important because if a wave form is more intense on the right side than the left, we'll know the sound is coming from the "left."

Second cue is arrival time, if you hear a sound on the right before you hear it on the left you know it came from the right. If sounds come in at the same time, then it's coming from your center image.

Wave form complexity. As each waveform bounces around the area you're at, it will become easier to detect the direction of the sound.

Phase is the final cue, and it's the most important when it comes to audio. When things are out of phase you can get some interesting results, if it's 180 degrees apart you'll cancel out that frequency. It pretty much means two wave forms are identical but arrive at different times. This is how you get that neat flanger effect.

-maz
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 10:21 AM   #3 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Standard reference levels, units of measurement.

In a perfect level we wouldn't have distortion, tape saturation or those ugly digital bleeps when you go past peak. You generally want to record the "hottest" sound possible, this is 0dB.

A dB is a decible, it's calculated by dispersing .775 volts of electricity over a 600 watt resistor. It's important to understand how you reach this calculation because in an analog world electricity is your way of creating sound, while in digital it's created with ones and zeros to mimic this.

You might read things like dBspl, dBm, dBu or even 0VU. Don't worry, I'll post what all these mean and their significance. You should also know how VU meters work over PPM meters.

dBspl is a measurement for the threshold of hearing. It stands for sound pressure level. It's measured .0002 dynes/cm^2. Little bit of extra knowledge for you there.

50-55 dBspl is your average conversation, 135 dBspl is where you get physical pain in the mid frequency ranges, and 160 dBspl is the sound of a DC10 air plane taking off, or ear damage.

+4dBm is 0VU, this is +4dBm to your standard dB (remember the transistor?). dBu is really the same thing in practice.

There are two different types of meters, ppm and vu. They really are quite different and will give you different read outs depending on what's happening with the signal.

VU meters take the average, in other words they don't change with transients. Think about humming into a mic, then start clapping while you hum. Each of the claps your hands do are called transient sounds. They will not cause the VU meter to move at all if your humming is a constant sound (highly unlikely). VU meters are more like human ears. The human ear tends to dampen transient sounds. That's why lots of clapping at the same tone, pitch and "volume" sounds louder than one person.

PPM meters are peak power meters, and will register transients. They give you the real read out of what's happening to the signal, and will let you know if you're going to distort easier.

You'll hear a term called "balistics" with meters, this just means how they are. Either PPM or VU.

-maz
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 10:30 AM   #4 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Signal to Noise Ratio

For starters, we're going to need a picture to make this part easier. Then I'll explain why I created something so cheap to get the point across.

Signal to noise is simply the span between the loudest and softest a sound source can produce (read instrument, voice, what have you). If you record too loud you'll distort, saturate the tape, clip the speakers. If you record too low the sound will be completely useless as it's in your noise floor.

Here's the image we'll be looking at:

Pretty simple, just a few lines in a rectangle but it's going to make my life easier describing how this works.

#1 is the loudest sound, it's where you saturate.

#2 is your average signal level, 0dB on a meter.

#3 is the noise floor. Where you have your analog hiss, it's around 60Hz for the most part, but you'll want to keep it to make it sound more natural. In the digital world this level is -20dB from analog.

Between #1 and #2 is called your head room, the area you have to peak and not distort.

Between #2 and #3 is your signal to noise ratio (I'll refer to it as s:n from now on).

Between #1 and #3 is called your dynamic range, the full range of frequency something can create.

Record too low and you're going to have a big s:n, not really good as your hiss will be more audible. So the game is to try to keep everything right around 0dB, it's not going to be 0dB completely, but try to stay around there.

-maz
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 11:05 AM   #5 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Wave forms and phase

Phase is very important when it comes to audio production. It shows up in everything because it can either give you a node or antinode, or can cancel out certain sounds (frequencies). In order to understand this we're going to do just a simple sin wave for starters. You should know that the sounds we hear are complex wave forms and nothing this simple, but it gets the point across.

If you don't know what a sin wave is, here's an example:


If you can imagine a line going through the center, the peak above is called a peak, imagine that. The downward curve is called a trough. This picture is one full oscilation. All an oscilator does is repeat wave forms, pretty basic idea isn't it?

Low frequencies will have higher bandwidth (the distance b/t peaks) than high frequencies. It's this way because low frequencies have a smaller amount of oscilations.

Frequency is really defined as the number of oscilations in one second. This is measured in hertz (Hz). You usually convert to kilohertz when above 1000 because it just makes sense, but there's nothing wrong with saying "1200 Hz) when describing a frequency range.

If you combine two peaks, or two troughs that is called an antinode, it doubles the amount of dB in the frequency, so if two peaks hit at +3dB then they will sound louder because it will be +6dB. If two troughs hit at -3dB then it will sound quiet because it's now -6dB. It's important not to confuse to troughs with a node though.

A node is when two frequencies hit in the same amplitude but at different times of 180 degrees. This will effectivly 0 out the wave and cancel all sound. It's easy to confuse this with two troughs because doubling the amount of attenuation can seem like the same thing.

Phase tends to be important because you never want to cancel out all of your sounds or frequencies. This can happen when you're recording with mics. This is also why there is a 3:1 ratio rule when using many microphones in a recording situation. So you don't end up canceling out what you're trying to record because mics really are stupid.

-maz
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 11:22 AM   #6 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Tidbits of information.

Frequency ranges:
Highs - 5khz and higher
Mids - 400hz ---> 5khz
Lows - 400hz and lower

Human ears can hear pretty much from 35Hz p to 17.5khz. Human ears are most sensative between 2khz and 4khz. This is actually the range of the human voice, makes sense huh?

Little trick, to increase "volume" without changing the amplitude (real volume) increase the mid range.

Harmonics are doubling frequencies, so 800hz and 400hz are harmonic to eachother. This means an octave is just the span between harmonics.

Ambient noise is the sounds happening around you, such as air conditioning, electric hum and what not. If you're in a crowded restaurant, the people speaking all around you would be ambient, while those closest to you would not be.

Fletcher Munson is a ratio of spl to hz. Pretty much lets you know that the bass frequencies need to be turned up, eg bass boost/loudness on your home stereo.

Pink noise is all audible frequencies that the human ear can recognize where each octive is at the same energy level.

Sweet spot is where the sound from your monitors is true. Means the room isn't changing the sound at all.

Doppler effect is when speed of an object changes the speed of the sound coming off of it. Easiest way to describe this is when you hear a car. It's speed changes how you hear a sound that is really the same throughout.

When somebody refers to something as warm that means it has an increased sensativity to lower frequencies. Bright just means the same but for higher frequencies

-maz

Last edited by maz : Aug 30th, 2003 at 11:27 AM.
maz is offline   Reply With Quote
Unread Aug 30th, 2003, 12:38 PM   #7 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Signal Processing 101

There are two types of signal processors. These are time based and dynamic based. I'll go into each one in this post, it's quite a bit of information, but I believe that it's really important that you understand how they work as they can add life to your songs or can completely destroy a good one.

Time based processing

There are three time based processors, these are reverb and echo. Now you'll notice how I will use the word "echo" inside reverb, you just need to realize that when I talk about echo processing it's different than the echo in reverb.

Reverb(eration) is many repititions becoming more closely spaced with time measured in the time it goes away. This is actually measured in seconds (quite important). It's measurement is called "delay."

There's four types of reverb methods, reverb plate, spring, acoustic chamber and digital. Digital is the most convienent, but can also suck badly if you don't have a good DSP for it.

Plate reverb is a combination of a suspended plate with contact mics connected to it. The signal will cause the plate to vibrate, the contact mics pick up these vibrations and turn them into sound waves (electrical signal). It creates a really nice sounding reverb, down side, it's really expensive to own one of these, and a huge space commitment.

Spring coils are common in guitar amps. It's just a suspended spring that will vibrate to create the reverb.

Acoustic chambers are rooms that have hard surfaces, such as ceramic, glass and are designed to bounce sound around to create the reverb.

Digital reverb is very convienent because you can do just about any type of reverb, and a good one will create great effects that the human ear can't notice. Generally the cheaper (such as the yamaha spx90) the worse it will sound, but can get by in a pinch. It's also nice that you can recreate the exact same reverb by the touch of a few buttons. The other ones you're not going to get the same reverb twice.

Reverb is used to make the human ear think the room the sound was recorded in is bigger or smaller than it really is. It's a great time based process that can really add a lot to a track within a song.

The equation for reverb is: RT(sub60) = V * 0.049/AS. RT is reverb time, V is volume, A is average absorption of the enclosure, S is total surface area in square feet.

Echo is the second time based process. It is not reverb. It's one or a few repititions of an audio signal measured in time between the original and first echo. This echo is measured in miliseconds rather than seconds.

Originally echo was created with a three head tape machine by sending the signal back through the record head before it could go back to the console. Now adays it's done digitally.

The effect of doubling uses echo with a very small delay, usually around .25 miliseconds and adding it to the "dry" (unaffected) signal. It gives a sound quite a full feel.

You can also do phasing or flanging. It's also called a comb filter. You take the dry signal, and change the time back and forth on the processed signal. By making the processed signal go back and forth over the dry signal you're going to start cancelling out frequencies, peaking them and doubling. It's a swishing sound when you actually listen to it.
maz is offline   Reply With Quote
Unread Aug 31st, 2003, 02:35 PM   #8 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default Clarification, Corrections & Amendments.

For the most part, the lump of information posted is, appropriately so, very basic. These factors have been observed by all, irregardless of weather you know it or not.

However, as Maz quipped earlier, I would pipe up with some corrections, so here goes. Sorry that this wont make things any easier....

Under tidbit of information;

---The frequency of the human voice obviously differs with gender, body size, race, and language spoken, but for males starts much lower than posted, at around 700 Hz for males, and goes to about 4k as mentioned. Females often dip as low, but are on average higher.

---Harmonics are (sometimes) alternating patterns of frequency integers, but do not necessarily equate out to octaves spanning harmonics.

Why? This is no easy answer (well it kind of is), so here goes.

A harmonic is essentially a fundamental tone. The resonation or repetitive cycle of a waveform. What is known as a 2nd harmonic is really the Overtone also called the 2nd partial (the first partial is the fundamental) . The first overtone is 1 octave up. So for example, a C note of 256 Hz (cycles per second) when resonating bears a 2nd partial of 512 Hz. The next resonant step would be a factor of the fundamental, or 768 Hz, but would sound as a note of G, in the octave shelf one higher (obviously.) So a ratio of 5:2 of the fundamental. This is known as a 'perfect fifth' and this is where what Maz said falls apart. An octave is not a span between harmonics, for, just beyond the next octaval harmonic (the 4th partial,) come a major 3rd (played in E @ 1280,) another perfect fifth and a minor seventh (B- @ 1792). And between the next C octave shelf (which is also now the eighth partial,) comes seven harmonic resonations before another C octave harmonic.

Essentially, harmonics are progressively dissecting subaternates based on multiples of the fundamental wave frequency.

Sorry for the run on.

To amend a thing or two,

White noise- Is equal energy at every frequency, so as to be full spectrum noise, per se. However, random 'capture' of noise, accounting for most of the common noise detectable in todays systems, are often (but not exclusively,) induced by higher frequencies.

It is certainly recognizable by the high frequency, random 'hiss' or 'snow' present in absolutely any tone you've ever heard. While an all digital tone generator will store the data of a sound in 1s and 0s, devoid of whitenoise, as soon as it is replicated by speakers delivered signal by electrical current (analog signal) it gains white noise.

In analog gear, white noise is an intrinsic decoration of any waveform, and within its randomness' carries harmonic information up to and beyond practical hearing, then replication (providing your speakers are better than your ears.)

Pink noise- is defined by equal energy at every octave (formula 1/f) and therefore, as the octaves increase, the energy decreases (by half.) This is why pink noise always sounds bassy. Pink noise bears logarithmic qualities and has statistically equal chance of possessing any frequency at an amplitude descending @ 1/2 pole.

Pole- A pole is defined simply as 6db/octave, differing the power as the Hz multiplies or divides by 2. For example, a general rule is that as the power doubles, it effectuates the dB by +3. Therefore, a 1 pole LPF (low pass filter) will trim the high frequencies at the given point (either sliding-scale or set cutoff freq.) by quartering the power to each successive octave.

I know, rudimentary stuff some, needlessly technical for most, but just fyi.
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.
Nolnova is offline   Reply With Quote
Unread Aug 31st, 2003, 05:49 PM   #9 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default In Practice

To quickly test what Maz said about phase cancellation, try this.

For those of you with soundforge, load a wave. A sinusoidal shape can be hurriedly constructed by going to Tools->Synthesis->FM... and OK the default 440 Hz A note.

Crop a single cycle, then copy it into a new file. To the new file goto Process->Invert/Flip. Copy it.

Highlight the first cycle, then goto Edit->Paste Special->Mix... If the volume levels are set even, you will hear no sound. By reducing the volume on one of the channels, you can hear the other become increasingly more prevalent.

While this is super basic, and rather un-inspiring, it illustrates a much more profound concept. Should you be using mechanical energy as your medium, like two speakers (with one of the speakers' polarized,) and sourcing from analog gear, which is micro-cosmically syncopated by electron-flow and resistance nuance, the importance of avoiding, or employing with intelligence this dramatic behaviour surfaces. To validate that, say, 100 watts of audio has phased itself out is remarkable, and drums up a slew of new questions about the physics of all nature. Think; rustling leaves, behaviour of light, tidal forces, and interplanetary physics...
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.
Nolnova is offline   Reply With Quote
Unread Aug 31st, 2003, 08:08 PM   #10 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default A Sweet Note

Just to elaborate on sweet spots...

This is one of those very basic concepts I was referring to, but it is so fundamental to production, that unless you garner an understanding of this concept, your production will suffer despite your skillset.

It is simple enough to detect, but the plethora of articles written on the subject, that I can attest have been frequent since the mid-80's, suggest that this is both an ever-evolving science and a delicate process of trial and error. The jury seems to be out on the finer points, but we need to start by addressing the issues.

1) The positioning of the speakers. This is the logical place to begin. The human pinnae (outer ear) can detect sound from about 270 to 300 degrees, but are most proficient at focusing sounds from the (from overhead) 10 o'clock and 2 o'clock position.

2) Control centred. Surround whatever device you sculpt the acoustics of your tracks with. If you mix down on the computer, it is most practical to flank your monitor with your, ehm, monitors. Otherwise, have them facing your mix desk.

3) Distance. They're called near field monitors for a reason. The cone shape, tweeter radius and size all cater to a close range. Long throw speakers are usually utilized to augment near fields, but were never designed to be stand-alone. Near-fields should be 3 to 8 feet (1 -2 1/2 meters) from your head.

4) Dampening. The last thing you want is to hear your track bouncing off the wall behind you. It retards your ability to condition your music with spatial dynamics. Ironically, clutter in the studio is a good thing, as objects dissipate sound beautifully. Professional sound dampers are, of course, optimum, but my inherent lack of organisation does the trick too.

Anyway, there are bound to be more factors that I've missed, some conditional, some universal, and people will undoubtedly hit on them in later posts, but these four are a must.
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.
Nolnova is offline   Reply With Quote
Unread Sep 2nd, 2003, 12:33 AM   #11 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default Wall of Sound

Names like Phil Spector and Thomas Dolby, whose music never initiated their names into the mainstream, where some of the largest contributors to audio production, handsdown. Both were master audio engineers, and helped pioneer how we hear the standard of sound quality.

In particular, equalization and spatial dimension are the two most commonly employed tools of the trade, two of the most pivotal aspects of a quality recording, and ironically enough, two of the aspects most mishandled by producers.

Phil Spector, despite launching the likes of Sonny and Cher, working with Johnny Cash and Ike and Tina Turner, and owning multiple labels, contributed nothing greater to every production on this planet than his 'Wall of Sound' engineering technique. Lots of us have heard the phrase 'wall of sound' in passing, and I'll try and touch on the notion as best I can.

Wall of sound is, in part, the structure used to build wide sounding, clean layered productions. Using equalization (varying the amplitude of frequency ranges,) and panning (lifting the volumetric presence of audio in one channel or the other,) a confluence of multiple instruments can convey a sense of depth, volume, dimension and feeling as though each instrument were being played live.

The idea is to cordon off each [acoustic] channel to a separate section, as though layering bricks in a wall. I'll sight a techno related example.

If a hi-hat were to be sonically centred around a frequency of 6,000 Hz, and a stab-like-string were also to share that range, say 5,000 Hz, an engineer would want to pan these sounds to separate points, essentially laying the bricks on the same height (frequency) of the wall, and distanced far enough apart to fix them flush to the wall.

Now say the stab-string was stereo to begin with, and therefore had its own breadth of dimension. Ergo, it demanded space for multiple bricks along that frequency height on the wall. Equalization (eq) is used to 'shave' off presence, or brightness, from whichever side (speaker wise,) the hi-hats would be occupying.

This shaving or equalization is also used to allow bricks just below and above other bricks to fit together. So say the stab-string sat 30% on the left channel, and right below, a bright snare 30% left in the 4,000 Hz region. Reducing the eq of the slightly higher than mean snare frequency (4,500 to 5,000 Hz,) as well as that bracket on the left channel of the stab-string (which would be its lower end because it's @ 6,000 Hz), would reduce frequency conflict on that point in the wall.

Now if these are the bricks (the snare and the stab-string,) that end up vertically adjacent to each other in the wall, then, just like in brick laying, it would be logical to stagger the panning in conjunction with the aforementioned eq'ing. So push the stab-string a touch to the left, and the snare 10% to the right of where it sat. Now the same would be done to accommodate for the earlier hi-hat I mentioned, with it easing off a conflict with the right hand side of the stab-string.

You see where I'm going with this.

In today's mastering world, we'd have to pre-adjust for the post-mastering tools like exciters, spatial expanders, multi-band stereo-master eq, and stereo imagers, but that's another post.

Most audio control devices have panning, and some eq. Professional production requires that the eq's be sweapable, or what are known as 'British eq's. On the computer they are aptly called paragraphic eq's. These allow you to pinpoint the variable height of the bricks you want to shave.

Oh, and as a final note at this time, a lesson long known by analog enthusiasts, subtractive eq'ing is many times more effective than additive. Additive should be used very sparingly, largely because it serves other purposes in a quality production, but the fact it can easily ruin the tone of the audio should be reason enough.
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.
Nolnova is offline   Reply With Quote
Unread Sep 2nd, 2003, 06:28 AM   #12 (permalink)
 
mike-hunter's Avatar
 
Join Date: Apr 2002
Location: Hamburg, Germany
Posts: 863
Default Re: Signal to Noise Ratio

Quote:
Originally posted by maz
#3 is the noise floor. Where you have your analog hiss, it's around 60Hz for the most part, but you'll want to keep it to make it sound more natural. In the digital world this level is -20dB from analog.
Isn't it called hum, when you have a rather specific frequency? Hiss is noise and noise contains all frequencies at once to some degree.

Doesn't make so much of a difference as you want to avoid both.

Mike
__________________
www.mike-hunter.de
www.trance-maniacs.com
mike-hunter is offline   Reply With Quote
Unread Sep 2nd, 2003, 06:56 AM   #13 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Re: Re: Signal to Noise Ratio

Quote:
Originally posted by mike-hunter
Isn't it called hum, when you have a rather specific frequency? Hiss is noise and noise contains all frequencies at once to some degree.

Doesn't make so much of a difference as you want to avoid both.

Mike
Actually hiss is in the noise floor as well as "hum" but they're pretty close in terms of what they are, extra noise in the signal. Electrical noise is usually the most common and it is at 60Hz.

-maz
maz is offline   Reply With Quote
Unread Sep 2nd, 2003, 05:18 PM   #14 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default White Picket Fences

While on the topic of noise floors, I might as well use this opportunity to talk about gates.

Gates are quite simply threshold based modifiers to previously defined dynamics. There are filter (VCF) gates, fx gates, trigger gates, envelope gates and others. But the most common gate, to the best of my knowledge, is the VCA or amplitude gate.

Amp gates are found on everything from compressor/limiters to fx boxes (I guess that isn't to broad an application ) but typically do the some thing. They allow passage of sound until a volumetric threshold is hit, then modify the sound that fails to achieve the threshold limit.

More times than not, that modification pertains to an outright cutting of all volume at the threshold and thereafter, however, due to the expansive variety of conditions subject to audio, gates must factor in a cut-off pole and algorithm. For example, a compressor/gate/limiter applied to a digital audio source can not have a VC (voltage controlled) hard gate system, because of flagrant bit spikes that give dyjutle sound its krispy edginess. It would pop, click and sputter, making more noise than it was intended to remove. So there are linear or exponential slopes attached to the removal of sound once the threshold is broken. The linear slopes deride the volumetric presence of the audio below the threshold, as though it were being compressed by a condition and without a peak- or trough-stop value, and the exponential slope uses millisecond-long trail-offs then curves down toward a total volume hack as the gear will allow. Both systems are standard.

I'll try my hand at some sort of ASCII diagram;

1.

12dB ...../\..........................
8 dB ..../..\.............../\........
5 dB .../....\............./..\.......
2 dB ../......\.........../....\......
0 dB----------------------------
-2 dB ''''''''''''''''\'''''''''/''''''''''\'''/
-5 dB ''''''''''''''''''\'''''/''''''''''''''\/'
-8 dB '''''''''''''''''''\'''/'''''''''''''''''''
-12dB''''''''''''''''''''\/'''''''''''''''''''''

2.

12dB ....../\.........................
8 dB ...../..\.............../\.......
5 dB ....|....|..............|.|......
2 dB ....|....|..............|.|......
0 dB----------------------------
-2 dB '''''''''''''''''''|''''|''''''''''''''''''
-5 dB '''''''''''''''''''|''''|''''''''''''''''''
-8 dB ''''''''''''''''''''\'''/''''''''''''''''''
-12dB''''''''''''''''''''''\/'''''''''''''''''''

Ok, that didn't turn out great, so let me explain. Figure 1 shows a triangle waveform, that except for the first cycle, shows a decreasing amplitude. Now lets say we apply a gate with the threshold at 8dB. That means anything below 8 dB is cut off. Figure 2 shows the resulting waveform all hacked apart. Notice the only portions still intact are those at, or exceeding, 8 dB.

Now this diagram also exemplifies why there are linear or exponential slopes on gates (and why ASCII is no longer used.) Obviously, if this diagram (no. 2) were indicating a pair of triangle cycles, we would be looking at only milliseconds of time. As all waveforms push and pull the speaker cone, and thus crosses the threshold to pass the centre point, a gate without a slope would be popping and clicking as rapidly as the frequency cycled.

So, a gate can be used to cut-out any audible hum (a tonal noise which would serve as the noise floor if present,) and noise (ever present, despite any measure. The Sun itself creates enough EM noise to give every piece of conductive metal a noise floor. Even if we weren't engulfed in our own hissy-fit of com-spec frequencies.)

Gates are a must when instruments are overdriven (distorted,) but are just plain good protocol when they can be ultilized transparently.

P.S. My ASCII diagrams didn't reproduce the spaces I made, so I had to re-edit them with dots as spacers. As a result, they look even crappier than they did initially. If anything doesn't make sense just pm me.
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.

Last edited by Nolnova : Sep 2nd, 2003 at 05:28 PM.
Nolnova is offline   Reply With Quote
Unread Sep 3rd, 2003, 08:22 PM   #15 (permalink)
DI Extreme Addict
 
DJ Squawks's Avatar
 
Join Date: Jul 2002
Location: Antarctica
Posts: 1,222
Default

I really appreciate all the deep information you're providing, Nolnova. You have a very nice way of explaining your points across; apparently you are extremely knowledgable in this field.

I'll add a tiny tidbit to what Nolnova said earlier regarding speaker positioning in terms of having optimum acoustics: asides from having the appropriate speaker distance and location from your head, it's also good to have optimum speaker height. Speakers should be head-level; the tweeters of the speakers should barely be above your ear level, although setting them at ear level is also ok. It is funny to see how many people out there have their speakers laying on the floor or on their desk at elbow level. Go out and buy some speaker stands if you must. You'll notice a difference.
__________________
There's too much blood in my alcohol system.
DJ Squawks is offline   Reply With Quote
Unread Sep 4th, 2003, 10:50 PM   #16 (permalink)
Regular Forum Addict
 
Nolnova's Avatar
 
Join Date: Nov 2002
Location: Canada
Posts: 483
Default Sweetness

TY Squawks. It's my pleasure to contribute to your skillset.

Your point is well made too. It is common to position monitors at head-level or above, and speaker stands, elevated desk stands and wall mounts and brackets are the industry norm. When pressed into using just one set of monitors, I prefer wall mounted, angled down and inward at the mixing desk.

The guidelines on this topic are easy to follow and flexible, and knowing your production style is the only wildcard, so it's subjectivity that inspires the set-up one chooses. Stretching your legs or pacing require a little more distance between you and your monitors (which I do a lot,) while sitting in a catatonic state for hours on end (which I do even more,) means that the speakers can be right up front.

What is more important, once you've garnered the sweet-spot, is assuring that the monitors you're using are of high enough quality to reproduce the audio properly.

A few of the factors that define the quality of the speaker are; the weight/strength of the magnet (heavier is better for response), the weight and construction of the box (thick and heavy again, 'ensures cleanliness), proper insulation and breathability (needed to reduce Q ), ohms (8 only), frequency range (manufacturers are liers), and to a lesser degree, things like; cone and tweeter material and size, diaphragm material (I think this is a big one, but others may not,) and, of course, wattage.

I can suggest some brands that have made the grade in professional studios I've worked in, and out of curiosity, I'd be interested in knowing who here uses what...

Alesis M1
Dynaco A series (love 'em.)
Yamaha MS-10 (use 'em.)
Both M-Audio and Roland have made monitor-speakers of notable fidelity.

One last note, to ensure your speaker isn't being stiffed of power, using a heavy-gauge speaker wire is always recommended. Furthermore, 8 ohms is the only resistance pro quality monitors should use. 4 ohm speakers, while delivering more actual juice to the magnet, colour the sound-image and fail to delineate detail.
__________________
"That it is a little thing merely to be a particular existing human being is not to be denied; but for this very reason it requires considerable resignation not to make light of it." -Kierkegaard.
Nolnova is offline   Reply With Quote
Unread Sep 26th, 2003, 10:13 AM   #17 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default

Welcome back to the world of analog recording. Now it's time for the goodness of dynamics processors. They vary from the time based processors by messing with the dynamic range of audio. There are five types of dynamics processors, these are compressors, limiters, gates, expanders and the ever famous equalizer. Now let's get some details on what they do, and
why we use them in situations.

Equalizers
Let's start off with the EQ. It's the most common, and everybody knows what they are. Even basic home stereos have them. They pretty much boost or cut certain frequencies depending on how they're built. You have many types of EQs, stepped, parametric, graphic and filters.

Stepped: These have two knobs. One is the frequency select and the other is your boost or cut. These are very limited in what you can do due to the fact that you can only select certain frequencies.

Parametric: Very similar to stepped EQs in the fact that they have two knobs per band. The only difference is you can sweep the frequencies. These are beautiful because you can select many different frequencies giving you better control than a stepped EQ.

Graphic: These are easy to distinguish, they are a series of
sliders, just like many home stereo units. These are pretty acurate and can get a good mix from them. The downside is how much tweaking to each fader you have to do. A 1/3rd octave has 32 sliders or a 32 band eq, 1/6 octave is 64. They create a good bit of noise, too much electronics and it's really too much EQ for the production studio.

Filters: These do cutting only, they're very basic. You have three types, low pass, high pass and band pass. A low roll off filter cuts the lows from a certain frequency, and high does the same for the highs. A band pass is a combination of low and high pass filters.

Filters are not active, while the other types of EQing are.

Also there's what's known as a notch filter, it's not as well known as the high and low pass. It's used to remove certain frequencies. Very narrow bandwidth on the range.

Before I get out of the EQ session I should explain what Q is. It's very useful because many EQs allow you to adjust the Q value when modifying the frequencies. Here's the equation:

Q=fc/bw

It's not that hard to understand. FC is the center frequency and BW is the bandwidth, or the distance between -3db on the left and right side of the FC, it's how wide the curve is.

Next post, compressors and limiters.

-maz
maz is offline   Reply With Quote
Unread Sep 26th, 2003, 01:43 PM   #18 (permalink)
maz
"Banned Useless Poster"
(Forums Moderator)
 
maz's Avatar
 
Join Date: Jul 2001
Location: San Francisco
Posts: 5,632
Send a message via ICQ to maz Send a message via AIM to maz Send a message via MSN to maz Send a message via Yahoo to maz Send a message via Skype™ to maz
Default Compressors & Limiters

Some terms you should understand before I describe what they do fully.

Input gain: This is the control that tells how much input is going to go into the compressor.

Threshold: This is where you want the compressor/limiter to start operating. So if you select -10dB signal below that will not be compressed, while signal above it will be compressed.

Output gain: This control determines the amount of signal that will leave the processor. Used to boost the reduced dynamic signal to a range where it best matches the level of a medium, or to be better heard in a mix.

Ratio: This determines the "slope" of the input-to-output gain ration. Pretty much determining the increase of input signal that's needed to cause a 1dB increase for the output signal. example: 4:1 ration means for 1dB increase at the output you would need 4dB at the input, so an 8dB increase in input would give you a 2dB increase at output etc.

Attack: It's set in milliseconds, determins how fast or slow the processor will turn down signals that exceed the threshold.

Release: Similar to attack, but how fast the processor will restor a signal to it's original dynamic level once it's fallen below the threshold.

It's important to know that fast attacks and releases on sustained sounds can cause a pumping or breathing effect by affecting the dynamic ratio too fast. For those you should use longer attacks and releases. If you're compressing a transient sound, like a high hat, then faster usually works better.

The reason that I included these two dynamic processors in the same post is because they are really the same thing. Once you go above an 8:1 ratio on a compressor it will become a limiter.

Compressors compress the dynamic range of a signal allowing you to boost it higher in the mix. It helps bring things out, makes the sound more full when in reality they're not. It works because you compress the higher portions of the dynamic range closer to the lower portions. Then when you amplify the signal you now have more of the signal around the same area, giving a more full sound, and therefore a louder sound, by keeping the higher and lower signals right around the same.

When you increase your ratio abovev 8:1 on a compressor it becomes a limiter. It's going to take a very hot signal in order for it to even work. Very useful to make sure things that are peaking really hot don't do that in the final mix down. With a large ratio like this you're not really going to compress much of the signal and change the dynamic range all that much, therefore it won't keep the signal in roughly the same area.

You usually have a fast attack and release when limiting as you're more than likely looking for transient sounds. It's also very useful to keep the signal from audibly pumping.

-maz
maz is offline   Reply With Quote
Unread Sep 26th, 2003, 02:24 PM   #19 (permalink)
DI Extreme Addict
 
Join Date: Aug 2002
Location: Claremont, NH
Posts: 2,861
Default

Quote:
Originally posted by maz
Before I get out of the EQ session I should explain what Q is. It's very useful because many EQs allow you to adjust the Q value when modifying the frequencies. Here's the equation:

Q=fc/bw

It's not that hard to understand. FC is the center frequency and BW is the bandwidth, or the distance between -3db on the left and right side of the FC, it's how wide the curve is.
Some more info that might help people:

Q is also known as resonance. Basically, the more resonance, the louder the frequencies at the cutoff. That isn't quite the whole story, however, as it also has a side effect of boosting the frequencies that aren't cut off (at least, this is the standard way resonance works... it might not be that way in all filters).

On a low pass filter, resonance gives a kind of honking sound when the cutoff is low, and a more bleepy sound when the cutoff is high. This gives that classic acid sound we all love...
__________________
-brash
brash is offline   Reply With Quote
Unread Sep 26th, 2003, 02:34 PM   #20 (permalink)
DI Extreme Addict
 
Join Date: Aug 2002
Location: Claremont, NH
Posts: 2,861
Default

A word about compression:

There are two common misconceptions, as far as I can tell:

1. A compressor makes things louder. Often a typical response to the question, "How do I make my kick louder/punchier/bigger/etc." is, "Compress it." While the one answering might understand compression, it ends up giving the one asking the question the wrong idea.

Compression actually makes sounds quieter. As Maz said (I'm paraphrasing here in layman's terms), it makes the loud sounds quieter. You can then take the compress signal and raise its overall volume, and you end up with the same sound, except the quieter sounds have been boosted (the louder sounds are just brought back up to where they were before). This makes it sound louder overall.

Because it is so common to raise the volume after compressing, it usually isn't even mentioned as an extra step.


2. Compression makes loud sounds quieter, and quiet sounds louder. For some reason I read this a lot in explanations of compression. This is just wrong. It only makes the loud sounds quieter.
__________________
-brash
brash is offline   Reply With Quote
Reply



Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


All times are GMT -5. The time now is 08:20 AM.


Powered by vBulletin® Version 3.6.8
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Search Engine Optimization by vBSEO 3.2.0
Copyright © 2010 Digitally Imported, Inc.