Principles of Reverb

From Audacity Wiki
Jump to: navigation, search
Peter 3Sep12:
  • Material here is transferred deirectly from the old pages in the Manual (now deleted).
  • The page title is purely a working-title. I welcome suggestions for a revised title prior to publication.

Reverb

A Reverb simulates the component of sound that results from reflections from surrounding walls or objects. It is in effect a room simulator. Some people think it's just a delay effect with some filters, but it's way more complex than that. Reverb effects (software plug-ins or external hardware units) provide an interface to their changeable parameters that need some explaining. Let's look at a simple room first.

Basic Simulation of a Room

Our model is a simple room with four straight walls, a sound source and a listener. In Figure 1 the arrows stand for the path of traveling sound.

The listener hears the DIRECT signal first. The DIRECT signal is also referred to as the DRY part of the signal when using any effect. Most digital reverbs produce two parts: The Early Reflections and the Reverb component.

earlyreflections

I. Early Reflections

Diagram of early reflections
Figure 1

The first Early Reflection reaches the listener milliseconds after the direct signal does.

The path of the Early Reflections is longer. The difference in time between the arrival of the direct signal and the first Early Reflections is measured in milliseconds.

II. Reverb

Diagram of reverb reflections
Figure 2

The sound reflects off the walls and objects in the room, and in time individual reflections disappear and the Reverb develops.

Predelay
The time between the reception of the DIRECT signal by the listener and start of the Reverb portion of the effect is called Predelay. This is a parameter in many digital reverb effects, and it is expressed in milliseconds (ms).

Reverb Time
The time difference between switching off any sound generator and the level of the reverb resulting from that sound dropping by 60 dB is called RT60.

This is usually referred to as the Reverb Time. When anyone refers to the reverb time of a real room or that of a digital reverb, RT60 is what they're talking about.

Most digital reverbs feature this as a parameter.

Dry and Wet

A reverb usually gives you the choice of how much effects signal(also called WET part) and how much of the original signal(also called DRY part) you'd like to the plug-in to output.

A processor doesn't have this option by nature. The Equalizer and the Compressor are processors. To mix the original signal and the processed signal doesn't make sense for these processors. Their output is always 100% WET.

Not only do they output 100% WET signal, but nobody usually mixes the unprocessed signals with the processed either. They do so with Reverb and Delay signals. Processors are usually placed in an INSERT, which means they sit in the channel. The EQs on your stereo or soundcard mixer work like this. The signal passes through them.


parameters

Common parameters of a digital Reverb

Usually controlled via Room type and/or Room Size parameters.

The EARLY REFLECTIONS enable the human brain to quickly identify the room size. They are therefore the most critical part of a reverb effect, if a room simulation is the goal.

Parameter Range : 0 to x milliseconds

Time between Direct Signal and start of Reverb portion.

The more space between you and walls or other reflective objects, the greater the Predelay.

Parameter Range : 0.1 to x.x seconds

The greater the enclosed(!) volume of the space you're in, the longer the reverb time. This is of course true only to a certain extent. The larger the room or chamber, the louder your source signal needs to be to produce any significant level of reverb tail. A good way to demonstrate this is to visit some huge cathedral and talk quietly. Then yell a bit and you'll actually hear some significantly loud reverb. The RT60 value is measured with a noise impulse, but the level at which that impulse was sent out should be known too.

Many reverb effects let you set different reverb times for different frequency bands. Most reverb units on PC and Mac architectures provide two bands, so you get to set the reverb time for each frequency band individually.

Usually a Highshelf and/or a Lowshelf EQ with variable Frequency.

These EQs are applied continuously to an evolving reverb of a piece of audio. Over time, as the reverb fades, so do certain frequencies, either in the lower or higher frequency range. The more less-reflective surfaces the sound bounces around from, the more the reverb is dampened the higher frequencies.

Freeverb features a Damping parameter that dampens the upper frequencies.


Usually a Highshelf and/or Lowshelf EQ with variable Frequency.

These EQs are only applied to the output of the reverb or perhaps even just the reverb portion of the Reverb effect. Some reverbs have Highshelf and Lowshelf EQs for the Early Reflections as well.

Freeverb has both a Hicut and Lowcut parameter.

Here's a screenshot taken from the reverb plug-in Renaissance Reverberator by Waves.com

Screenshot of Renaissance Reverberator
reproduced with permission

Here we have the Highshelf and Lowshelf EQs for Damping(left) and for the final output(right). The upper middle display depicts the Early Reflections placement and level, as well as the Reverb portion start and fading. Above this display is the Reverb Type selector. This changes the placement and level of the Early Reflections. The Size slider has influence on the Early Reflections as well.

The surface of a wall as well as different shapes and sizes of objects in a room influence the way sound reflects, bends or spreads from off them.

Blank concrete will reflect sound quite well across the frequency spectrum. A large thick curtain or padded walls will reflect sound poorly. A large club filled with people sounds completely different from an empty one. People, comfy chairs, curtains, carpets, beds and many more objects reflect only part of the sound that hits them.

A Virtual Room with Reverbs

An example by the numbers

As an example I'd like to show you how to create a reverb for small club with a band.

Let's see what we have to do with parameters in the reverb fx in sample editors. First we determine some basics about the room we want to simulate.

Our room (the club) is about eight meters on all sides (for simplicity) and three meters high. The band will be playing no more than four meters away from us. The room volume will be about 200 cubic meters.

This means that the DIRECT SOUND will get to you within about 11 ms (milliseconds). This is the amount of time we'll need to take off the other values.

The first EARLY REFLECTIONS (off the left and right walls) travel around nine meters before reaching you, so they'll get to your ears about 15 ms later! (9 meters at 343 meters/sec -> 26 ms minus the 11 ms the DIRECT SOUND takes to travel to your ears).

I'm not taking the floor into account because of the obstacles, such as your own legs and a table, that prevent reflections of sound from the band from reaching you. It is also probable that your sitting close to a wall and EARLY REFLECTION time may drop considerably. Experiment and listen very carefully.

The reason I'm deducting 11 ms from the 26 ms is that in a sample editor we already have the DIRECT SOUND. In Cool Edit an EARLY REFLECTIONS parameter doesn't exist. In Soundforge however there are a few ER times to choose from. We choose whatever is closest to 15 ms. Freeverb unfortunately does not have an Early Reflections feature.

Please note that the timing of the EARLY REFLECTIONS are defined by the rooms characteristics such as shape and size. The sound color of the EARLY REFLECTIONS is defined by the surfaces in the room.These are parameters that are available to professional reverbs only (hard- or software), so don't worry about not finding that stuff in great detail in either Soundforge's, CoolEditPro's or plug-ins of other editors.

The PRE-DELAY and REVERB TIME depend on more than just room size. Wall and floor material, as well as anything else that reflects, disperses and dampens sound will influence the reverb in quite a few ways.

Our club has lots of wood on it's walls and floor and people are sitting at tables and the air's probably filled with a little smoke (except in California :).

The REVERB TIME, a parameter found in almost all plug-ins, should be set to no more than one second. In small, crowded rooms where little REVERB occurs it is quite important to set the EARLY REFLECTIONS right. Experiment with the reverb time keep it lower than 1.2 seconds and above 0.7 seconds. Freeverb puts all this information in to the SIZE parameter.

The PRE-DELAY can be set to 30 ms or more. Larger rooms will develop REVERB later since the reflecting sound will travel a while before the reflections become hard to identify but it also depends on the reflecting materials. Freeverb once again doesn't know any Predelays, but since you're reading this to learn how things work, it's still worth spelling out.

Both Cool Edit and Soundforge have parameters referring to the DENSITY of the REVERB part of the sound. Materials such as polished marble will reflect the sound very well and disperse the sound very little. Wood disperses the sound a lot and a carpet practically swallows a large chunk of it. For our Club we need to set the density high since our material disperses the sound a lot(humans,wood and obstacles). No parameter like that in Freeverb and yet it sounds great. It's easy to use as well, so I don't really mind when using it.

Higher frequencies are swallowed a lot faster than low frequencies. Many reverb algorithms refer to this in some form or another. Cool Edit has the 'High Frequency Absorption Time' as a parameter for this. Sound Forge gives you a choice, whether to do this at all for both low and high frequencies and even lets you set them but doesn't provide any time parameters. Professional Reverb hardware such as Lexicon PCM 70,80&90 let you you control frequency and time of this, for low and high frequencies. Freeverb has the DAMPEN parameter for this.

This can effectively add to room's liveliness. A room full of people will absorb high frequencies a lot faster than an empty room, in which only the materials would make the difference, so in Cool Edit we'll set the 'High Frequency Absorption Time' very low, since we're in a crowded club. For Soundforge activate the 'Attenuate high frequencies above' box and set the frequency parameter to around 2000-2500 Hz. Experiment a bit. The higher this frequency and the longer the Absorption Time parameter in Cool Edit, the emptier and and colder your room will seem. In small rooms this can cause the room to sound very metallic. Sewer pipes and oil pipelines e.g.

The Reverb EQs are the final processing stage and can change the sound dramatically. These Hishelf and Lowshelf EQs are found in almost ANY Reverb plug-in and hardware unit. Freeverb has both in form of Hipass and Lowpass filters. These are not EQs. The parameters change the target frequencies of the filters. Again, listen.

The last thing you have to do is set the balance between the WET signal and your original DRY signal. Here you can do as you please but remember that only very smoothed hard surfaces in big cavernous space create very loud reverbs. A church might do that if there aren't a big number of people and wooden benches in there, but a club doesn't :). Remember that in Audacity it is smarter to DUPLICATE the part you wish to add reverb to, and run the effect over that duplicate. The volume of that reverbed material is easier to control, so set the WET parameter in Freeverb to 0 dB (-&rt;100%) and the DRY parameter to -endless(->0% left most slider position).

So when you create your next 'soon-to-be-the-hit' song or want to put things in perspective, be sure to check the possibilities of adding dimension and scope to your material through clever use of reverbs. Put your snare to the front, the piano a little to the back, and the sax just a little in front of the drums. Experiment. With these basics you'll be able to get better results in less time.

Btw, Snare always gets reverb. How much and what kind up to you, but it's generally good thing. It's also important to decide whether you're trying to simulate a real live band or doing a multitrack sound, which boasts all kinds of nice ideas, like doubling(another take of the same thing on top of the first take), extreme panning and in general NOT reality :).


Dry versus Wet

Have you heard of those silent rooms, where they test speakers and microphones ? They're great if you don't happen to be inside of one. After a few minutes you hear the blood in your ears roar like an ocean. You hear only your inner voice when you speak. The room swallows almost everything else. Most people can't stand being in a room like this for more than 10-15 minutes. I am told that this is very different to wearing ear plugs. Those only attenuate sound from the outside. A silent room kills it completely.

This room is horrible to listen to, because it basically features no room reverberation, i.e. it doesn't reflect anything at all. It doesn't reveal any information about its size or surface composition.

You may be surprised to know that many vocals are actually recorded in similar environments. Vocal booths are little rooms, in which the walls reflect sound as little as possible. During mixing, digital reverbs are used to create a virtual space.


Digital Reverbs

Audacity for Windows and MacOS9&X includes the VST plug-in Freeverb, which is one fine reverb, and ideally suited to take your first steps with digital reverbs. It features simple controls that require little explanation.

For details about Freeverb, check out the effects menu page. For information on how digital reverbs work and what reverb is, check out this page of this manual.


Keeping effects on seperate tracks

You shouldn't use reverbs on original audio tracks. First duplicate the audio and use the reverb on that duplicate. The reverb should be set to output 100% WET (i.e. 100% effect signal and 0% of the original signal). That way, you can control how much reverb should make it in to the final mix, right up until the last minute.


Basic Calculus of Acoustics

This is so terribly useful for so many things. You can use this stuff to calculate just how bad parallel walls are for sound, because of Standing Waves. What's that ? Coming attractions include exciting info on how to improve the acoustics of your room and what standing waves, bass traps and reflecting will do for your listening karma.

You may enjoy finding out that a $1000 subwoofer won't work in a small 3x3 meter room, 'cause the big energy bass action will happen beyond your walls. After all, the first energy peak happens at 1/4 of wavelength. Check it for yourself and get smart the next time you buy those 'gotta get buried with/under/in them' speakers.

This all happens in plain fresh air:

The speed of sound is 343 meters per second at 20 degrees Celsius

+/- 0.6 m/s for every degree Celsius more/less ( -> 340 m/s at 15 degrees Celcius)

speed of sound = wave length * frequency
 
time [milliseconds] = time[ms] for one meter * distance [meters]

Examples (at 20 degrees Celsius > c = 343 m/s):

distance [meters] time [milliseconds]
0.34 1
3.43 10
6 18
10 29
50 188



Personal tools

Donate securely by PayPal, using your credit card or PayPal account!