Convolution is one of the most musically useful, creative, and fascinating techniques to hit the desktop in a long time. It allows you to “merge” two audio files to produce effects that range from controlled reverbs to unpredictable sonic hybrids. The technology has fantastic potential for the electronic musician and can serve a variety of purposes.
Convolution is most commonly used to model real acoustic spaces, allowing any sound to take on the characteristics of those spaces. That approach, often called sampling reverb, is the focus of several convolution plug-ins, including Waves IR-1, Tascam GigaPulse, Audio Ease Altiverb, and Emagic Space Designer. These products take plug-in reverb to new heights of realism and sound quality. Not only are they useful in music production, but their space-emulating features are also a boon in post-production environments.
But convolution also has more esoteric purposes, offering an infinite range of filtering options and allowing any two arbitrary audio files to interact in unusual ways. For sound designers and computer-music composers, that use of the technique has great potential and can provide unlimited variations of their audio material.
In this article, we''ll explore 13 different convolution programs. Our main focus will be on sampling reverbs, but we''ll also discuss using convolution software for sound design. We''ll look at tools that work as plug-ins, as well as those that are specific to particular host programs.
We''ll begin with an overview of the programs as a whole, then offer a more detailed, minireview of each. For a list of links to files that you can use in your own convolution experiments (and experimentation is a key to exploring convolution), see the sidebar “IR Libraries.” Before we start, we''ll offer a short explanation of convolution and what it can do. You''ll find in-depth coverage of the topic in John Duesenberry''s article “Convolution Number Nine” in the June, 1999 issue of EM, available online at the EM Web site (www.emusician.com).
Convolution is a signal-processing technique in which the spectra of two audio files are multiplied. Most commonly, one of the files, called the impulse response (IR), is short and represents the acoustic signature of a room or some other real space. The other file, often called the source, takes on the characteristics of the room when it is multiplied by the IR. In effect, it has the quality of having been recorded in that space.
IRs are created by sampling the space in question. Typically, a wide-spectrum sound, such as a balloon pop or swept sine wave, is recorded in the room being sampled to determine what impact the room will have on that sound (that is, what the response of the room will be). The reverberant qualities of the room can then be extracted from the recording through a process called deconvolution and applied to any new sound.
When creating impulses, you need to choose which kind of mic you''ll use, the sampling rate, and the number of channels (stereo IRs mixed to mono can result in comb filtering and phase cancellation), but those and other factors are beyond the scope of this article.
Many of the programs that we will examine come with their own library of IRs that were carefully sampled in all kinds of spaces, including cathedrals, concert halls, recording studios, bathrooms, living rooms, and even car interiors. IRs come in a variety of file formats—including proprietary formats that can''t be read by other programs—but the most common are simple WAV files. The majority of the programs we will be looking at can use (either directly or after conversion) WAV IRs, and you will find hundreds of such files on the Web.
You can make your own IRs without ever leaving home by sampling electronic devices such as reverb or multi-effects processors. And if you download IRs that are in a common audio format such as WAV, you can edit them, adding fades or EQ before applying them to your source material. As you will see, the possibilities are endless.
The programs we''ll be looking at fall into two broad categories: plug-ins and host-specific software. The plug-in group consists of Audio Ease Altiverb 4, Christian Knufinke SIR 1.008, DelayDots SpectrumWorx 1.1, Voxengo Pristine Space 1.1, and Waves IR-1 1.0. In the host-specific category we''ll look at the convolution features in Adobe Audition 1.5, BIAS Peak 4 (ImpulseVerb), Magix Samplitude 7.22 (Room Simulator), Sony Sound Forge 7 (Acoustic Mirror), Steinberg Nuendo 2.1 (Acoustic Stamp), and Tom Erbe''s freeware program SoundHack 0.892. We included one plug-in—Emagic Space Designer—in this category even though it only runs in Logic Pro 6.4 (and version 6.3 or higher of Logic Platinum, Logic Audio, and Logic Gold). We also included Tascam GigaPulse Pro 1.0, which runs both in GigaStudio 3 and as a VST plug-in.
Two other convolution reverbs were not released in time to be included in this roundup: Trillium Lane Labs TL Space plug-in (see the sidebar “Convolving in TDM”), which will offer native and TDM versions for both Mac and Windows; and Red State Sound''s RevolVerb Lite (http://revolverb.hostrocket.com/), a Windows product that was still in beta at press time.
Prosoniq''s Rayverb (see the sidebar “A Ray of Difference”) is a sampling reverb that mainly uses technology other than convolution. As a result, we didn''t feel that the program fit our criteria for this roundup.
Additionally, we chose not to cover some tools due to space considerations, such as Sounds Logical WaveWarp, which is aimed more toward DSP engineering applications; and a handful of Linux convolution tools. We saved the topic of hardware convolution processors for another article.
We considered making direct comparisons of all the programs by testing them with the same impulses and the same source sounds, but after noting the vast range of differences in the implementation of convolution, we decided against that approach. Instead, we''ll explore a variety of IRs and source files and document our results both here and at the EM Web site (see Web Clips 1 through 10).
The programs we reviewed cover a wide range of feature sets, user interfaces, and prices. Rather than doing head-to-head comparisons, which in many cases would be comparing apples to oranges, we opted for a broad overview, with the aim of letting you know what the strengths and weaknesses are of each of the programs covered. Armed with that information, you can choose the software that best meets your needs.
Despite their differences, there are some common features to all the programs. The most obvious one—and the reason they were chosen for this story in the first place—is their ability to apply convolution processing. As you might expect, they don''t all do it the same way.
The plug-ins—Altiverb, Space Designer, Pristine Space, GigaPulse Pro, and IR-1—perform their processing nondestructively in real time, whereas ImpulseVerb, Acoustic Mirror, and SoundHack write the convolution results to an audio file. The convolution features in Audition and Room Simulator can be run as effects inserts or as destructive processes. Many of the programs have a preview feature that allows you to check out the effect of an IR on a file before writing it to disk.
Share and Share Alike
Another feature that all the applications share is the ability to use third-party IRs, although some of the programs make the importing process easier than others. The ability to import IRs not only allows you to access a greater range of sampled acoustic spaces for reverb purposes, but it also means that you can modify any source with any type of audio file. Imagine using your singing voice as a source and convolving it with a horse''s whinny, a wailing lead guitar, or the sound of breaking glass.
You''ll find major differences in the programs when it comes to the file formats that they can open as IRs. Space Designer and Acoustic Stamp are the most versatile, handling AIFF, Sound Designer II (SDII), and WAV files; SpectrumWorx, Pristine Space, and SIR restrict you to WAV files; and IR-1, Acoustic Mirror, and Room Simulator support WAV and their own proprietary formats. Altiverb requires that you use its own file standard or import noninterleaved SDII files, and Audition will only open files in its own IMP format (although you can make IMP files from any type of file that Audition will open). GigaPulse can read raw WAV files, but requires a few steps to convert them into its own Impulse Set format. According to Tascam, an upcoming release of GigaPulse will offer drag-and-drop conversion of individual WAV files into Impulse Sets.
The maximum length of an IR supported by each program varies greatly. Room Simulator has no maximum, while a number of the others are approximately five to six seconds at a 44.1 kHz sample rate. Length may or may not be an issue, depending on whether you''re doing straight reverb (where short IRs are the norm) or sound-design work (where longer IRs could be useful).
The software examined here gives you mono-to-stereo and stereo-to-stereo operation. Several programs offer other options, such as Altiverb''s mono-to-quadraphonic and stereo-to-quadraphonic channel configurations and GigaPulse''s mono-to-seven-channel setup. Pristine Space scores big here, with eight discrete channels each of which can hold a different IR. Not only do the channel possibilities vary from program to program, but sometimes you also get different capabilities with the same plug-in, depending on the host software.
The programs all offer help in various formats. Some offer documentation in PDF or HTML format only, while others merely provide contextual help. Altiverb, Audition, Space Designer, and Room Simulator include both printed and PDF manuals.
Most of the programs have functional and good-looking user interfaces. Many show you photos of the space from which the IR was sampled. Altiverb can display photos, scrolling photos, and mic-position diagrams of the venues from its library. It even lets you open a larger window for viewing purposes. Both Acoustic Mirror and GigaPulse can associate an image with an IR file.
Space Designer and IR-1 offer slick-looking controls in their main windows that let you drag control points to change envelope values. SpectrumWorx uses a modular approach, providing 16 slots in which you can insert not only its convolution module, but also any of the program''s 43 other spectrum-processing modules. Its controls are small, but they still do the job. Room Simulator and SIR have a graphic display on which you can draw the frequency response of your IR. Pristine Space offers a single display that toggles between its six main functions.
One area in which there are big differences between the convolution programs is parameter control. The sampling-reverb plug-ins offer high and low damping, wet-dry controls, and often predelay. IR-1 gives you controls that come the closest to resembling those on a traditional reverb unit. GigaPulse includes a number of familiar reverb parameters, and its large Placement Selection window lets you interactively adjust mic position and the placement of the listener. Additional GigaPulse parameters (Perspective, for example) include mic modeling, and others can take your impulses into unknown territory.
Quality and Quantity
Overall, we were impressed with the sound quality of the convolution reverbs. In most cases, they more than held their own against conventional reverb plug-ins. By virtue of their IR libraries, many offered a much wider range of sound possibilities. That said, standard reverb plug-ins tend to offer more conventional parameter editing than the convolution tools in our roundup (with the possible exception of IR-1).
As to whether you''ll want to replace your hardware reverbs with convolution plug-ins, you''re going to have to decide that for yourself. In our admittedly subjective and unscientific comparisons, we found our hardware reverbs to be slightly warmer than the convolution plug-ins, but they were similar.
Of course, convolution software excels when it comes to number of sounds, ability to save and recall, and flexibility. Moreover, the ability to use any sound as an IR gives these programs unlimited sonic potential (see the sidebar “Five Tips for Better Convolution” for suggestions about successful convolving).
Keep in mind that convolution programs are processor intensive, so if you don''t have a relatively fast CPU, you''re going to be in trouble. In fact, many of these plug-ins and programs won''t even run on outdated computers. If you''re unsure about whether your computer can effectively run any of these programs, check the company''s Web site for specific system requirements (see the sidebar “Manufacturer Contacts”).
We''ve organized our program summaries by type, splitting them into two categories: generic plug-ins (VST, AU, MAS, RTAS, DirectX) and host-specific tools, alphabetized within each group.
You''ll notice that we''ve included retail prices for each of the individual software write-ups. When comparing prices, bear in mind that in a number of cases, the convolution feature is integrated within a larger host-software program. Therefore, the only price we can provide is for the entire program.
The following plug-ins run within any compatible host program. As with any plug-in, the performance you''ll get is a function of the host and your computing platform.
Audio Ease Altiverb 4.0, $495 native, $795 HTDM. On the market longer than any other convolution reverb software, Audio Ease Altiverb combines great sound, ease of use, excellent IRs, and a simple-but-effective set of parameter controls. Depending on how your sequencer is configured, it can run in a number of channel configurations, including mono-to-mono, mono-to-stereo, stereo-to-stereo, mono-to-quadraphonic, and stereo-to-quadraphonic. The latter two are designed for use in a surround-sound mix.
Altiverb gives you separate wet and dry knobs; a button to mute the direct sound completely, which is handy when using Altiverb as a bus effect; a predelay control (up to 200 ms); and high- and low-frequency EQ knobs (see Fig. 1). A large knob controls reverb time, but it only allows you to decrease it, not increase it. (Audio Ease reports that Altiverb 5 will include more editable parameters.)
Altiverb supports snapshot automation, offering ten slots for storing various IRs. You can automate them with your sequencer, which makes it possible to change IRs mid song. However, loading an IR is processor intensive, and it takes about a quarter-second or so on a dual G5/2 GHz Mac. As a result, setting an automated patch change to happen on a beat is problematic, because there''s sure to be an audible pause before it loads. The snapshot feature will work well if you can set up the IR change to happen during a pause or a rest in the track. Altiverb also responds to automation of its various knobs just as any other plug-in would. Be aware that the AU version currently doesn''t support any of the automation features, including the snapshots.
You will also find buttons that switch Altiverb between high- and no-latency settings (in the RTAS and HTDM versions, the choice is between high- and low-latency settings). The high-latency setting reduces the processor load and works well when using Altiverb as an insert. If you use that setting when Altiverb is functioning as a bus effect, the latency can be quite pronounced.
Previous versions of Altiverb were known for being processor intensive, but Altiverb 4 is substantially more efficient. When comparing its CPU usage to IR-1 and Space Designer—using OS X''s Activity Monitor—Altiverb consistently used less CPU than IR-1 (with both set to their full settings) but more CPU than Space Designer. In its high-latency/low-processor load setting, Altiverb was about equal to IR-1, with the latter set to its most efficient setting.
Altiverb comes with an excellent library, which includes IRs from studios (the Cello Studios echo-chamber IRs were particularly impressive), stadiums, stairways, concert halls, cathedrals, and classic reverb units. A number of the IRs in the library offer you choices of multiple mic positions within the same venue.
Altiverb is not so user friendly when it comes to importing IRs. It requires that you use noninterleaved SDII files rather than WAV files. Other plug-ins let you import IRs directly through their interface while the plug-in is running. But with Altiverb, you have to put them into a specific folder and then restart the host program.
Altiverb does encourage users to come up with their own IRs, and Altiverb IR Preprocessor 2.1, a utility program that helps you make your own, comes bundled with it. You also get a sweep file (used in the IR creation process) and a number of how-to PDFs.
Pros: Easy to use. Great-sounding IRs. Useful built-in help feature. Excellent display of information and venue pictures. Help provided for making your own IRs.
Cons: Automation not supported in AU version. Audible pause when switching between Snapshots. Latency can be a problem when using low CPU-load settings. Loading other audio files as IRs is cumbersome.
Compatibility: Mac OS 9—VST, MAS, and RTAS; Mac OS X 10.2 or greater—AU, VST, MAS, and RTAS.
Christian Knufinke SIR 1.008, free. Although SIR is the only freeware plug-in in this roundup, its tool set is both powerful and unique in several respects. The program can load fairly long IR files—it tops out at a length of around 30 seconds—and a graphic display of your IR provides an intuitive aid as you edit.
Working in SIR''s single screen, you''ll access several parameters that can dramatically change the sound of your audio source (see Fig. 2). The Length function, for example, stretches and compresses the IR from 50 to 100 percent, and the Stretch function raises or lowers the IR''s sample rate. You can also taper the onset of the IR using a combination of the Attack and Time sliders.
Other features more closely resemble traditional reverb software. The Predelay feature alters the time between the onset of a sound and the first reflections; like the other controls, it updates almost instantaneously even while a sound is playing. Two of SIR''s sliders affect stereo width: Stereo In works on the source signal, and Stereo IR works on the impulse. The Reverse button does what its name implies. The large FFT EQ display at the bottom of the screen allows you to manually tweak the frequency response of the IR using as many as eight filter points.
SIR''s CPU Consumption option automatically adjusts the amount of CPU power the program uses so that you can continue to work with other tasks. The automatic Gain Compensation features helps keep your impulses from clipping. Other handy features include a list of all the files in the directory from which you first load a file, separate wet and dry level sliders with on and off buttons for each, and buttons that apply –6 dB cut or +12 dB boost to dry and wet signals, respectively.
The sound of the plug-in is excellent, and its feature set gives you lots with which to work. The only downside is a hefty fixed latency of 8,960 samples, although you can find automatic latency-compensation options in some programs (such as Nuendo) or adjust your tracks manually as needed.
SIR has a huge following among members of the convolution community and is in active development. A new version claims to provide zero latency. SIR makes a great introduction to convolution, both for reverb and for other purposes. At this price, how can you go wrong?
Pros: Free. Intuitive parameters.
Cons: High latency.
DelayDots SpectrumWorx 1.1, $99. Unlike the other plug-in convolution tools in this roundup, DelayDots SpectrumWorx offers convolution as part of an extensive toolkit that includes several dozen spectral-mangling tools (morphers-blenders, combiners, pitch shifters, and the like). In fact, this kit is one of the most extensive spectrum processors we''ve seen. The plug-in works as an effects insert, and once you load an IR (WAV format only), you''re live and ready to tweak.
The SpectrumWorx interface resembles a hardware rack. Although there are not a lot of sliders or knobs to work with, the available controls are manageable and fairly easy to adjust (see Fig. 3). You can also use MIDI Control Change messages to modify any of the parameters in real time.
There are four settings for modifying the convolution process, and it''s clear that SpectrumWorx has more than just reverb on its mind. First is the thd setting, which can be used to gate the IR signal or to function as a blend factor, depending upon which convolution module you are using. The values range from 0 dB to –99 dB in increments that start at 0.1 dB and rise to several dB as the value increases.
Next are the lband and rband controls, which are found in all of the program''s modules and determine to which range of frequencies the processing will be applied (lband controls the currently loaded file and rband controls the IR). Although the parameters of these controls are not completely intuitive (the range is 0 to 100 percent), they work effectively.
The final convolution control is op, and it offers different types of convolution processing, including conv, which multiplies the respective files'' magnitudes (powers) and adds their phases; and rconv, which does just the opposite. There''s also cmagn, which multiplies only the magnitudes of the two signals; and separt, which performs convolution on the magnitudes and phases of the two signals independently. Another option is slowco, which (according to the manual) is an “experimental mix of the above.” It''s pretty hard to predict what each type will result in, but if you''re open to experimentation, you''ll have a lot with which to work.
There are global in, out, and wet-dry mix controls along with dedicated buttons to access presets and IR files. Like some of the other programs, SpectrumWorx also has its own browser that lists all of the files in the drive in which the currently loaded file resides and from which you can quickly pick a new file for use as an IR. If you want, you can use the program''s Use Sampler feature to load an IR file from your system and tweak it with the program''s numerous other processes in advance of using it for convolution (for example, vocoded IRs).
Right-clicking on the main info window displays a drop-down menu from which a number of additional options can be set. Here you can adjust the window size for the processes that rely on spectral analysis (larger windows give more frequency resolution but less temporal accuracy), pick a window shape (windowing is used to help the analysis routine better determine the frequency of any given chunk of samples), and even choose from several skins. You can also select whether the processes will use the two channels of the file currently loaded in your host software (Sidechain mode) or whether the IR should be a file you load from your drive (External mode).
SpectrumWorx comes with a bunch of presets (you can''t tell which ones load the convolution module simply by looking at them), and you can also save your own. The PDF manual includes an introduction to audio processing in the spectral domain (the phase vocoder in particular) that can help focus your experiments with the different modules. According to the manufacturer, a new release, with an updated graphical user interface (GUI), new modules, assignable LFOs, and more, will be available by the time you read this.
Pros: Unusual spectral-processing options. Modular configuration.
Cons: Nonintuitive parameters for convolution.
Voxengo Pristine Space 1.1, $139. Pristine Space offers a whopping eight independent channels of convolution processing. You can configure the channels as eight mono or four stereo pairs. You could also use all eight channels in series to process a single audio source. This might be useful, for example, if you wanted to apply a variety of mic positions to the same file.
The program uses a simple and efficient patching system to assign any of the eight potential audio inputs to as many as eight outputs. Pressing the “?” icon at the top of the screen (see Fig. 4) takes you to a setup page where you can configure the number of ins and outs as well as tweak other aspects of the program for best performance on your system (such as assigning a latency value). The setup area provides access to the help file, which is also available as a program-group shortcut.
In the middle of the screen you will find color-coded buttons representing the ins and outs that you''ve enabled. Click on a button, and you''ll see a list from which you can choose the desired destination for a channel. Each of the eight slots has its own wet- and dry-mix parameter as well as an independent quality-control setting (low or maximum).
At the top of Pristine Space''s screen is a window in which you can draw envelopes (linear only) to control six of the program''s main parameters: Volume, Stereo Width, Stereo Pan, Lo-Pass, Hi-Pass, and Equalizer. You toggle the envelope window to access the parameter you want by clicking on the appropriate button underneath the display. The program shows a waveform of the currently loaded impulse, which helps you to accurately fine-tune envelope times. When you''ve created the envelope that you need, you can copy it to the same parameter of any of the other slots.
Four knobs—Offset, Length, Delay, and Gain—provide additional control over the IR file. You can fine-tune settings by clicking and dragging the right mouse button, but you can''t type in an exact value.
There''s an auto-gain control, which is especially useful if you are loading IRs that produce different output gain levels, as well as a Reverse feature for reversing the IR file. To cut down on the amount of tweaking you might need to do with multiple impulses loaded, you can link the Offset and Length controls of one slot to those of one or more other slots.
Once you load a WAV file into the Impulse slot (according to the manufacturer, AIFF format will be supported in the next release), Pristine Space detects any other files that you might have in the same directory and lets you move among them by clicking on left- or right-pointing arrows. That''s just one of many nice touches that make Pristine Space useful for trying out a variety of impulses on the same source material.
If you go to the Voxengo Web page, make sure to take a look at Voxengo''s Impulse Modeler. Like a CAD program, Impulse Modeler allows you to design virtual spaces and to output IRs in WAV format that represent the signatures of those spaces.
Pros: Eight discrete channels. Flexible envelope control. Handy patching/routing system.
Cons: Can't type in values for parameter control.
Waves IR-1 1.0: $1,200 HTDM, $800 native. IR-1 is one of the most full-featured convolution plug-ins on the market, albeit the most expensive. It is the only cross-platform plug-in in this roundup, and it is compatible with virtually all of the major plug-in formats.
What sets IR-1 apart from the other convolution products is its array of editable reverb parameters, which give it more of a conventional reverb feel than its competitors (see Fig. 5). It offers parameters such as reverb time and reverb size (both of which can be increased or decreased), a control for density, and one for resonance. The predelay section is particularly flexible, offering time (up to 500 ms), gain, and on/off controls for the direct sound, early reflections, and the reverb tail. IR-1 gives you more predelay control and more predelay time than Altiverb, Space Designer, or GigaPulse.
You also get low- and high-frequency damping and a 4-band paragraphic EQ to tailor the frequency response of the reverb. The EQ controls are reminiscent of those in the company''s plug-in equalizers and offer the most options of any of the convolution reverbs in this article. Other features include a graphic control on the reverb/decay envelope, a wet/dry control, a reverse button, and output meters with output control sliders.
IR-1 sounds great, especially when used with one of the IRs from its included library, which is both large and meticulously recorded. There is a huge range of sampled spaces and devices, from famous concert halls (including the Sydney Opera House and the Ryman Auditorium in Nashville) to recording studios, car interiors, and classic reverb units.
There are also a number of options in terms of channels and processing usage, including mono-to-mono with one convolution; mono-to-stereo with two convolutions; stereo-to-stereo with two convolutions (Efficient Stereo); and stereo-to-stereo with four convolutions (Full Stereo). (According to Waves, a new version with surround-reverb capabilities is expected to ship sometime this fall.) To further save processing power, you can toggle IR-1 into Low-CPU mode using a switch in the upper-left-hand corner. Another option for cutting processor load is to shorten the length of the convolution using Direct Convolution Length Control.
But even with all the choices offered for lowering CPU usage, IR-1 can still be quite processor intensive. It''s not a big problem on a fast computer such as a Mac G5, but on slower machines, you''ll need to be careful not to overtax your CPU. Using IR-1 in Efficient Stereo mode and running it as a bus effect is a good way to get the most from the program without maxing out your computer.
Latency was never a problem with IR-1, at least not on the dual G5/2 GHz Mac on which it was tested. Even with the plug-in set to its most CPU-efficient settings, the latency remained at a virtually undetectable 11 ms (when working on 44.1 and 48 kHz files).
If you want to open up your own IRs, or those you have found from third parties, IR-1 lets you do so, provided that they are in WAV format. Unlike other programs, IR-1 doesn''t offer utilities for making your own IRs. The focus seems to be more on using IR-1 as a reverb processor, and for that purpose it is a formidable product. But don''t sell it short as a sound-design tool: its wealth of sonic options also make it well suited for that kind of work.
Pros: Full-featured reverb-parameter editing. Excellent IR library. Well-designed interface. Solid documentation. Cross platform. Wide range of formats. Low latency.
Cons: Pricey. CPU intensive.
Compatibility: Mac OS X 10.22 or greater—AU, AudioSuite, MAS, RTAS, VST; Windows (Service Pack 3), XP—AudioSuite, VST, DirectX.
The next group of convolution processors are part of specific host programs. Most are features of larger, multifaceted applications.
Adobe Audition 1.5, $299. The convolution feature in Adobe Audition 1.5 may not be as extensive as some of the others in this roundup, but it has unique aspects that you won''t find elsewhere. The tool is aimed at tweaking the original space in which the IR was measured by adding echoes to the IR file. There''s a lot of control over the echoes that get added, from picking the size of the Finite Impulse Response (FIR) filter that is used to generate them to adjusting the high- and low-frequency ranges in which echoes are allowed to build up. Although it''s not your average convolution processor, Audition can produce interesting results.
Audition''s convolution works in both Edit (destructive) and Multitrack (nondestructive) modes. Load a file into a track in Multi view, then drag the convolution icon from the Effects tree onto the audio track. A dialog box will open with all the settings displayed in a single screen (see Fig. 6). If you''ve got any IMP files on your drive (a few are included with the release), then you can open them. But you''ll probably end up making your own, especially if you want anything out of the ordinary.
To make an IR, you must first load the file that you want to sample, then highlight the range that will become the IR. Audition has a fairly low limit on IR duration (about five seconds), so keep it short at first. Once the range is selected, click on the Add Sel. button, and the IR will be loaded. By default, the amplitude of the IR is heavily attenuated (often reduced to only a fraction of the original), so you may have to enable and adjust the Normalized View switch to see the waveform in the display. After you adjust the Shift (for timing compensation) and Global Volume setting (if needed), you can save your file to disk for future use.
Unlike most of the other software, Audition allows you to make your own IRs from scratch. To do so, however, you''ll need to know a few things about room acoustics. You''ll have to create echoes one by one and choose their attenuation level individually. Although you can get nice comb-filtering effects without much effort (add a full-strength echo every millisecond for 10 ms or so), simulating a convincing reverb is a complex thing.
While using the Audition convolution plug-in, an error message appeared repeatedly that said there was a problem “in the envelope.” Adobe is currently looking into that problem. Regardless, if you''re an Audition owner, take a look at the convolution option and see how it can expand the program''s already sizeable audio arsenal.
Pros: Unique room-design feature.
Cons: Limited file-format support. Maximum five-second IR length. Nontraditional parameters.
Compatibility: Windows XP.
BIAS ImpulseVerb, $499. BIAS Peak, the premier 2-track editor on the Mac, has always offered a menu full of DSP features in addition to its editing functions. The program has had a convolution feature called Convolve since version 1.0. But the processing power of the Altivec velocity engine in the G4 and G5 made it possible for BIAS to include ImpulseVerb—a full-fledged convolution reverb—into Peak, starting with version 4.0.
Although ImpulseVerb doesn''t offer real-time processing like a plug-in would, it does provide a real-time preview function with a wet/dry control (see Fig. 7) that allows users to listen to the reverb and experiment with the mix level before writing the results to disk. (By using the Save As or Save a Copy As commands, you can apply ImpulseVerb while leaving your original file untouched). Be aware that the preview function uses a huge amount of CPU resources, much more than those of the real-time convolution plug-ins mentioned in this article. That said, during the testing for this review, the preview feature worked flawlessly on a relatively slow G4/733 MHz and a speedy G5/dual 2 GHz.
Using ImpulseVerb couldn''t be simpler (which is good, because Peak''s PDF manual devotes only about one page to it): you open an audio file, select ImpulseVerb from Peak''s DSP menu, and a window opens with a pull-down menu from which you can choose to use the clipboard''s contents as your IR, or pick from a library of over 100 included IRs. Although that number is small compared to what you get with most dedicated convolution-reverb plug-ins, it''s still a nice collection considering that ImpulseVerb is only one feature among many in Peak.
The IR collection is broken up into five categories: Acoustic Modeled Spaces (physically modeled rooms of various materials), Grand Spaces (cathedrals), Intimate Spaces (bedrooms, bathrooms, elevators, and hallways) Medium Spaces (ballrooms, theaters, churches, and even a quarry), and Misc (reverb units, synthetic IRs, and effects IRs). Overall, the IRs sound good.
Using a third-party IR, or any audio file as your IR, is easy. As long as Peak can open a file—and it opens many file types—you can use up to 5.9 seconds of it as your IR. Open the file, select and copy a portion of it, click back on your source file (to make it active), and then open the ImpulseVerb window and choose Clipboard.
ImpulseVerb doesn''t offer much in the way of parameter control, except for the ability to change the volume envelope, which opens in a separate window.
If you''re already a Peak owner and you haven''t checked out ImpulseVerb, do yourself a favor and try it. It produces some very impressive reverb and sound-design effects. Given that it''s not a real-time processor, it isn''t likely to become your main reverb. However, for sound designing and adding reverb to 2-track material, or if you''re a Peak owner and don''t want to spend the money for a separate convolution processor, it offers a lot.
Pros: Easy to use. Reads many different file types as IRs. Real-time preview function. Reverb sound on par with convolution plug-ins. Decent IR library.
Cons: Preview function is CPU intensive. Not much parameter control. Scanty coverage in the Peak manual.
Compatibility: Mac OS X 10.2 or greater.
Emagic Space Designer 1.1, $999. Included in the Emagic Logic Pro 6 bundle, Space Designer is a convolution plug-in that''s robust enough to compete with third-party plug-ins such as Altiverb and IR-1; however, it runs only in Logic. Space Designer can be opened in mono-to-mono, mono-to-stereo, or stereo-to-stereo configurations, and it includes a number of controls to shape and mold reverb sounds. Although it doesn''t have as many traditional reverb parameters as IR-1, Space Designer gives you plenty with which to work (see Fig. 8). There are separate sliders for direct sound and reverb level; a predelay control (up to 200 ms); low-shelving EQ (20 Hz–4 kHz ± 18 dB); and graphic control of the volume, density, and filter envelopes.
The filter offers lowpass (6 dB and 12 dB per octave), bandpass, or highpass filtering, which can substantially change the tonal character of the reverb. If you want a more unconventional sound, you can turn on the filter''s resonance control, use the Reverse button, or use the Crossfeed slider, which allows the left side of a stereo IR to be processed on the right and vice versa.
The Sample Rate control lets you reduce the frequency response of the IR. (In cases where the IR''s sample rate is lower than that of the song''s audio, you can actually increase the IR''s sample rate.) Each time you lower the rate, the pitch of the reverb drops an octave, and the length of the IR doubles. If you''d prefer it not to increase the length, you can click on the Preserve Length button.
Perhaps because it''s designed expressly for Logic, Space Designer was consistently more CPU efficient than IR-1 and Altiverb—its Mac plug-in competitors.
Space Designer''s IR library is split into a number of categories: Delays, FX Reverbs, Indoor Emagic Rooms, Outdoor Forests and Fields, Reverb Units, Synthesized IRs, and Vintage Gear. Together, there are over 1,000 IRs. The library doesn''t have IRs of well-known cathedrals, concert halls, and recording studios the way some of the other collections do, but it does give you a lot of choices, including IRs that were sampled from well-known reverb and delay devices.
If you want to open third-party IRs or use other audio files as IRs, Space Designer makes it easy because it can handle several different file formats, including SDII, WAV, and AIFF. Overall, its sound quality is good and it holds its own against the other Mac convolution plug-ins.
Space Designer also has a unique feature that generates what Emagic calls synthesized IRs, based on the parameter settings. They can be generated at will and give you random IR variations. Although you can''t create an actual audio-file IR from these, you can save the settings and open them in Space Designer at another time.
Pros: Flexible, lots of parameters. Good for both conventional reverb and sound design. Sample-rate control for IRs. Unique Synthesized IR feature. CPU efficient.
Cons: IR library lacks well-known cathedral, concert hall, and studio IRs.
Compatibility: Mac OS X 10.2 or higher.
Magix Room Simulator: classic version, $629; professional version, $1,249. Room Simulator, found in Samplitude 7.0, is one of the most powerful and flexible tools in this roundup. The feature runs in a variety of modes, including as an Object effect (Objects are the individual audio clips that make up a track), a track or aux insert (professional version only), and as a destructive DSP effect. Its main interface includes a graphic display of the impulse and three sets of sliders for adjusting various parameters (see Fig. 9). A second window offers advanced settings for even more detailed control.
You can start your exploration by using one of the large number of included IRs in Samplitude''s RAP format, among which are a substantial collection of files sampled from the TC Electronic M3000 processor. Some of these are used in the presets (Cool Room, Garage, Tunnel), and for traditional reverb, they are the best starting points. You can also use your own WAV files as IRs, including those on your drive and any that are currently loaded in a project.
Of course, half the fun is tweaking an IR for your own purposes, and Samplitude has lots to offer here. You can reverse or time-stretch the IR to create a variety of effects or use the drawing feature to make manual adjustments, such as adding or removing early reflections. There are separate sliders for boosting or attenuating the early and late portions of the IR. A third slider, found in the Advanced window, is for adjusting the envelope''s attack portion. And because Samplitude is a powerful multitrack audio editor (like Nuendo and Audition), you''ll have endless resources for manipulating your IRs before you apply them to your source.
Room Simulator offers separate controls for wet and dry levels and has a level meter for keeping track of the effect on overall gain. (It''s a good idea to keep your levels low when experimenting with different IRs.) Another screen offers a filter graph on which you can make fine adjustments to the frequency response of your virtual spaces. This screen shows a real-time display of the IR''s frequency response, so it''s easy to tell where any problem spots might be.
Like most aspects of Samplitude, you can customize a number of the Room Simulator''s interface elements. For example, the impulse display can be toggled between logarithmic or linear, and you can adjust the IR display''s zoom level. There''s also a control to adjust the quality of the convolution process and another to determine how much CPU power is devoted to the process. The Samplitude manual offers an excellent overview of the Room Simulator and includes solutions to common problems.
An entirely separate feature, called Convolution, is also part of the Samplitude arsenal. It has a minimum of controls and works only with files that are currently loaded in the program—you can think of it as Room Simulator Lite. The forthcoming release of Samplitude 8 will feature 5.1 surround real-time room simulation.
Pros: Flexible implementation. WAV and native-format file support.
Cons: No IR-extraction option.
Sony Acoustic Mirror, $499. Available at one time as a standalone effect and included as a feature in Sound Forge since version 5.0, Acoustic Mirror is one of the more senior convolution tools in this group. You won''t find slick graphics, skins, or 3-D buttons here. Instead, you will find a straightforward interface with several sliders, clear text, and intuitive parameters driving a fast and powerful convolution engine.
Dedicated windows provide access to the four main work areas of the program (see Fig. 10). Load a sound file into Sound Forge, and then open the Acoustic Mirror menu option (Effects/Acoustic Mirror)—you''ll find a variety of controls for tweaking the interaction of the loaded file and the IR. You can use any WAV or Sound Forge Impulse (SFI) file as an IR by selecting it in the Browse dialog box, and once you click on Preview, you can swap an existing IR for a new one without stopping playback.
If you don''t want the effect to begin with the start of the file, adjust the Response Delay slider to offset its start time from -500 to 500 ms. (Use a negative number if you want the impulse to be in progress when the source begins to play.) There are high- and low-shelving filters on hand to modify the spectral content of the resulting sound. Boosting the highs will sometimes be necessary (convolution can produce dull-sounding results if you''re not careful). You can also adjust the wet and dry mix settings individually. Adding a bit of dry signal might be useful when you don''t want to lose the character of your source completely.
Because convolution can often produce sounds that build up into utter chaos, an envelope is available to determine how much of the impulse is used. This is a good parameter to explore when you want only a little of the IR interacting with your source. You can choose to enable the envelope and limit the decay time directly from the main window. But if you want to tweak the envelope itself, click on the tab at the bottom of the interface to access the Envelope screen. Here you''ll see a graphic display of the IR, which makes it easy to tweak the breakpoints and levels for the type of time-varying control that you want. You''ll also find a button to create a preset of the current settings.
The Summary screen is a sparse window that has only three sliders—Dry Out, Wet Out, and a slider (also found in the General window) to adjust the quality-versus-speed ratio of the convolution process. The Summary window also displays information about the IR, including its length, sample rate, and number of channels, as well as any other comments you want to add. More significant is the Recover window, from which you can extract your own IRs. The process is quick and painless, and if you have any trouble, there are tips in the help file.
Be sure to grab the huge collection of IRs that Sony has available at its Web site. It''s one of the best sources for real-space ambiences that you''ll find.
Pros: Quick, near-real-time updates. Simple and logical interface. Built-in extraction tool.
Cons: No parameter automation. Aging interface.
SoundHack 0.892, free. If you''re a Mac user who''s looking to get into the convolution game but you don''t want to spend any money, check out Tom Erbe''s SoundHack, a standalone sound-processing program that has been around for a long time. SoundHack is not a convolution reverb per se: it doesn''t come with any IRs, and its interface is fairly basic. On the other hand, you can''t beat the price.
The process of applying convolution in SoundHack is somewhat convoluted. First, open a file (SoundHack can open numerous file types). Then, choose the Convolution command from the Hack menu and browse to an impulse file. Set the length of the IR (the longer you set it for, the longer it will take to process) and select from various preferences and parameters (see Fig. 11). Click on the Process button, and SoundHack will open a dialog box, asking you to name your processed file. (SoundHack''s editing is nonreal time, and it writes a new file to disk for each edit.) According to Erbe, it is best to save the file in the 32-bit floating-point AIFC format, because the format''s large dynamic range can handle the unpredictable nature of the file after convolution.
When the processing has finished, use the Gain Change command, which finds the peak amplitude and then normalizes the file. Now you can save it as a 16-bit linear file. Once you get used to it, applying the convolution effect is a relatively quick process.
There isn''t a preview function, so if you don''t like the result of the convolution, you''ll have to try it again with different settings. According to Erbe, the only way to adjust the relative levels of the IR and the source file is to mix the processed file with the original file using a multitrack audio editor.
SoundHack performs a number of other DSP processes, including Spectral Mutation—which gives you similar results as convolution—and pitch shifting. For that alone, SoundHack is worth the download. According to Erbe, VST plug-in versions of the SoundHack processors—including convolution—are on the way.
Pros: Free. Useful for sound design.
Cons: Cumbersome interface. Several steps required to complete convolution process. Few user-adjustable parameters. No IRs included.
Compatibility: Mac OS 8, OS 9, OS X.
Steinberg Acoustic Stamp, $1,499. Acoustic Stamp, found in the Process menu of Steinberg Nuendo 2.1, has many features that make it well suited for producing virtual reverb, but it''s equally at home with other types of convolution as well. At the top of its single window is a display of an envelope that is superimposed over a waveform of the impulse. A vertical zoom slider is on hand to ensure that the IR is viewable, regardless of its amplitude level (see Fig. 12). The envelope updates in response to changes in several of the parameters, so you''ve got lots of visual feedback as you edit settings.
Clicking on the Load Impulse button brings up a file browser along with an option to play files from disk, a feature that all manufacturers should adopt. The maximum impulse length is 12 seconds, and you can choose whether to use left, right, or both channels of the IR. Nuendo ships with only a few impulses—three halls, opera, shower, and stage IRs—but you can also use files in WAV, AIFF, and SDII formats.
Five controls at the top of the screen are used to adjust the impulse envelope. You can fine-tune the length and level of early reflections using a combination of the First Reflections and First Reflections Level sliders and set the reverb length using the parameter of the same name. There''s also an attenuation control that affects the portion of the impulse after the early reflections and an overall-gain setting. You can control many of the parameter values with sliders, with scroll arrows, or by typing directly into the data fields.
To the right of the envelope are three sliders to filter the IR: Lowpass, Bandpass, and Highpass. (Increments are 0 to 100 percent for each.) A second area of controls becomes accessible when you press the More button. Two of these, Pre-Crossfade and Post-Crossfade, can be used to gradually fade the convolution in or out from 0 to 10,000 ms (in 1 ms increments). The Wet and Dry sliders are linked by default; however, by using the Alt key, you can move them independently. There is also a Tail Length control for when you have a reverb time that is longer than the original file''s length.
The Preview command lets you hear how the processed file will sound, but even on the Pentium 4/3.02 GHz machine used for this article, audio playback was intermittent during preview. (Because you can undo the processing using the offline Process-History feature even after saving, you might want to simply skip the Preview command entirely.) Actual processing time was quick, however, and the finished file sounded as anticipated. Also, subsequent parameter changes won''t update until the next time you playback the file from the beginning. (Playback of the loaded audio file automatically loops). Overall, Acoustic Stamp is a handy destructive-convolution tool that complements Nuendo''s other processing features nicely.
Pros: Support for multiple file types.
Cons: Maximum 12-second impulse.
Compatibility: Windows 2000/XP, Mac OS X 10.2.5 or greater.
Tascam GigaPulse Pro,, $599. Tascam has raised the bar for virtual sonic reality with its GigaPulse Pro convolution software. More than any of the other programs in this roundup, GigaPulse gives you options to place your sound in precise room locations with great accuracy. With some of the included IRs, you can, for example, pick from one of 18 different instrument placements in the sampled room. By using the Perspective control, you can adjust the perceived distance from the performer to the selected mic. In effect, you become the recording engineer and can rerecord your instruments (or other samples) in the room.
But GigaPulse is not just about mic positions and room ambiences. In fact, it offers features that closely resemble physical modeling. You can use the included impulses of instrument-body resonances, which allow you to, for example, cross a trumpet with a cello. And because you can run multiple instances of GigaPulse simultaneously, you can use it for several tasks at once. Cascade Mode allows you to run two impulses in series, and it is more efficient than using two instances.
GigaPulse has a highly graphic interface, which allows you to view an image of an instrument with two mics hanging over it and pick the mic from which you want the convolution processing to emanate. Its controls are easy to adjust, and the interface is clean and well organized.
GigaPulse is tightly integrated within Tascam''s GigaStudio 3 sampling software (different versions of GigaStudio include one or both GigaPulse versions—Pro and SP). That allows developers to create Giga libraries containing instrument samples with convolution as one of their parameters.
Bundling convolution properties with a sample has the benefit of cutting down on the number of individual samples you''d need a library to have: rather than record a piano in ten different rooms or mic positions, you can create that effect by applying the proper IR. Or, instead of sampling a variety of pedalings, you can include pedal models that were created using convolution, such as those included in the new GigaPiano II.
This article looks at GigaPulse Pro from within GigaStudio 3, in which it resides as an NFX effect (see Fig. 13). (A standalone VST version should be shipping by the time you read this.) Opening the interface reveals a set of controls that include familiar reverb parameters such as predelay, left, right, and master level (with bypass); and wet/dry mix. A large window that resembles a surround application is on the right side of the screen, and a seven-channel surround routing matrix is located below that window.
Under the matrix is a dialog box that you won''t find in your average reverb software program. The Mic Replacement feature lets you impose the characteristics of one mic model (there are many models, including Neumann U47 and Shure SM57) in place of the original. In addition, the Original Mic feature lets you neutralize the sound of the mic used in the recording, as long as the specific mic model appears in the list of options.
GigaPulse Pro has an option called Tail Model that can synthesize the tail portion of an impulse. Although some may prefer the naturalness of a longer IR to the synthetic model, Tail Model is an efficient and good-sounding alternative that cuts down on the length of the IR you would need for long reverbs, as well as the processing time that they would require.
GigaPulse ships with a number of impulses, including a variety of concert halls and Hollywood soundstages. Some were recorded using multiple mics—often as many as seven—in multiple positions. One group was made by sampling a surround-reverb-processing unit. All told, GigaPulse''s ambitious feature set and tight integration with GigaStudio makes it a force with which to be reckoned.
Pros: Extensive feature set. Built-in mic modeler.
Cons: Current release requires multistep process to open third-party WAV files.
Compatibility: Windows XP.
As the breadth of this article demonstrates, there are plenty of convolution choices available on both Mac and Windows platforms. If you have not experienced this technology yet, do yourself a favor and try it. Chances are good that you will find it to be useful and inspirational.
EM Associate Editor Dennis Miller tested and wrote about the Windows software for this story. EM Senior Editor Mike Levine tested and wrote about the Mac side.