YOU SAY YOU WANT A CONVOLUTION
A convolution reverb like the IR-1 can analyze what happens when a signal with known characteristics, such as a wide-range frequency sweep, plays through speakers in a room (or through an effects device or other processor). It then records the resulting spectrum to one or more audio files. Waves calls these “sweep response files” because the IR-1’s impulse is created by a logarithmic frequency range sweep; other programs may use a pistol shot, or other sound that contains a wide spectrum of frequencies. The convolution reverb can then load the recorded file, and “impress” its characteristics on another signal. The process resembles sampling in the sense that it captures the characteristics of something that actually exists, rather than synthesizing a particular sonic characteristic.
(When sampling hardware devices, remember that the IR-1 is primarily intended to capture real room responses for reverb. When you sample processors that include effects like pitch shift and modulation, the results may be unpredictable — although possibly interesting, as I learned by sampling Eventide H3000 effects. On the other hand, sampled delays and reverbs should sound very close to the originals.)
STEP BY STEP, SLOWLY I CONVOLUTED . . .
Determined to have my sound, I took the plunge into impulse response recording. The instructions in the Waves documentation were sparse, so I thought others would find my method for stereo or mono sampling with Waves’ IR-1 helpful. Once you’ve mastered this procedure, you can likely extend it to surround sampling with the IR-360.
To use IR-1 or IR-360 you must have a computer-based digital audio system for a host, so we’ll assume you can use the same system to record your sweep responses.
1: Create a session. Start a new session (or “project,” “song,” etc., depending on your software) at the highest available sample rate and bit depth, up to 96kHz/24-bit. Create six mono tracks. Set the input to “monitor” mode rather than any auto-monitor option (the usual default).
2: Choose a sweep file. IR-1 includes a disc labeled “Sweep Files and Audio CD.” Open it as a data disc in your computer, then locate the sweep file that matches your session’s sample rate and bit depth. If you don’t have the disc available, download sweep files from www.waves.com (the choices range from 44.1kHz/16-bit to 96kHz/24-bit).
3: Import the sweep file. Import the desired sweep file into tracks 1 and 2, and position both to start at exactly one second. Route track 1 to the left main output and track 2 to the right. To avoid feedback, tracks 3 thru 6 should not be routed to the main L/R output. Now “save as” a copy of this session as a template for future use, and continue working with the original.
4: Make a destination folder. During installation, IR-1 creates a folder named “IR-1Impulses V2” that contains all of the Version 2 program’s impulse response files. On a Windows computer, this folder’s default location is C:\Program Files\Waves\Plug-Ins\IR-1Impulses V2. Find this folder, then create your own folder within it to store your effects.
5: Prepare to record. We’ll consider three possible modes: mono (mono in/out), m/s (mono in/stereo out), and full stereo (stereo in/out). We’ll cover each as if you’re sampling a hardware device, but the method for sampling a real space is identical (see the sidebar “Sampling a Space with Mics and Speakers” for information on suitable speaker and mic setups). When sampling a room, the “device input” is the input of the power amp or powered speaker(s) that play back the sweep tone, and the “device output” is the output of the mic preamps that pick up the impulse. (Note: When sampling a space, begin playing the sweep at a very low level to avoid damaging the speakers, then raise the level to produce an adequate sound pressure level at the microphone position. Somewhere around 80 to 85dB SPL should be good, but the exact level is not critical.)
- Mono: This is primarily for mono hardware effects. Connect one of your computer system’s main outputs to the device input, connect the device output to a record input on the computer system, route it to track 3, arm the track, and play the sweep. See that the signal flows properly to the device and back again, and adjust your levels at each stage for a strong signal with no clipping. Let the sweep play all the way through, because with some effects the output level will be much louder toward the end of the sweep.
- M/S (mono/stereo): Follow the above general procedure, but send the left device output signal to track 3, the right to track 4, and arm both tracks. Adjust gain as needed, making sure that the L and R levels are the same at each point in the signal path.
- Full Stereo: This requires two passes, and creates four response files. Connect the left and right out from the computer to the left and right in of the sampled device. Send the left device output to tracks 3 and 5, and right out to tracks 4 and 6. Arm tracks 3 to 6, play the sweep, and set your levels. Be sure to set identical input gain on all four tracks, and do not change levels between record passes. (Note that while you can create a stereo effect by recording L and R responses in a single pass, then using IR-1 “efficient” mode to process the files, there’s no real need to do this. The two-pass method takes little extra time, and once you have a full stereo effect, you can always load it into IR-1 “efficient” to conserve processor power if needed.)
6: Record sweep responses.
- Mono or M/S: Start recording from zero time and be sure to run past 26 seconds. The actual sweep lasts only 15 seconds, but the program expects to see a full 10 seconds after the sweep ends to allow for a long reverb tail.
- Full Stereo: For the first pass, mute the right output from the computer system and disarm tracks 5 and 6. Record as described above. For the second pass, disarm tracks 3 and 4 and arm 5 and 6. Mute the computer system’s left output and unmute the right. Record as before; on all record passes, be sure to begin at zero and continue past 26 seconds.
7: Trim the sweep response files. Trim the recorded tracks to discard any signal before 1 second and after 26 seconds. This ensures your effect will have the correct predelay. (After you’re proficient with recording room samples, try taking the process to the next level by changing the predelay time, thus making the apparent listening position sound further from, or closer to, the stage. Do this by trimming the response to begin a little earlier or later than I second; but avoid cutting off the response’s start, and remember each region must be exactly 25 seconds long.)
8: Save each trimmed sweep response file as a .WAV file. The destination should be the folder created in step 4. In Pro Tools, use the “export selected as files” command in the Audio Regions List Menu. With other software, you may need to “render” the selected regions to save them as new .WAV files. Name each .WAV file for the particular effect you’re sampling, with a suffix to indicate the type of sweep response it contains — when making several recordings of the same effect with different parameter settings, give each variant a unique effect name. With some programs it’s easiest to let the software name the files as they are saved, then rename them manually. Name the files as follows:
- Mono: [effect]_M.wav
- M/S: [effect]_L.wav and [effect]_R.wav
- Full stereo: [effect]_LL.wav, [effect]_LR.wav, [effect]_RL.wav, [effect]_RR.wav (in that order when exported from tracks 3 to 6).
9: Create the impulse response file. Insert IR-1 into a bus in your session. Choose the version that matches your sampling mode (“mono,” “m/s,” or “full”). Open the IR-1 window, click on Load, then select “import sweep response from file” from the Load menu (Figure 1). In the “open” window that pops up (Figure 2), navigate to the folder containing your .WAV files and load them in the order required by the program: L, R for m/s, and LL, LR, RL, RR for full stereo. After selecting all the required .WAV files, IR-1 will crunch the numbers for awhile and store the result as a .WIR (Waves impulse response) file in your effects folder. The program will automatically name this file based on the first .WAV file you loaded.
10: Save a preset. In the main IR-1 window, click Load and select “Import impulse response from file” from the menu. Select the .WIR file created in step 9. Next, back in the main IR-1 window, click on Save and select “save to new file.” This creates an .XPS file; name it as desired. Next, you will be asked to enter the preset name, which will appear in the load menu. Give both the .XPS file and the preset the effect name you chose in step 8.
11: Check it out! Remove the IR-1 insert, reinsert it, and load your new effect. If everything went correctly you should see your effect folder in the load menu and find your new preset inside). Import some music into the session, and give a listen. When you’re satisfied with your effect, you can delete the .WAV files or move them to another location for safekeeping. Be sure to leave the .WIR and .XPS files as they are. If in the future you want to rename or move the .WIR file for some reason, you can do so, but afterwards you will need to perform step 10 to create a new preset.
Congratulations! You are now a member of the convolution reverb sampler’s club.
Product type: Convolution reverb plug-in for creating ultra-relatistic ambiences.
Target market: Studios that want a convolution reverb effect, but who also want to create new and original effects.
Strengths: Excellent sound quality. Ships with over 120 useful impulses. Relatively low CPU requirements. Can use your own impulses to create novel effects. Many adjustable parameters.
Limitations: Documentation on creating your own impulses is sketchy. Not cheap.
Price: $800 (list)
SAMPLING A SPACE WITH MICS AND SPEAKERS
In addition to playback and recording equipment, you will need one or two monitor speakers, two identical preamps, and a matched pair of mics. Either cardioid or omni-directional mics will work, depending on how you intend to set them up. Every element in the signal chain should have a frequency response as close to perfectly flat as possible. For suggestions about what equipment to use, see the detailed descriptions of the effects supplied by Waves.
For mono/stereo recording, center a single speaker at the performance position. This is considered best for a reverb you intend to apply to monaural sources such as a lead vocal, or single instruments. To capture a full stereo image, place two speakers symmetrically left and right of center. I know of no “rules” about how far apart the speakers should be; probably the best approach is to record a number of different responses with the speakers set at different, carefully documented spacings.
There appears to be more of a consensus on mic placement, although there’s certainly no harm in experimenting. The orthodox methods are either to position two omnidirectional mics roughly 1.5 meters apart, or a pair of cardioid mics in an X-Y configuration at an angle of 110º, with the capsules exactly 17cm (6.7") apart. Distance from the speakers will depend on the characteristics of the space you are sampling and the type of sound you want to create. Again, multiple recordings at various distances is a sensible approach.
To sample my barn, I chose a single speaker (a Mackie HR824) and a pair of Beyer M201 microphones in an X-Y pattern. As the space is fairly small, my main purpose was to capture an ambience that would be useful on vocals and solo instruments. I’m pleased with the results so far, but I intend to try again soon with a pair of the new Neumann KMD series digital microphones (which I’m scheduled to review in an upcoming issue of EQ — stay tuned).