Anyone who has played Tascam's GigaSampler or GigaStudio (originally developed by NemeSys Music Technology) readily understands why it is rapidly becoming the de facto standard for PC sampler users. Because GigaSampler plays samples from a hard disk rather than from RAM, it doesn't suffer from the same memory-storage limitations that plague RAM-based samplers. That means you can use full-length samples without loops to better reproduce an instrument's sound. Granted, some situations call for looping samples in GigaSampler, but the looping doesn't compromise the sound quality or realism, because the loops can be so long they're virtually undetectable. Moreover, GigaSampler isn't bound by the same constraints as other samplers regarding the number of samples that constitute an individual multisample Instrument (see the article “Soft Sampling” in the October 2001 issue).
Combine the freedom to create a vast number of full-length, high-quality samples with GigaSampler's extensive sample organization and MIDI control parameters, and you have the capability to create and play sampled instruments that reproduce with astounding accuracy the nuances of the actual instruments.
Although this massive instrument-design potential may be liberating, it can also be intimidating to the neophyte GigaSampler user. The program's supplied tutorial shows how to create simple Instrument libraries but doesn't demonstrate how to create more complex libraries like those that are available commercially in GigaSampler format. What's more, the tutorial doesn't provide instruction in the all-important pre-production phase of development.
Pre-production is the key to creating an efficient and well-organized GigaSampler library, whether it's for private use or for commercial release. Beginners, however, often bypass pre-production and jump right into the sample-recording stage of development without clearly mapping out their goals or anticipating roadblocks. Instead of wasting time making mistakes, you can learn the secrets of effective pre-production planning by examining the approaches taken by professional sample-library developers. You may even find that mastering GigaSampler's potential is easier than you thought, as long as you're willing to properly lay the groundwork with a solid pre-production plan.
CLIMBING THE DECISION TREE
The first step in the pre-production design process is to gain a clear and detailed picture of what you want to sample: a specific instrument, a collection of instruments, loops, or sound effects. Your decision at the initial stage will determine the approach you take for the whole project, so it is very important to plan carefully (see Fig. 1).
If you want to create a guitar library, for example, you could design one that concentrates on reproducing individual notes from a specific electric, steel-string folk, nylon-string classical, or 12-string guitar. Likewise, you could design a library that includes an assortment of different types of guitars or loops of guitar riffs.
What you choose to include in your library determines its eventual production requirements. For example, specific instrument libraries are usually highly detailed, capturing as many nuances of the instrument as possible. That means you will need a lot of samples and probably a complex design structure. Instrument collections, on the other hand, may not include all the playing nuances of the single-instrument libraries but may still require a large number and variety of samples with a different kind of organization.
Jim Corrigan's Nashville High-Strung Guitars, for example, focuses on recreating an acoustic-guitar sound employed in country-music arrangements. Larry Seyer's Upright Acoustic Bass reproduces the plucked sound of an upright bass for jazz, bluegrass, swing, and other nonclassical styles.
You can also focus on a particular playing style but apply it to a collection of instruments in the style of Christian and Lane's Orchestral Percussion, Sonic Implants' Afro-Cuban Percussion, and Jim Corrigan's Celtic Instruments. Loop libraries are even more focused on a particular musical style, as evidenced by CDs such as Numerical Sound's Sly Dunbar Reggae Drum Splash and Bigga Giggas' Harmonica Essentials: Blues Licks.
If you want to sample a specific instrument, you may find that narrowing your focus and reducing the variety of sounds can yield a higher-quality result. For instance, Peter Ewers's Symphonic Organ Samples library uses only five of the most common organ-stop combinations on a historical French-cathedral instrument. By concentrating on a few no-compromise patches, he was able to deliver an excellent rendition of the instrument complete with dedicated reverb tails from the cathedral itself. If he had tried to incorporate all the possible organ-stop combinations, producing or selling the CD-ROM wouldn't have been feasible. Likewise, Seyer excluded bowed sounds from his Upright Acoustic Bass library because that playing technique did not fit with the intended musical applications.
MANO A MONO
Before you move on to the next stage of pre-production, decide if you want to record your samples in mono or stereo. Your decision should be based on a number of aesthetic and technical factors. Although it seems that everyone samples in stereo, sometimes it's preferable not to. For example, Bigga Giggas' Rhodes for GigaSampler and Hohner D-6 Clavinet use mono samples because the original instruments had mono outputs. Loop libraries are almost always stereo because loops are often the bedrock of an arrangement. Sound effects are usually recorded in mono, so the end user has more control over the placement and movement of the sound in a stereo or surround-sound environment.
Corrigan also suggests examining an instrument's sound-radiation pattern to determine whether its samples should be mono or stereo. For instance, a trumpet might be sampled in mono because its sound-radiation pattern emanates from the bell and is therefore more localized. On the other hand, an instrument with a more complex radiation pattern, such as a guitar or piano, will probably sound better if sampled in stereo. If you're trying to capture the sound of the trumpet and the room in which it is played, however, stereo sampling is a better bet.
Corrigan broke his own rule when he created his Nashville High-Strung Guitars library. He used separate mono recordings of the low-voiced and high-voiced guitars panned hard left and right, because that's how the sound is recorded with live players in Nashville studios. As you can see, the musical context alone might determine the final approach.
A couple of technical considerations may also influence your decision to create mono or stereo samples. If you plan to create a GigaSampler Instrument by layering samples, keep in mind that GigaSampler lets you layer two stereo or four mono samples per note. (GigaStudio allows you to layer four stereo or eight mono samples per note.) If you know ahead of time how many layers you'll need, you can decide if sampling in stereo is necessary or even possible.
In addition, GigaSampler uses one of its Dimensions (sample control sources) just for stereo samples. The program lets you set the resolution for each of its five possible Dimensions (seven in GigaStudio) up to a total of five bits (or 32 splits). For instance, mono samples don't use any bits, but stereo samples use one Dimension bit. That leaves four bits (16 splits) to be divided among the remaining Dimensions. (The number of splits drops exponentially.)
In addition to the bit used for stereo, two bits (four splits) are used for the four Velocity levels, and two bits are used for the four sample layers (see Fig. 2). That effectively uses up all the possible Dimensions. If you suspect that you might need to use many Dimensions for other purposes, be sure to use mono samples.
THE NEXT STEP BEYOND
The next pre-production step involves analyzing how a performer actually plays the instrument in the style that you're trying to capture. That is important because you must generate a performance model that you can later apply to various MIDI controllers. Moreover, the model will influence how you design the GigaSampler “translation” of the instrument, and it will give you an indication of the types and numbers of samples that you'll need.
Strummed instruments, such as guitars, ukuleles, and banjos, may need a variety of sampled chords that are switched with trigger keys. Plucked instruments may require string-bending or damping effects; acoustic and electric pianos rely heavily on Velocity control and a sustain pedal.
A pipe organ, on the other hand, is more like a wind instrument; its sound doesn't decay into silence like a piano's. For greater realism, you have to loop a sampled organ's sound and use separate release samples to imitate the action of stopping the bellows from blowing air into the corresponding pipe.
MAKE MINE MIDI
Once you have determined your instrument's behavioral model and have selected which of its articulations you want to capture, you must choose the appropriate MIDI controllers for your instrument. Naturally, you will have to set up your library to work with keyboard controllers, because they're the most widely used type. However, other controllers may work as well or even better.
For example, Seyer's Upright Acoustic Bass has an Instrument that lets keyboard players “switch strings” while playing by using trigger keys. The CD-ROM also includes four Instruments, one for each string, that you can load into separate MIDI channels to play with a MIDI guitar or bass controller for a more natural playing option. In a like manner, percussion samples often produce more natural-sounding percussion tracks when played with a percussion controller.
LIVE OR MEMOREX?
Once you have established which controllers you will use, you must determine whether it is possible to play your Instrument in a way that generates all of its nuances in real time. For some instruments, such as pianos and percussion, the answer is a definite yes. For other instruments, you may have to assign different articulations to separate Instruments and load them into GigaSampler at different Levels. (Levels let you layer patches.) For example, the gadulka (Bulgarian fiddle) sound included in Quantum Leap's Rare Instruments CD contains separate Gadulka and Gadulka Resonance Instruments. To reproduce the authentic sound of the instrument, you load Gadulka into Level 1 and Gadulka Resonance into Level 2 on the same MIDI channel to hear both the bowed and sympathetic strings.
Sometimes it isn't possible to incorporate all the desired articulations into a single GigaSampler Instrument. In that case, you may have to create separate Instruments for the extra articulations. For instance, Thomas Skarbye's Scarbee J-Slap Bass crams a huge assortment of articulations into its Studio Bass multisample. Even so, separate Instruments were created for slides and fingered harmonics that have to be accessed through patch changes. You might be able to play and generate patch changes quickly enough to play all the articulations in real time, but in most cases, you'll have to take a couple of recording passes in your sequencer and perhaps do a little editing as well.
Another alternative, which Scarbee J-Slap Bass employs in its Jam Bass patch, is to create an Instrument that has fewer articulations but that still offers a realistic sound. Seyer's Upright Acoustic Bass has an Instrument named Optimal Upright Bass that takes the best of the capabilities of the Switchable String Position Bass to make it easy to play in real time (see Fig. 3). It's easier for the performer to use this patch because it lacks the extra element of key switching to move string positions.
Determining whether it is possible to play your instrument in real time implies that you have started the process of assigning articulations to Dimensions. It isn't as complicated as you might think. You should have already determined whether the samples are to be mono or stereo, so that's one Dimension you have assigned. You may not know exactly how many Velocity levels (second Dimension) you will eventually use, but a maximum of four usually is sufficient for most projects. Depending on the instrument, you may not need Velocity splits at all; that would leave you with even more room for other Dimensions. Other common Dimensions are MIDI Sustain and Mod Wheel.
Keep in mind that assigning Dimensions is preliminary at this stage; you might change some Dimension assignments later. However, remember two things. First, you should document what you do and update that documentation as you change things. Second, you should try to assign Dimensions in a way that makes sense. For example, don't have the Mod Wheel control the sustain samples when the Sustain Pedal is a better choice. GigaSampler guru David Govett recommends that you assign Dimensions to the available features of most MIDI controllers rather than assigning them to controls that you may have but which are not commonly used by other people.
Believe it or not, you are now at the final stage of pre-production, where you are about to put your preliminary library design to the test. Before you begin sampling, make sure your instrument is in tip-top shape and that the instrument and related equipment don't have mechanical or electrical problems. You may also find that it is necessary to take extra steps to improve upon the instrument's natural sound. As an extreme example, Corrigan developed a program to generate open tunings using different string gauges to eliminate fingered string buzzing in his Nashville High-Strung Guitars library. Nick Phoenix, producer of Quantum Leap's Rare Instruments library, used a Roland VP-9000 processor to stretch the sound of a rag-dung (10-foot Tibetan trumpet) in real time as it was sampled.
Now record some test samples of the instrument and some of the articulations you wish to capture. Listen to the samples to check for flaws inherent to the instrument that may not have been obvious when listening with the naked ear. Jim Van Buskirk (founder of NemeSys Music Technology) recommends that you create and play a mini-Instrument with a three-octave range so that you can test the samples and the preliminarily assigned articulations together in context. Use this alpha test to determine what's right about the Instrument design and to isolate any problems.
Once you've run some test samples and you're ready to start sampling in earnest, create a recording template similar to the one shown in Fig. 4. That example comes from the session work for Nashville High-Strung Guitars. Corrigan used 12 tracks to indicate the 12 major-seventh chords he needed to record. On each track, he indicated the different samples he would need for that particular chord. Reading from left to right, he inserted a part for Left Guitar Upstroke, Left Guitar Downstroke, Left Guitar Damped, Right Guitar Upstroke, Right Guitar Downstroke, and Right Guitar Damped. In a similar manner, you should give the tracks and parts in your template names that reflect, in an organized manner, the samples that you want to record. It's important to have a template that moves the sampling sessions along in an efficient and logical manner and to have consistent sample names that are easily identifiable.
JUST DO IT
You are now ready to move into the production phase. Don't forget to name samples logically and to use MIDI note names or numbers at the end of the sample file name. As stated in the GigaSampler tutorial, the Instrument Wizard uses that information to speed up the process of mapping samples in the Instrument-creation process. In addition, remember to create and appropriately name folders so you can more easily organize and access samples by type. Afterward, you should be able to create your Instrument using the Instrument Wizard to lay out sample maps and the Dimensions to control how those samples interact when you play the Instrument. Good luck; I hope to hear a great GigaSampler library from you soon.
Zack Pricewishes to thank GigaMasters Jim Corrigan, David Govett, John Thomas, and Jim Van Buskirk for their invaluable advice and feedback. Additional thanks to Garth Hjelte for his input regarding sample-translation software.
MAKING NEW CONVERTS
You've just bought Tascam's GigaSampler or GigaStudio but haven't had the chance or the cash to invest in a CD-ROM library, and you're not ready to dive into creating your own sample library from scratch. You also have a boatload of samples in other formats that you want to use in GigaSampler. What do you do? Fortunately, GigaSampler and GigaStudio include a program called S-Converter that lets you convert Akai S1000 and S3000 files on CD-ROMs to GigaSampler format. GigaStudio also includes A-Converter, which lets you rip tracks from audio CDs so you can create loop and sound-effects libraries in Giga format more easily. In addition, Instrument Editor 2.0 in GigaStudio lets you load SoundFonts 2 (SF2) and Microsoft Downloadable Sample (DLS) files for easy conversion to Giga-format files.
Several third-party programs can convert between different sampler formats including GigaSampler. Chicken Systems' Translator 2.1 ($149.95; www.chickensys.com) and Amazing Sounds CDxtract 3.5.1 ($79.00; www.cdxtract.com; see Fig. A) convert a variety of sample file types, but they read only CD-ROMs or files on removable and fixed hard drives. Neither reads sample files from floppies. It is, however, possible to find utility programs that can put floppy-disk images onto a hard drive. Translator even provides those free utilities as part of its full package. Both CDxtract and Translator come in GigaEdition packages ($79.95 and $59.95, respectively) that translate a variety of sample formats into just GigaSampler format.
Another file-conversion program with a long history is FMJ Software's Awave Studio 7.3 ($89.95; www.fmjsoft.com). Not only does it convert a variety of sample formats but it also reads and converts numerous tracker program files, disk images of various Ensoniq synthesizers, and a variety of obscure audio-file formats. Unlike CDxtract and Translator, Awave Studio only reads CD-ROMs that are in ISO9660 format. It can, however, read floppies if they're in MS-DOS format.
All of the conversion programs do a fine job of converting sample libraries, and all use the Windows Explorer — style user interfaces. However, no one program is capable of translating all file formats, so check each company's Web site for updated file-conversion information.