Posted in: Recording by Maya Tong on March 24, 2009 | 1 Comment
Wanna make music like pros? My series of tutorials guide you to understand how our modern music is produced. Here, you’ll find prime concepts, main points and tips of MIDI synths and samplers.
Hello, I’m Maya. I am a multimedia artist who is very fond of music composition and recording, as well as a researcher in modern music production. My specialized areas in music include composition for MIDI orchestra, electronic wind instruments and surround audio mixing for Hollywood film score. To know more about me, please visit http://mayatong.blogspot.com
In my series of music/audio-related articles, I would like to guide you from beginner to veteran knowledge of modern music production, in terms of software and hardware. Other than just explaining each concept, I will try to include reasons, pros and cons, and tips. That should inspire some usages or ideas that even advanced people have never paid attention to. I will update my articles very often, so stay tuned!
In my last article (Part I – sequencer), we have talked about the purposes of music software are for (I) arranging and sequencing music, (II) generating instrumental sound, and (III) manipulating sound, and also talked about software used for sequencing and arranging music, in a way similar to our classic maestros writing on a staff. In this article, we will talk about a kind of music software generating instrumental sound, called softsynth and sampler, in modern music production. I hope this article can make your mind clear on a question like, ‘What software is used to generate musical sound?’
While classic composers conducted a live orchestra to turn a staff into beautiful melody, modern composers obtain melody by acquiring instrument sounds from MIDI equipment. In the old days, it was done by hardware synthesizers, usually called sound module or tone generator (contained built-in sound only), and sampler (designed for putting user own sounds in). Famous sound modules and samplers were from Roland, Yamaha, Korg and E-MU etc. Since a decade ago computers become more powerful, all these hardware modules or samplers turned into software. We now have two kinds of popular software that generates sound, they are (1) softsynth and (2) sampler.
figure 1: a typical hardware sound module (Roland XV-5080). This module also had a sampling function, which loaded user’s samples like an ordinary hardware sampler.
Softsynth is derived from ’software synthesizer’. Sometimes older softsynths are also called ‘rompler’. The earliest well-known softsynths are like hardware sound module, contained only built-in sound ROM (usually less than 100MB) and limited parameter controls. You can, however, add audio effects like reverberation, chorus or delay, as well as some hardware-inheritable controls like key-shifting, master-tuning and polyphony-limiting (polyphony determines how many notes can sound simultaneously. Modern softsynths can handle over 200 notes at a time while old ones only a few dozens.) These softsynths were general-purpose, which means each contained quantities of instruments like pianos, strings, winds, percussion and electronic instruments, much like a General MIDI (GM) hardware module. The quality of each instrument is comparatively lower than the newest softsynths that contain only a few specific instruments (discussed below). However, some general-purpose softsynths are still used by modern musicians, examples are Roland Virtual Sound Canvas, Cakewalk TTS-1, Luxonix Purity and Steinberg Hypersonic.
figure 2: a typical look of a general-purpose softsynth. It is Yamaha S-YXG50, one of the earliest popular GM softsynths.
Since the occurrence of soft samplers, softsynths faced their direct competitors. In terms of sonic quality, samplers were usually deemed much better because of the detail for each instrument captured in samples. Famous samplers beat GM-based softsynths by releasing GM samples (over GB large), like Conexant GM500 and Native Instruments Bandstand. When everyone thought softsynths soon became vintage stuff in MIDI world, they made a quantum leap in recent 4-5 years. Synful Orchestra, the earliest well-known softsynth which claimed they used advanced synthesizing technology to reproduce articulation realism of orchestral instruments with a small sound ROM (contains phrases but not single-hit sound), made a playable softsynth as expressive as real performance. The sound ROM (hundreds MB) was small compared with samples (over GB). The result was really amazing as, at the time, musicians were upset with softsynths (and even GM samples) that sounded mechanic from note to note due to no solutions to “smooth out” the transition in between (i.e. legato, as will be discussed below). From the time, rather than having quantities of instruments in a softsynth, newer ones tend to take the following approach – They designed softsynths for specific purpose or specific instruments by using some kinds of mathematical modeling synthesized in real-time with (or without) a small sound ROM to produce a quality, ‘playable’ instrument. Examples are Mr. SAX T. and The Trumpet from Sample Modeling, Simple-media Super String, Soundfonts It MileStone and Wallander WIVI.
Bear in mind that these softsynths, are not like their predecessors, reply heavily on synthesizing technology than just playing back a huge amount of captured samples. In this way, musicians can save a lot of memory resources for other usages. However, due to real-time synthesis a powerful CPU is required when using a lot of them together.
Unlike rompler, soft sampler (or just sampler) can temporarily store a much larger amount of custom-made samples for playback purpose. The actual memory limitation is the size of your hard disk and RAM. Currently, popular samplers are Native Instruments Kontakt, IK Multimedia SampleTank, Steinberg HALion, Logic EXS-24 and Tascam GigaStudio.
Since samples are actually a bunch of digital wave files collected together, we can sample an instrument as detailed as we want. Sample library (a bunch of sampled instruments collected together) is designed to use with samplers to provide high-quality playback of each instrument. A high-quality sample library usually contains hundreds of articulation samples, from short to long notes, as well as dynamics layers for each instrument. By ‘assigning’ these samples and layers to corresponding notes on a staff, composers can make a realistic melodic line like a real instrumentalist playing with different dynamics and articulations.
Sample Libraries succeeded in reproducing acoustic instruments, especially for classical and modern film orchestras, as these instruments normally cannot be reproduced realistically by any softsynths due to sound ROM limitation. Examples of high-quality orchestral libraries include Vienna Virtual Instruments, EastWest Symphonic Orchestra, ProjectSAM Symphobia and Miroslav Philharmonik. More and more composers utilize these libraries to compose scores for movies, capable of sounding as big as a Hollywood orchestra.
Beside conventional western instruments, the current trend of sample libraries is towards ethnic instruments: Duduk, Erhu, Shakuhachi and Dilruba etc. frequently appear in libraries nowadays.
The main differences between these two creatures are (1) sonic quality and (2) storage size. Softsynths are generally deemed as lower quality, due to the fact that they usually rely on sound ROM, which limits to the size around hundreds MB to a few GB of instrument sounds. Each instrument in the ‘preset’ ROM cannot be manipulated individually while sampler can do so even a digital wave of a note in a sampled instrument. On the other hand, high-quality samples counts up to hundred GB, which include thousands of high-quality samples to represent an instrument precisely. However, the amount of memory used in hard disk and RAM is too large (samplers usually offer users to ’stream’ a sampled instrument into RAM for quick access in real time), which burdens composer with hardware investment.
In the old days of softsynths and samplers, samples that are captured professionally (i.e. clean recording, broad representation of sound frequencies, each note sampled chromatically etc.) contented musicians. Softsynths and sample libraries gradually offer higher bit depth (24-bit instead of 16-bit) and higher sampling rate (48kHz, 96kHz instead of 44.1kHz). However, ‘high sonic quality’ cannot resolve the following problem: since notes are sampled individually and only once per note in each dynamics, it is impossible to easily reproduce some subtle effects, such as legato (the subtle change of pitch between two adjacent notes) and note variations during repetition (e.g. the same note played repeatedly, usually very fast). Unfortunately, human ears are sensitive to these subtle effects when judging an instrument is real or not. To address the problem, two popular techniques, usually done by scripting in a sampler or a synth, are developed. “Legato function”, a famous tool first developed by Vienna Sample Library, to solve the problem regarding legato passage. “Round robin” technique, first used by Kontakt sample libraries, is alternating samples of the same note being played when repeated notes occurred. Modern competitive libraries frequently rely on these techniques to make their samples more playable. However, for modern synths and samplers, the following rule still holds in general: samples are usually higher in sonic quality while softsynths are better in playability.
Soon the discrimination between synth and sampler becomes minimal. A proof of this phenomenon is human vocal synthesizer and sample. Yamaha Vocaloid and EastWest Symphonic Choirsare examples. These vocal synths and samples create believable human singing with meaningful words by synthesizing specially-scripted engine and vocal samples. The process usually takes a lot of CPU power than normal instruments in synthesizers and samplers. It seems to be the trend of future synths and samplers, which makes intensive use of special-designed scripting engines with high quality samples, as it is possibly the way to offer musicians a playable MIDI instrument with high sonic quality.
figure 3: besides voice synthesis engine, Vocaloid has its own vocal editor (like a piano roll) to edit every word in a lyrical passage.
Instead of sampling a single shot of a note, a loop contains sampled phrases with additional information like key and tempo (so users can change the key and tempo of a loop when composing music). Sophisticated samplers that load loops are capable of editing a note, or the order of notes, inside a phrase, but not as effective as their single-shot counterpart. For most people, using loops are the easiest way to compose a piece of music without using a staff. Reason, a famous sampler/sequencer from PropellerHead, provides tons of quality loops and editing interfaces for different loop libraries, which is very convenient to composers who rely on loops to make their music.
figure 4: one of the coolest things of Reason is that you can connect all synths and samplers graphically via cables, much like dealing with real hardware modules and cables.
Choosing among Synths, Samples and Loops as Instruments
The basic rule to choose suitable sound generator software is to consider (1) composing style, (2) hardware limitation and (3) creativity.
In terms of composing style, for beginners who don’t have classical training, loop samples are a good choice to compose a piece of music in no time. On the other hand, GM synths are best for sketching your music on a staff, as you can take as many instruments as you need from the synth. Specific-purpose synths and samples are for people who are fond of composing specific instruments, and want some playable instruments in hands. For people who strive for achieving high quality sound, especially writing scores for TV, movies and games, sample libraries are almost the only choice satisfying your need.
In terms of hardware, a powerful CPU tends to be advantageous when using heavily-scripted synths and samples. A big size of a sample hard disk (100GB+) and a large amount of fast RAM (4GB+) are vital if you intend to run many sample libraries together. A fast spinning HD can help a bit in loading samples for very big libraries (assumed you have not enough RAM for streaming). GM synths are usually the best if you have limited HD space and RAM. For loops, hard disk space is more vital than the amount of RAM.
In terms of creativity, there is no boundary that restricts you to use either samples or synths. Sometimes you get the best result when layering loops with samples, or a piano synth may sound aesthetically better than a sampled piano, who’s know? Most composers own hundreds of synths, loops and samples for different situations. However, the bottom line is that you should start with a few favorite synths and samples, and to be familiar with them, before looking for another synth or sample – This process could be endless.
Of course, there are some websites contain samples/synths being downloaded for free. Some are high-quality too. GIYF again 
Okay, I hope you can understand better for different synths and samplers. We will go on talking about in detail those software manipulating sound, which are vital for mixing and shaping your sound, in the next article, so stay tuned!
sk October 24th, 2009 at 6:51 am
Nice infos – very helpful overview