MIDI Vs. MP3/M4A: Before we go any further and look into the aspects of MIDI file creation, I would like to address one of the most common misconceptions that I have come across through questions that are asked of me or assumptions that are made about sound files in general. That is The Difference Between MIDI and MP3 (or other audio formats).

MP3/M4A Audio: An MP3/M4A file is a compressed version of a Wave file. Both contain actual digital data representing the frequencies and wave shapes of sounds that have already been recorded through the means of a microphone or other reproduction form. It is to computer audio what a CD is to a CD player or a cassette to a tape player. The content of the file IS sound. That is why vocals and other forms of natural sound can be stored in an MP3 or Wave file. Very little alteration can be done directly to an MP3/M4A file without changing the contents of the file to a state of distortion or non-recognition. Similar formats that stream audio, such as Real, Quicktime or Windows Streaming Media are compressed even more so than even MP3/M4A, but at the cost of further sound degradation.

MIDI: A MIDI file contains data in the form of instructions to tell a MIDI compatible musical reproduction device, such as a sound card or digital piano, what notes to play, how loud, how long, and using which instrument. Even for the best MIDI files, the quality of output is dependent largely on the quality of the device that is reproducing the sound. The sounds are not stored in the MIDI file; rather they live in the MIDI device that is reproducing the sound. MIDI is adaptable in the sense that you can change the type of instrument playing a certain sound. For example, a grand piano can be changed to a honky-tonk piano or a flute to a clarinet. However, standard format MIDI files cannot store voice data or other "live" sounds.

Advantages: In cases where the content needs to be as close to the original as possible, particularly in the case of vocals or "live" sounds, MP3 files are the clear winner. The problem is with bandwidth, or the size of the file. A MIDI file of a piano rag may be around 25k in size, where an MP3 recording of the same rag is typically well over 100 times that. This drastically alters the number of tunes that can be stored in a given server space, as well as the download time for each file. In an instance where a smaller MP3 file is available for faster loading on slower connections, this usually comes with a major loss of fidelity, thus rendering the final sound as highly inferior to the original source. If your sound module or computer card are of high quality, MIDI will retain the same quality of sound regardless of connection speed. If you have a lesser quality sound card, most browsers either contain a MIDI player or allow you to install a plug-in for playing MIDI files. This converts the MIDI data to pre-recorded waves of fairly good quality, creating an on-the-fly Wave recording for playback on even the cheapest sound cards.

Conclusion: For many years I used MIDI for the content of my site in order to bring the user maximum content quickly. Since the majority of my work is done with the piano as the intended medium, the lack of vocals on the few pieces where they might be expected were of lesser consequence, but those performances are available on high quality audio CDs and as of 2015 on medium resolution MP3 recordings on my site. Now, on to MIDI reproduction.

Originated through the cooperation of a number of electronic instrument producers, through an effort headed by the Roland corporation, MIDI (Musical Instrument Digital Interface) is a protocol that allows all compatible musical instruments, computers, and playback sources to communicate with each other with minimal alteration of the intended original performance. It is now self-regulated by the Midi Manufacturers Association (MMA), which currently sets MIDI standards. Being that it is a digital medium steeped in binary math, most MIDI signals are sent in a range of 0-127 (7 bits with a check bit left over). There are sixteen channels available to most MIDI devices, with the possible separation of channels into eight physical MIDI ports, creating 128 logical channels. Each channel is similar to the channels on your TV or radio. They carry separate MIDI signals destined for the same or multiple MIDI devices. If you have a sound card or device capable of reproducing more than one channel at a time, you can potentially hear between ten and sixteen channels of MIDI with a different instrument assigned to each channel.

The resolution of the MIDI clock varies between control devices, with the possible division of 96, 120, 240, 480, or 960 clock ticks for each quarter note. This provides for astonishing accuracy for recording or playing back performances. The MIDI signal is usually sent between devices through a five pin DIN style cable. The MIDI signal consists of several possible elements, the most common which are listed here with their typical range and other parameters:

• Note: Channel #1 to 16; Note #0 to 127, with 60 equal to Middle C (C4); Velocity #0 to 127; Clock Ticks 1 to ????
• Control Change: 7=Volume, #0 to 127; 10=Left/Right Pan, #0=Left to 127=Right; 64=Sustain Pedal and 67=Soft Pedal, #0 for Off, 127 for On. For more control change numbers and parameters consult your playback device manual.
• Program Change: #0 to 127 instrument patches. For devices with multiple program banks consult your playback device manual for bank access protocol and codes.
• Pitch Bend: #-128 to +128. The amount of actual bend depends on the range set on your playback device.
• Aftertouch: #0 to 127. Aftertouch is pressure put on the key after the initial note is played, and can be expressed as channel aftertouch (CAF) where all keys are affected or polyphonic aftertouch (PAF) where only pressed keys are affected.
• Exclusive: Open. May consist of many parameters, and is usually targeted toward a specific brand and model of playback device.

Most computer sound cards and a large number of keyboards and external modules use the GM or GS MIDI Instrument standard originally endorsed by Roland and approved by the MMA. It consists of a specific set of instruments identically mapped to set bank numbers. For example, Grand Piano is always mapped to instrument patch #1; Strings are always mapped to instrument patch #49; Banjo is always mapped to instrument patch #106. All MIDI devices should default to MIDI channel 10 for percussion and drum sets, and those using the GS MIDI Percussion standard have each drum sound mapped to a similar set of specific notes. Some MIDI devices, including many manufactured by Korg, Kurzweil, and Yamaha, do not directly conform to the GS MIDI standard. Most computer sound cards made since 1993 will conform, and Windows, OS/2, Motif, Apple and Lenix environments can remap non-compliant soundcards or external devices with a little work. For further information on the instrument and percussion mapping of your specific device, check your owners manual. For MIDI remapping, use Windows, OS/2, Motif, Apple or Lenix Help. The majority of mobile devices available since 2010 support full MIDI reproduction through an interpretive sound font, including for Apple IOS, Windows and Android smartphones and tablets.

Since there is such a wide spectrum of MIDI compatible equipment available, I will try to present just a basic overview of various MIDI setups. I use Roland and Korg devices, but much of the information here should apply to Peavy, Kurzweil, Yamaha, Casio, and Proteus among others, as well as a majority of PC sound cards. Your specific needs will obviously be subject to taste, space, experience and finances. The necessary equipment and/or software for MIDI recording and reproduction are listed here.

Input Devices: An input device is anything that creates a MIDI signal from user input. The most obvious input device is the keyboard. There are other input devices attached to software input, but they will be listed in the recording devices section.

A Korg Triton Studio 88 Key Workstation/Sampler, essentially an input device with lots of extras.

If it is a keyboard that you intend to use, don't short-change in this area. For good MIDI, a quality keyboard is essential. There are four general classes of keyboards: dummy keyboard controllers, dedicated pianos, multi-timbral keyboards/synthesizers, and workstations. The dummy keyboard or keyboard controller is the simplest input device. It is a fully weighted and balanced piano keyboard with MIDI output but no sound. Dedicated pianos typically have no more than eight piano or piano-related sounds, and often have their own amplifier and speakers built in. Many are inherently heavier than other keyboards. Multi-timbral keyboards and synthesizers have one or more banks of instrument sounds, and are capable of producing multiple layers or channels of sound. Workstations have the added bells and whistles of a built in sequencer that provide an all-in-one MIDI recording/playback device. If you can afford the full 88 keys in any of these configurations then go for it. This applies particularly to ragtime music, which in most cases requires more than the 61 keys of lesser keyboards. You may also consider getting a basic 88 key digital piano with one or more external sound modules for instrumental sounds. Also look for touch sensitivity, knowing that both spring controlled and weighted keyboards are available. Some even emulate the two-step feel of real piano hammers. The latter variety are most like a piano in feel and response, but may not be best for reproducing other instruments.

A Yamaha Disklavier™ piano that records and plays back MIDI signals using a real piano action.

An alternative similar to the digital piano is the MIDI piano or a MIDI conversion kit for an existing piano. You get the touch and sound of a real piano with player-piano like capabilities in terms of MIDI. It sounds appealing, but the MIDI sensors on these pianos need regular calibration in conjunction with the piano's normal reaction to its environment. Recording methods and results from such a piano largely depend on the acoustic environment in which it is located and recording microphone quality, whereas a digital piano records with consistent sound quality. MIDI pianos also do not allow the flexibility or touch that a keyboard will in terms of emulating instruments other than piano. They are inherently expensive, currently adding a minimum $1,000 to the cost of the piano alone.

If you want to do more than just record piano tracks then there are some controls and capabilities that you will want included with your keyboard. These include direct MIDI patch access, micro-tuning capability, a pitch bender/modulation wheel, and an expression (volume) pedal. You should have a minimum of one MIDI In and MIDI Out port on the device, and optimally a MIDI Thru port which passes on the unaltered MIDI In signal. A second MIDI Out port may yield some advantages depending on your setup. There are other MIDI input devices including those designed for guitars and electronic wind instruments, as well as some audio signal converters. All of these are designed to convert the incoming audio signal or pulse into a usable MIDI signal, thus allowing you to play a banjo using a guitar, or emulate horns with an electronic sax or clarinet. You also gain advantages offered by these instruments that are usually unavailable to a keyboard, such as string bending, controlled slides, or wind embouchure.

Playback Devices: An output device is something that processes a MIDI signal into audible sound. As discussed above, some input devices also contain the elements of an output device, but there are many stand-alone output devices. Computer users are most familiar with the sound card.

A Roland Multi-Timbral Sound Module with synthesis capabilities, going far beyond anything heard by early ragtime performers!

Dedicated MIDI users likely own one or more multi-timbral sound modules, which are able to reproduce nearly any instrument sound (and some synthesized sounds) concurrently on multiple channels. Output devices generally reproduce sound through one of three methods: Synthesis, or the creation of sound from a combination of waves or a wave table; Digital processing, which processes and plays back a digitally recorded and stored sound; Soft sound, which is similar to Digital processing, except each note is played back from a bank of sounds loaded from a file. The last method applies mostly to advanced 32 or 64 bit sound cards and digital samplers. Some external devices also contain built in sequencers for recording and storage, making them pseudo-workstations, since they do not include the keyboard. Since all of these devices have diverse sound qualities for different timbres, the one you choose should contain a pleasing combination of the sounds you use the most. For example, I prefer Roland strings and percussion, Proteus horns, and Korg pianos, so I own a variety of devices for reproducing these various sounds. Note also that if you plan to do any CD production from your computer, you will want a digital sound card or USB/FireWire connected device with a high-end analog or optical input. These typically run $400 and above currently.

An aspect of output that should never be overlooked is the method of audio amplification used for your setup. Don't expect realistic sound from your average set of computer speakers. If you are tied to using a computer sound card for reproduction, invest in a computer speaker system that has an external subwoofer. Otherwise, a component high fidelity system is your best bet for a home studio, even for piano-only sequences. If you take your show on the road, a decent 10" or greater two-speaker system with a minimum 100 watt amplified mixer or head is acceptable for most small auditoriums. Anything above a 300 seat room warrants use of the house sound system, or a rented system if nothing else is available. No matter where you are listening, don't short change your sound reproduction methods as they can actually have an effect on how you create MIDI files.

A screen from Cakewalk home studio, combining MIDI with digital audio recording and production on your PC.

Sequencers: A sequencer is a software driven device or software program that records a MIDI signal and allows you to edit individual notes and performance parameters, then save the performances for playback. For computers, this is usually a program such as Cakewalk. Sequencer programs vary in quality from simple 16 track bar displays to programs such as Finale which are capable of accurately printing out music from the MIDI input. Computer sequencer programs usually allow a wide diversity of viewing methods for information, ranging from textual representations of note parameters to graphical views of tempo. Cakewalk is currently among the most widely used sequencing software, and advanced versions of this program can also print very presentable music. If printing is a capability you desire, expect to pay a little extra for it. Other enhanced features include varying synchronization algorithms for video and motion picture film, as well as the synchronized inclusion of sound files, and a variety of software triggered events. Sequencers are also available as a component of keyboard workstations and as stand-alone hardware devices. Both of these usually have a limited alphanumeric display that allows viewing of one note at a time, but they otherwise contain the same mass editing capabilities of their computer counterparts. They also provide portability, an advantage that is discussed below. Most software sequencers and many external sequencers can create PC compatible standard midi format (SMF) files.

With a sequencer you can record multiple tracks of information, such as the instruments required to reproduce a Traditional Jazz band, or the left and right hands of a piano rag. Some sequencers allow you to redirect any of the sixteen channels of information to multiple output ports, so you can utilize different output devices without running all MIDI signals through all devices. Sequencers also allow you to edit note parameters, remove bad notes, add missed notes, change dynamics for a note or range of notes, insert or delete measures, alter tempos, and a variety of other music management chores. In essence, a good sequencer gives you the power to create almost anything you want. But don't let this fool you into thinking that it can correct poor playing, as the quality of the end product largely depends on how well it is played to begin with.

Portability Vs. Studio: For the most part, your MIDI setup will be either one of permanence or portability. A desktop PC centered workstation, or anything directly integrated into a MIDI keyboard creates less portability. Add in rack devices such as digital delays, compressors, and DAT (Digital Audio Tape) or hard drive/flash drive recorders, and you'll find yourself doing virtually all of your recording in the same place. I don't mean to imply that keyboard workstations are not portable, but my older digital piano alone weighed 115 pounds, and my current one around 50. The older one used to go places with me, but not without a lot of groaning and effort. External sound modules and sequencers can contribute to portability. When I play gigs I like to take my "band" with me, as I have produced a number of pre-recorded accompaniments for Ragtime songs. My sequencer and two sound modules easily fit in a gig bag, and require only a piano on the stage and some amplification to create the desired effect. Add to this the ability to control the starting and stopping of sequences with a pedal and it's easy going. If you prefer to keep everything in place in your studio, I recommend getting a DAT or hard drive/flash drive multi-track recorder to record your sequences digitally, so if you use them when you perform out you will still have a modicum of control and accessibility not available with standard audio cassettes.

There are many ways to approach how a MIDI file can be produced. Each has advantages and disadvantages, depending on the tools used and the musical capabilities of the individual using them. When deciding for the first time how you want to approach MIDI production, or if you feel like you need a change in methodology, it is helpful to remember the GIGO theory: Garbage In, Garbage Out. In other words, look first at the end result you hope to achieve, then learn and utilize the method which will obtain that end result in conjunction with your current playing ability.

If your MIDI tool set is limited or you are less secure in your musical abilities, you may be happy with just recreating the notes on the page to give you a better perspective on how a piece could sound. From there you can play with tempo and dynamics to some extent to map out expression. You may not get hand played or "live" results, but your recording could act as a template for a piece you are learning. If you are using a mid-range MIDI keyboard and sequencer program, you have more flexibility in what you produce. Even if you are not recording in real time or real tempo, you will have more control over dynamics, and can create a somewhat realistic tempo mapping if you are recording metronomically. If you have a full keyboard with both piano and keyboard sensitivity settings (there is a difference between the two in reference to key acceleration principles), and a full-featured sequencer program or hardware sequencer, you can create pretty much anything within or slightly beyond your current musical abilities.

Before you use any of the following methods, realistically analyze the end result you desire and/or feel that you can achieve. This may save much frustration, heartache, and time. Over time, you will acquire the ability to churn out not just piano-based files but fully orchestrated arrangements. The end result will nearly always be your reward for the amount of effort you put in. At any time during this process you run into specific snags, please refer to this page, or you may write to me by clicking on head in the top frame. Again, remember GIGO, and try to enjoy the creation process with the end result in mind.

Advantages: You only need a rudimentary understanding of music and the software you are using. For a person who is not MIDI keyboard capable, this will be the fastest method. It is easier to see and correct mistakes than with some other methods. You may also learn enough over time to apply your own creative muse and write an original piece, then have the advantage of being able to print it out for others to view or play.

Disadvantages: Most dedicated music printing software has limited playback and interpretation capability. The same is true for sequencer programs when the notes are put in through the keyboard or by mouse. When playing back printed music, notes will typically sound for 80% to 100% of their designated length, creating a possibly undesired legato effect that comes across as unnaturally smooth. All of the notes will be quantized, which means that they will fall exactly where printed within a measure instead of approximately, as a hand played recording would. Dynamics will likely be limited to literal representations of what is contained in the file, and crescendos/diminuendos are rarely utilized. It will sound mechanical and expressionless for the most part. There are a couple of programs that allow playback using the original MIDI input expression, or will interpolate it into a hand played sound through a de-quantizing process, but when creating a piece one note or chord at a time, this could result in uneven or stilted playback.

Advantages: You only need a rudimentary understanding of music and the specific steps required by the sequencer or sequencing software. This method is somewhat faster than computer keyboard or mouse entry. When using a software sequencer it is easier to see and correct mistakes than with some other methods. You also have the advantage of hearing chords as they are entered, rather than having to play them back after clicking them in one note at a time. You also gain more control over the dynamics as you can enter notes at varying volumes on a touch-sensitive keyboard.

Disadvantages: The entered notes are often adjusted to play only for the incremented time, which creates a choppy sound on playback. Alternately, they may be adjusted to sustain until the next note entry point, creating a continuous legato sound. In some cases they may all be adjusted to one predetermined length. In all cases, this requires extra editing to readjust undesired note lengths, which may end up consuming more time than other entry methods. As with the Basic method, all of the notes will be quantized, which means that they will fall exactly where printed within a measure instead of approximately, as a hand played recording would. It will sound mechanical for the most part, and will require extra tweaking to interpolate it into a hand played sound through a de-quantizing process, although this could result in very uneven playback depending on note lengths.

Advantages: If you are musically inclined, this may produce faster results than the Basic or Step methods. It will also create a more hand-played sound, albeit with some limitations. Playing against a slow metronome also tends to improve your playing in general, as well as repeating a section over and over until it sounds right. Therefore, this method also encourages good practice habits. You ultimately have more control over the dynamics, pedaling, and note lengths than with the Basic or Step methods. If you are using the hands separate entry, you can divide the hands between two tracks for a more accurate interpretation of your playing when imported into a music notation program.

Disadvantages: You are limited to metronomic constraints, and creating rubato effects with a tempo track or through tempo mapping can prove to be extremely difficult in terms of getting desirable results. In most cases, you will learn very quickly just how accurate or inaccurate your note timings are, particularly if you quantize your performance and end up correcting a lot of misplaced notes. When playing slower or using both hands to play the notes for one hand, you are more likely to hold the notes down for a longer time than you would at tempo, creating more of a legato effect than may be desired. The same is true for pedaling. Importing your performance into music printing software may require further editing due to literal interpretations of note lengths, and will likely result in a visual cacophony of dotted notes with lots of miniscule rests interspersed throughout. Some programs do have a filter that can correct this condition.

Advantages: If you are musically inclined, this method will produce faster results than the Basic, Step or Metronomic methods. It will yield a more realistic "live" sound, with the only limitations resulting from your playing or editing abilities. Repeating a section over and over until it sounds right will also improve your overall performance of the piece in the future. You ultimately have total control over the dynamics, pedaling, note lengths, and tempo expression.

Disadvantages: As stated above, your results are limited to your playing and editing abilities. Therefore, if either or both are limited or questionable, this can quickly become an extremely time-consuming and mood altering methodology. Mass editing and the location of starting and ending points becomes more difficult than with other methods because of the lack of clear logical measure boundaries. Importing Real Time files into most music printing programs and editing them to make sense is extremely difficult or close to impossible, ultimately depending on your skill and patience levels.

Orchestration is such a broad topic that I will only attempt to outline the implementation of it in MIDI. Since the combination of instruments, playing styles, rhythms, etc. is usually a matter of personal taste with a set of very wide-ranging guidelines, it is inappropriate for me to insist that my way is the way and will work for you. It may, or may not. I have been creating MIDI orchestrations for the better part of two decades in genres that range from classical orchestra for Handel's Messiah to Contemporary Christian Rock and Top 40 tunes. My Ragtime and Traditional Jazz band orchestrations make up a large part of that work, but are by no means inclusive. I suggest that you also branch out, using the ideas here, and try a wide range of musical styles. Perhaps you may attempt to emulate a recording you know of, or orchestrate a piece composed for solo piano. For ideas of just how far MIDI orchestration can be taken, including examples of Ragtime, Traditional Jazz, and Mariachi bands, I refer you to my vocal recording set, "Perfessor" Bill Sings, Volumes 1 to 3. Since the orchestrations on these recordings were produced using multiple sound modules, they cannot be easily interpreted to a single GS MIDI sound card standard. Even if they were converted, many of the tracks require up to 60 notes of polyphony to work, which is beyond the range of most sound cards, and why I instead refer you directly to the end result. For methodology on the mechanics and accepted practices of orchestration in any musical style, I refer you to your local bookstore or library, or even a thorough listen through your CD or tape collection. Read, listen, absorb, apply.