All the effort that you put in composing, sequencing, editing and orchestrating your project deserves the best final results. The mix stage is where everything comes together. All your tracks, EQ, effects, etc. are channeled down to a stereo track, or to a multi-channel mix in the case of a surround project. Lets take a look at how to approach the final mix and how to solve a few of the most common issues that often puzzle the modern composer/producer.
For a generic hybrid project composed of MIDI and audio tracks, the first step that needs to be taken in order to approach the final mix is to decide how to handle the MIDI tracks. If your session contains a combination of both MIDI and audio tracks the first thing to do is to decide if the MIDI tracks need some extra tweaking from an audio (eq, reverb, effects in general) point of view. The sound associated with the MIDI tracks is generated by the MIDI devices externally from the computer (unless the tracks are assigned to a software synthesizer). Therefore in order to add effects to them you have to use the external processing power of your mixing board and outboard gear, or the built-in effects of the MIDI devices. This technique is definitely faster and can save you some time but it is also limited by the number of channels and features of your mixing board, and also by the number of audio outputs present on your MIDI devices.
If you decide to opt for this solution, then the procedure is fairly simple. The audio and software synthesizer tracks of your sequencer will be mixed taking advantage of the internal plug-in effects installed within the applications, while the MIDI tracks assigned to external MIDI devices will be processed and mixed using the equalizers and effects connected to your mixing board or, in the case of a digital board, through its internal effects. After applying all the effects and making the changes needed, you will mix down the main output of the mixing board to a stereo audio track of your sequencer. As I mentioned before while this technique is probably the fastest it has some limitations and it also presents a few problems. For example it is usually pretty hard to match the reverb, eq. and dynamic effects of the plug-ins hosted by the sequencer and the ones of the outboard gears. By using two different sets of effects (plug-ins and external gear) most of the time you will in fact accentuate the already noticeable difference between audio and MIDI tracks. In addition, if your MIDI devices dont feature enough separate outputs, it is hard to effectively apply eq. and effects to single MIDI tracks.
A better solution is to bounce all the MIDI tracks to audio and then mix the entire project inside the audio sequencer. This approach will guarantee a more coherent and balanced mix. The drawback to this technique is that it has to be done in real time for each track, making it very time consuming. If your project is 5 minutes long and you have to bounce 20 MIDI tracks it will take almost 2 hours to finish. Keep in mind that this figure can vary depending on how many input and busses are available respectively on your audio interface and your mixing board. The higher the number of inputs on the interface and the higher the number of busses available on the mixing board the faster the process will be.
DSP power is another issue. By adding 20 more audio tracks (the ones converted) to the already existing ones you will stress the CPU more, limiting and its capacity in terms of plug-in count. Nevertheless, this technique is preferable as long as you have a powerful enough computer to handle a high count of audio tracks and plug-ins. By recording each MIDI track individually as audio material you will have greater control over EQ and effects such as reverb and dynamics. To record each MIDI track as separate audio tracks, you have to solo the MIDI track you want to record, record enable an empty stereo track in the sequencer, route the channels of your mixing board receiving the output from the MIDI device selected to a bus and connect the output of the buss to the input of the audio interface attached to your computer. If you have multiple busses available on your mixing board you can actually transfer several tracks (one for each buss) simultaneously. If your mixing board doesnt have a buss system you can use its main output as a temporary buss and connect the main output of the board to the input of the audio interface (make sure not to send the output of the audio interface back to the board in order to avoid a feedback-loop). While this can sound a bit intimidating by the description take a look at Fig. 1 and Fig. 2 to compare the two options respectively.
Fig 1: Signal routing to record MIDI tracks as audio with a multi-buss mixing board
Fig. 2: Signal routing to record MIDI tracks as audio with a mixing board without busses
In the first example (Fig. 1) the mixing board features a main set of output (L-R) and 4 busses (1 through 4). The inputs from MIDI device 1 are bussed to buss 1 and 2 and then sent (through the busses) to input 1 and 2 of the audio interface. In the same way, the inputs from MIDI device 2 are bussed to busses 3 and 4 and then sent to the input 3 and 4 of the audio interface. MIDI device 3 will have to be bounced in a second pass and bussed to any of the 4 busses that will become available. Notice how all the inputs of the board are bussed to the main output so that we can listen to them while bouncing.
If the board doesnt have a buss system you will have to use the main stereo output to send the input from the MIDI devices to the inputs of the audio interface. In this case you have to remember to send only the channels of the board that you want to be recorded to the main (L-R) output. This process is longer since you will have to record them one by one.
In any case, remember to record the signal on the audio tracks as loud as possible but without distortion. In the digital domain distortion can be a disaster. In the analog world a little bit of distortion or saturation can create a nice warm and compressed effect, but in the digital domain as soon as the signal goes over the 0dBfs you will get nasty digital distortion that you wont be able to remove from the audio track. As a general rule, try to keep your recording meters between 12dBfs and 0dBfs.
When it comes to recording volume settings before transferring the MIDI tracks to audio, there are two approaches you can follow. If you have a fairly simple automation set up where you only used CC 7 and CC 10 to automate volume and pan, I recommend suspending the automation for the MIDI tracks and resetting all the MIDI volumes to 127 (maximum) in order to get the highest possible output from each MIDI device. You will have to recreate the automation for volume and pan later at the final mixing stage. If you, on the other hand, have created a complex automation for your sequence in order to recreate the natural attack and release of live instruments, I recommend leaving the automation On and deal with overall volume issues later at the mix stage.
Before starting the mix, take a few minutes to reorder the audio tracks, this will definitely speed up the process of mixing later. I recommend ordering the tracks by section: woodwinds, brass, strings, synthesizers, piano, guitar, bass and drums/percussions. Keeping a standardized order for the tracks is particularly useful over time since you will start getting used to it and it will get easier to orientate yourself in complicated mixing sessions. The order I suggest works well because it follows the standard order with which the sections are arranged in a score layout. By following this template, another engineer will be able to look at your sequence and find any instrument in a matter of seconds. After transferring the MIDI tracks to audio tracks make sure to keep the MIDI data, do not delete them. They will come handy if you want to go back to the original MIDI tracks and make some changes. Make sure you mute the MIDI tracks after each transfer and move them to the bottom of the track list. Keep in mind that depending on the sequencer you use, you might have limitation in terms of audio tracks or voices you can use. If the track count is limited you might have to create sub-mixes of certain instruments in order to be able to handle a high audio track count. Even though I highly recommend keeping as many separate audio tracks as possible, here is a list of instruments that can be grouped without limiting your options too greatly.
High woodwinds can be grouped together because of their similar sonorities and acoustic response to reverberation. In this category you can include clarinet, flutes, oboe, and English horn
Low woodwinds in general require less reverberation in order to maintain a clear and intelligible mix. In this category you can include bass clarinet, bassoon, and contrabassoon
The saxophone section can be set as a separate category mainly because of its distinctive phrasing and sonority even if it covers a wide frequency range
Trumpets constitute the higher range of the brass section, for this reason they usually require different settings in terms of reverb and equalization
Trombones, French Horns
Trombones and French horns cover the mid and low range of the brass section. In an ideal situation you want to keep the two separated in order to have a better control over the equalization of each sonority
Even though violins and viola feature substantial differences in terms of overall sonorities, the two cover the high and mid-high range of the string family and for this reason they can be treated together in a mix where track count is an issue
String Basses, Cellos
String basses and cellos cover the low and mid range of the string family. They usually require substantial different settings in terms of equalization, panning and reverb than the violins and violas
Synthesized leads can be grouped together unless their sonic features and purposes are completely different
Synthesized pads usually can be easily an effectively mixed together since they can share most of the panning and reverb settings
Piano, El. Piano, Keyboard Synthesizers
Preferably the acoustic piano would need to be on a separate stereo track
Make sure to separate the acoustic guitars from the electric guitar since they have different needs in terms of reverb and equalization
(See Ac. Guitar)
The bass drum needs to be separated from the rest of the drum kit mainly because of its need to be fairly dry (no reverb) and because of its peculiar equalization
Snare Drum, Toms
While in an ideal situation the snare would be separated from the toms you can sometimes have them on the same track since they usually both require a similar amount of reverb
HH, Cymbals, Shaker, Tambourine
These instruments can be assigned to the same groups because of their frequency range and reverb settings
The four main factors that you have to consider when making decisions in term of tracks grouping are: frequency range, reverb (and in a more general way overall effects), equalization and pan. The frequency range covered by each instrument and section plays an important role in grouping tracks. Usually, instruments that cover a similar frequency range share similar settings in terms of pan and reverb and therefore can be sub-mixed on the same stereo track. The frequency range has thus a direct impact on the reverb settings and panning of the parts. Usually instruments that cover the lower frequencies of the audible spectrum need little to no reverb since a high amount of reverb would cause the mix to be muddy and unclear. For the same reason, high frequencies usually can handle and need more reverb and therefore instruments that cover such area of the audible spectrum can be sub-mixed on a common stereo track. It is a bit more complicated to group several tracks based on their equalization needs though. Usually each instrument has peculiar characteristics regarding frequencies that are specific of its sonority. If you have to make a choice though try to keep instruments inside the same section separated by frequency range to which they belong (low, mid and high).
Once you have the audio tracks in order, and (if necessary) sub-mixed according to the criteria just listed, it is time to start getting an overall balance among the tracks. This is an important step. Before adding effects and working with equalization, create an overall good balance between tracks. Even though there is not a specific approach or technique for creating a draft mix, I recommend starting from the ground up, meaning from the foundation of the orchestration. Begin with the bass drums and then move to bass, then to rest of the drum kit followed by the piano and keys, guitars, strings, brass, woodwinds and then leads (which could include synthesizers, acoustic instruments or vocals). Every time you add an instrument go back and adjust the volumes of the other tracks, if you are doing things right you will probably have to apply small changes to the tracks that are playing back already.
In order to achieve a perfect balance among instruments and tracks keep in mind few important aspects. Each style and composition has a certain feel and overall sonority that needs to be achieved. Since you are the composer and producer of the project you have the advantage of knowing the material being mixed extremely well. Focus on the elements that are important in that particular project. Mixing these days is very much like a companion to orchestration. By raising the volume of a certain section or instrument you basically rewrite the dynamic marks that you had in mind while you were writing and sequencing that part. I like to think of mixing as conducting the orchestra by indicating crescendos and decrescendos, sforzando and pianissimo. After getting a decent rough mix, I recommend saving it as a snapshot or storing each level as automation data. This will speed up things later if you will have to program automation data for some of the tracks. When trying to get a good balance look for two or three elements of the mix that represent the most characteristic and important features of that particular project and work the other instruments and parts around it. This is a good approach to avoid ending up with a massive wall of sound where nothing is clearly distinguishable. Lead instruments and especially vocal tracks are a good starting point in terms of featured tracks. Always have two or three parts that are more exposed in comparison to the other background parts.
The way you position your instruments in the horizontal left and right axis has an impact not only on the stereo imaging of the virtual orchestra but also on the balance between sections and instruments. Even though panning can change drastically from project to project I would like to give you few recommendations that I find particularly useful. You will find that by changing the panning you will have to adjust the volumes of some of the tracks. Instruments that are panned hard left or hard right tend to be more exposed and separated than other instruments left more into a central position.
There are two main criteria that you should follow when it comes to panning instruments on the stereo image. The first one is balance and the second one is frequency placement. Even though balance may seem a pretty obvious concept, it is often forgotten, and if overlooked during the mix process it will have to be fixed at the mastering stage later. Balance is achieved by panning sections and instruments so that you reach constant equity between the left and right channels. You want to avoid having one of the two sides constantly more active and predominant than the other. Check the balance between the two channels by listening carefully to the mix and by watching the meters of the stereo master track of your sequencer. A good starting point to keep the balance between the two channels is to have a clear idea of the instrumentation featured in the project and on the importance that each instrument has inside the arrangement. Avoid extreme pan settings for lead vocal and instruments that are featured constantly in the piece unless similar parts such as a vocal counter-voice can counterbalance them or another lead instrument. For short and temporary featured solos or passages it is ok to have instruments hard panned especially if you can alternate sides between several featured solos (i.e. a short electric guitar solo panned left followed by a short saxophone solo panned right). Before deciding on the panning I recommend graphically sketching out the instrument featured in the project and their placement in order to come up with a sort of blue print of the mix). Avoid placing too many instruments in the center without taking advantage of the clarity and spatial openness offered by the stereo field. Pads, keyboards and strings offer a good starting point to open the stereo image by either recording the patches in stereo and pan hard left and right the stereo track or by carefully panning each instrument using more extreme settings.
The second principle that you have to keep in mind when working on panning and instrument placement is based on the frequencies featured by each part. The so-called frequency placement approach is based on the fact that low frequencies are harder to place in space than high frequencies. This is due to the fact that the brain perceives sounds in space according to the difference in phase of the waveforms received by the two ears. Since low frequencies have longer periods a small difference in phase is harder to perceive and therefore for the brain low frequencies are more difficult to place precisely in space. Thus usually it is more natural for the ears to listen to audio material that features low frequencies instruments placed in the middle of the stereo image rather than panned hard left or right. Following this approach you should avoid using extreme panning settings for instruments such as bass, bass drum, string basses and any other low frequencies based instrument.
Remember that rules can be broken if there is a good reason to. For example if the project you are working on features a virtual ensemble mimicking a live ensemble such as a jazz quartet or a classical orchestra or a string quartet then it is better to use the placement that the live ensemble would follow in a live performance setting. A typical example would be a jazz quartet composed by piano, bass, drums and saxophone. In this case you can place the instruments with more freedom by panning the piano slightly left, the drums in the center and the bass slightly right. The saxophone could be placed either in the middle or slightly to the right to counterbalance the piano. The same can be said for a string quartet, for example, where the cello (which is the instrument that covers the lower end of the spectrum in such ensemble) can be panned right as it would be in a live performance setting. As you can see panning, as all the other elements from sequencing to orchestration, can have a clear impact on how real your virtual ensemble can sound. Try to keep realistic settings also in terms of panning and try always to reproduce as much as possible a real live performance situation no matter which style or ensemble you are sequencing for. Keep in mind that you can always use creative and original solutions when it comes to panning and that rules can be broken for creative reasons.
While low frequencies sound usually more natural if panned in the center, high frequencies on the other hand can be panned at any degree ranging from center to hard left or right. An example is offered by the hi-hat and cymbals of the drum set. These instruments can open the stereo image if panned with more extreme settings. Crash and ride cymbals can be panned respectively hard left and hard right (or vice versa) as they would be found in an acoustic drum kit. The same can be said for shakers, tambourines, and high-pitched percussions in general. The final goal for a successful panning is to open up the stereo image of your mixes without creating unbalanced positioning of the instruments. By taking advantage of the entire stereo image and accurately planned and precise panning settings you can greatly improve the intelligibility and clarity of the production.
As you can see, bringing your mix from an agglomerate of different tracks to a cohesive and clean work or art can be a daunting and time consuming process, but it is also a very rewarding one. It takes several trials and errors before reaching the perfect mix (and believe me we are all still looking for it!), but you will be surprised how much your mixes can improve by applying some of these principles and techniques. Now stop reading and go back to your mixing board and good luck!