A Guide to Digital Editing and Audio Mastering

What Is Mastering?

The Mastering processing (the correct technical term is pre-mastering) is the production link between your finished stereo mixes and the CD glass master or cassette running master tape that are used to manufacture albums.

There are different levels of services that are available during this all important stage.

The Basics

After finishing your final mixes, you will receive a reel of tape or a DAT of individual stereo mixes.

There are several ways to combine individual mixes into an album sequence. Dat to Dat (ala double deck cassette) works but is a truly inelegant solution because the spaces between songs can never be done accurately and if the level balance between songs isn't correct, you either have to remix the offending song(s) or go through a mixing board (usually analog) to change it. Manufacturing plants will not accept DATs unless you authorise them to do the PQ coding (start IDs, etc) which they will usually do on a Sony 1630.

The Sony 1630, a digital recorder that uses video tapes, is an effective assembly machine and has been used by the recording industry for years. However, there are some problems associated with this method. The most important negative is this format's potential for uncorrectable errors because it is not a particularly hardy platform. 1630 transfers can sometimes degrade your masters and unless you pay for a reference CD, you will never know whether your master has been sonically changed until you receive your boxes of manufactured CDs. The Write Once CD (CD-R) is an excellent alternative to the 1630 and will be explained in a later article.

In the 90's, the most practical and cost effective way to master is on a hard drive editing system. Manufacturers such as Digidesign (with 3rd party "plug-ins") and Sonic Solutions for the Mac offer reasonably priced, sophisticated digital workstations that can be used for mastering. Outfitting a studio with this type of equipment is much less expensive than the gear of even five years ago. As a result, the price of digital mastering/editing studios range from $30. to $80. per hour versus $150. to $200. per hour at the traditional houses. My view is that the skill and experience of the mastering engineer and the acoustics of the studio are as important as the gear, so since the prices of equipment allow many more people into the editing and mastering game, it pays to screen carefully. Actually, it pays to screen carefully in any price range as I've also have had some bad mastering experiences at $200.00 an hour!

Computer based editing and mastering services include items such as change gain, mix, equalize, compress and sample rate conversion which are technically called digital signal processing or DSP. DSP is accomplished by performing mathematical calculations (called re-quantizing) on your digitized sound file. Please note that your sound file will be permanently changed when processed and not every digital workstation handles processing in a musically pleasing manner For example, some digital EQs sound edgy and unmusical and your music could end up sounding grainy and lifeless when compared to the original master. The reason is that DSP can't be done in a sonically pleasing manner at the CD standard of 16 bits and so has to done at a higher rate.

So what should you look for? Ask the studio if their workstation does all intermediate processing and storage at 20 to 24 bits or make sure that the software that you will be using (for EQ-ing, etc.) processes at 20 to 24 bits. Once your DSP is completed, the sound file has to be pulled back down to 16 bits. The particular functions you should be concerned about are dithering and noise shaping. Dither is a technique used to mask the noise created by re-quantizing when the 20 or 24 bit file is resolved down to 16 bits and noise-shaping techniques move the noise to a frequency that is less disturbing to the ear. Some manufacturer's noise-shaping techniques sound a lot better than others, such as the Waves products (L1, Q-10, etc.). If you would like more detailed information, there is an excellent article in the October '95 issue of Mix magazine entitled "The Secrets Of Dither" by Bob Katz.

For the balance of this article, I am going to assume that you have a DAT master tape as DAT is currently the most widely used mixing format. I will also presume that you are going to a mastering house with a hard drive workstation.

Before addressing your album sequence, the mastering engineer has to check your DAT to see what sampling frequency rate your songs were mixed at. Some engineers are still mixing at 48 kHz even though the CD standard is 44.1 kHz. DATs submitted to the manufacturer at any other sampling frequency will have to be converted at the plant. The sound will change upon conversion so its best to do it now, first in the mastering stage, before any other work is done. The only exception to this is if you know for sure that you are running your final sequence through an analog chain back to DAT. In this case, conversion on input to hard drive is unnecessary.

There are several reasons why mixes are still being done at 48 kHz, from lack of familiarity with the CD manufacturing process to the use of consumer DAT recorders or even DAT recorders that default to 48 kHz on start-up.

The sampling frequency of 48 kHz is a higher sampling rate and therefore actually sounds better than lower rates. The reason is that there are more samples, or "snapshots" of the analog waveform being taken at 48 kHz than 44.1 kHz. Therefore when we convert a higher sampling rate to a lower one, a small portion of the signal is lost or slightly changed. This is usually audible in the high end but sometimes in the quality of other parts of the frequency spectrum.

The mastering engineer will convert your 48 kHz mixes by using a digital converter or may use the conversion function in the editing program. A decent converter, such as the Alesis AI-1, does a good job but some software programs do not. In that case, it is better to simply go analog, i.e., record into the hard drive through the analog converters instead of going digital and crunching numbers after entry. Its worth taking the time to ask the mastering engineer to let you hear both methods -- the difference is sometimes very noticeable.

Obviously its better to mix at 44.1 kHz since you are working with the final tonal spectrum of the CD format instead of listening to frequencies that may or may not be there on your final DAT master.

After choosing the best mix of each song (or the best parts of 2 or 3 mixes to edit together into one song), the next step is to turn them into an album sequence that makes artistic sense. If you have access to a DAT player or double deck cassette player, making rough cassettes of the proposed sequence at home is OK. A better approach is to ask the mastering studio to make you cassette copies of 2 or 3 running orders to take home and listen to for a few days before making the final decision.

The next step is to adjust the relative volume levels of each song so that the whole lbum is comfortable to listen to as a sequence.

Individual studio mixes that are recorded digitally are usually done with meters hitting the zero point on the loudest peaks, even with ballads or slow songs. For example, playing a fast song and a slow song side by side, mixed to the same meter levels, usually will not sound balanced. Volume levels have to be adjusted by "ear" (within correct technical parameters), not meters in order to make a sequence sound balanced. Incidentally, most mastering engineer prefer to work with mixes that peak in the minus 3 dB range (below digital 0). This gives them enough "headroom" for overall or frequency specific level raising without having to lower the volume digitally and then raising it again before cutting the master.

Some of the mastering programs have an adjustable pre-roll function that make volume balancing easy. A flick of the mouse can instantly take you into the back end of a song according to how you set the pre-roll time function. I usually set a time of 50 to 60 seconds which almost always the hottest point any song. I then flip through all the songs to get a sense of how the sequence should be balanced. Depending on the mastering program, there are several different ways to adjust the volume balances, from simply lowering the level(s) of the hottest song(s) to using digital or analog compression. The key technical point to check is that the final master sequence is printed hot enough.

Sometimes a few or all of the songs have to have their levels raised digitally. If the levels average out on the low side, your CD will sound weak when compared to major label CDs. Adjustments can be made at the plant before making the glass master but most mastering engineers and record producers do not like having any extra processes done to their music without being there to approve the results.

Another item that should be checked at this point is the beginnings and endings of each song. There is always noise, or a noise floor before and after each song so each song file should have the unnecessary parts cut off. This is analogous to cutting an extra word from the front of a sentence on your word processor. This is now the time to check the fades to see if they are exactly what you intended. I am always surprised at how many mixes that are otherwise done well have sloppy fades. However, a good mastering engineer can fix these easily. In fact, because of the sophistication of the mastering programs, some producers don't bother doing fades at the mixdown stage anymore. They leave it to the mastering session where they can quickly (and non-destructively) experiment with different places and different lengths.

After the song order is chosen, the levels are balanced and the fronts and backs of each song are cleaned and re-faded (if necessary), the next step is to decide on the spacing between songs. If you analyze a few albums in your collection, you will notice that some have a set time between songs, i.e., 3 seconds, and others have the spacing set according to the fade, tempo or style of the song before and after. If you make a decision to go the artistic route, most computer mastering programs make variable track spacing technically an easy job to hear and change, again using the pre-roll time function.

The following are some of the more sophisticated mastering services, the ones that can sonically really polish your album and help take it to a level that will compare favourably to major releases.

If you have not witnessed some of the miracles that can happen in the mastering suite, it is important to understand that the final stereo mix is not necessarily final. Although all the individual elements of the multi-track recording have been blended into two tracks, every instrument still has its own fundamental frequency and overtones. If the mastering engineer can isolate the dominant frequency of a particular instrument that needs to be altered, that instrument's presence can be significantly boosted or cut in the stereo mix. With the flexibility of digital editing, the transformation of multiple mixes into one have reached a level that was just not possible with razor blade edits.

Editing Of Different Takes Into One Final Track

Although I briefly addressed this topic earlier, it is vitally important to understand the incredible depth and power of digital editing so that you don't feel you have to accept a less than perfect complete take.

When mixing, it is common to print two or more versions of each song to DAT. Generally, one mix is chosen as the "final". When listening at a later date, it is discovered that one of the alternate takes may have a better mixed section, i.e., the intro or instrumental solo, than the same section on the final. Digital editing allows the seamlessly combining of two or more takes into a new "final" track. A skilled digital editor can replace a bar, a beat, or even a small portion of a beat with laser-like precision.

Digital editing can be used to rearrange, lengthen or shorten a song by expanding or leaving out existing material. This is the most efficient way to create more than one version of a song, i.e., the "club mix" or "radio edit".

If you bring good documentation to the editing session, each individual edit should only take from 2 to 15 minutes, depending on the complexity of the idea. All of the above are done digitally, without causing any audible program degradation (if using a professional system with 20 bit or higher resolution) and without touching your original master. A good way to edit is to use a system that allows you to "capture" the areas of the song that you want to piece together. The "regions" are then put in a list and are seamlessly strung together (with cross-fades if nessary). This is called "playlist" editing. When chosing to work directly on the song file, most digital editing systems allow the last move to be "undone" so virtually any idea can be auditioned without making a commitment until fully satisfied. Some systems are considerably faster and easier to work with than others, so if you have to do a lot of editing, do some research. The same evaluation should be applied to the engineer.

Adjusting The Levels of Different Sections Of A Track

Sometimes a mix can be improved by simply adjusting the volume level of a particular section of a track.

The following are representative of common problems that we are asked to fix:

1. A 4 bar acoustic guitar intro seemed to be too soft in relation to the full band entry on 5th bar.
2. During a long, smooth fade in, the client decided that the the 1st entry of the kick and snare need to be louder.
3. A drum solo in the middle of a song was too soft.
   During an intro, an single acoustic guitar arpeggio was too loud.
   The above problems were easily fixed by digitally raising or lowering the volume level of the section in question. Even though other elements of the track may be affected by the global level change, these types of adjustments are usually successful because the dominant problem has been addressed.

Equalizing

With parametric equalization (EQ), the mastering engineer can often isolate the dominant frequency of a particular instrument or voice so that instrument's presence can be significantly boosted or cut in the stereo mix. Any section or the entire track may be worked on to make various elements louder or softer, richer or thinner. The more sophisticated digital EQ programs are so precise that often the tone or the level of a kick drum can be changed without affecting the level of the bass guitar.

The following are typical problems that we recently encountered:

1. A 10 track album was recorded over a period of 2 years, in 3 different studios with 4 different producers. Obviously, when this album was sequenced, the songs weren't going to match sonically. Solution: we found one song that we felt had the best overall mix. We then adjusted the frequencies that needed work and digitally created a new, EQ-ed version. The new version was used as a "template" or reference and all the other songs were adjusted until they sounded like they belonged on the same album.
2. An album that was generally well and uniformly mixed had 1 song where the vocals were too loud in relation to the vocals on the rest of the album. Solution: We found the frequency of the vocal and lowered it. Of course there are other elements of the mix that are in the same range, such as guitars. However in this case, it was possible to find the right combination of frequency and bandwidth, and although some of the other elements of the track were also adjusted, the overall goal was satisfactorily achieved.
3. An album was recorded in a studio where the noise floor was a little too loud. When ever there was a section with solo bass (i.e. a pick-up) or solo guitar, your attention was drawn to the hum and hiss in the background. Solution: We found the frequencies of the hiss and hum and reduced them just enough to keep your attention on the music but not enough so that the sound of the program was drastically changed. EQ-ing is a good way to reduce noise but even better are dedicated outboard gear and software programs. If your masters are particularly noisy, make sure the studio has the equipment to deal with it. DINR by Digidesign is the most common software noise elimination program and is excellent.
4. An album was mixed in a studio where the acoustics over emphasized the bottom end of the frequency spectrum. The mixing engineer wasn't aware of the studio's deficiency and lowered the bass frequencies during the mix to compensate for this and as a result, when the album was played outside that studio, it didn't have enough bass or kick drum. Solution: Find and raise the relevant frequencies. Digital EQ is so precise that sometimes the level of a kick drum can be changed without affecting the level of the bass guitar.

When researching mastering studios, inquire as to what gear they use to EQ their files. Some programs, such as Digidesign's Sound Designer, have decent built in EQ and other functions but there are superior and relatively expensive "plug-ins" that can be added. For example, the Q10 by Waves, is a 10 band equalizer that is better sounding and more versatile that the one that comes with Sound Designer so if you anticipate a substantial amount of EQ work, make sure that the mastering suite has the tools to deal with your problems. Once again, the engineer's "ears" and experience are an equal consideration.

De-essing, or reduction of sibilance in the vocal track of a mixed master is done by using a dedicated outboard unit or software programs. If there are only a few occurrences in a song, we get good results by finding the offending waveforms and simply reducing their volume level.

Compression and Limiting

An important used of compression is raising the overall loudness of a track when simple volume adjusting (as explained earlier) isn't enough. Rather than simply raising or lowering the volume, compression works by reducing the loudest peaks in the music and raising the lowest ones. The net result is a raise in either the "apparent" volume level or the actual volume level. This process is done, to the extreme, with most television commercials. This is why they always seem to be louder than the soundtrack of the program you are viewing. Singles mastered for radio are usually compressed more than album tracks, for the same reason. Additionally, radio stations run their programs through compressors so if your tracks are already compressed, they will be less affected by the station's gear.

Sometimes a song has a peak(s) that does not actually sound louder than the average wave form but prevents the raising of the overall volume of the track (called normalizing). In this case, in order to raise the level of the track, the mastering engineer has to locate the major peaks on each song, reduce their volume and then normalize the file. The better way is to use a "brick-wall" limiter such as the L1 by Waves when adjusting the final volume levels. This program looks ahead for the peaks and automatically lowers them while raising the rest of the waveforms as close to digital 0 as you want. We run all of our masters through this popular plug-in program. The IDR (increased digital resolution) feature adds a sonic depth and polish that subtly makes almost every album sound better. This is a particularly impressive piece of software and so far nobody mastering at our studio has said no to "L1-ing" their album! The net result of using a good compressor/limiter is that your CD will end up louder than most other indie CDs while retaining control of how much dynamic range (loud to soft) you want to sacrifice. If your album has not been run through a compressor/limiter like the L1, it will sound weak when played next to a major label product.

Cross-Fades

Occasionally, a client wants one song to be fading out as the next one fades in. Most programs have an effective way to create a variety of cross-fades.

Additional Adjustments

Sometimes is it beneficial to add a light reverb program to an entire track or tracks. This is done in cases where a dry track(s) stands out unfavourably when compared to a majority of more reverberant tracks. Reverb is sometimes added to an entire album to give it a "sheen" or polish that was lacking in the original mixes.

Occasionally we receive mixes where the level is too low on one side. If the difference is audible, it is best to balance the two sides. Almost all programs can deal with this problem but some take much longer than others to audition the adjustments.

Although not a common occurrence, sometimes an important element of a song is left out of the final mix and not missed until the mixing sessions are over. An example of this would be a percussion part, background vocal or additional guitar part. If there is no budget to remix, the following technique works well if your chosen mastering studio has a multi-track hard drive recorder. Record the mixed track to multi-track tape and then record the missing part. When satisfied with the part, record it to one side of a DAT tape and a guide mono mix of the original mix to the other track. The mastering engineer will record the new part and the replacement track simultaneously into the editor and will achieve synch by matching the wave forms of the mono track to the original mix. When they are matched, the other track will automatically be in synch and can then be blended into the original mix.

By using a multi-track hard drive recorder, sound effects and voice-overs can be added to an already mixed stereo track. In this case, the added tracks are literally pulled or slipped into the correct place. When satisfied with the placement and blend, the four tracks are digitally mixed to stereo with no audible loss in fidelity.

While slightly out of the scope of a mastering article, pitch and rhythmic correction of individual tracks are not uncommon and can be done with digital editing gear. The following are representative of problems we have been asked to fix.

1. We got a great vocal performance but there's a few notes out of tune/out of time and the singer will a) never recreate that attitude or b) has left town.
2. The kick drum is not in time in a few important places and we didn't notice until after the drummer was gone.

In both of these cases, the hard drive recorder is synched up to the multi-track and the track needing work was transferred. Using the editing program, the kick drum, vocal phrases or individual words can be easily moved around. Because the hard drive, is "locked" to the multi-track, the placement can be fixed and recorded back to the multi-track before mixing. Out of tune vocals are a little more difficult to fix, but if the editing program has a good pitch correction function, this can usually be done successfully. I sometime find it more natural sounding to simple copy a word or phrase that was sung better in some other part of the song and paste it over the out of tune version. The latter method also works sometimes with a stereo mix, i.e., copy the whole beat, bar or section that has the good performance and paste it over the bad one. Of course, care has to be taken not to destroy the rhythm of the overall performance!

Analog vs Digital

Sometimes harsh or brittle sounding mixes can be improved by running the program through a warm sounding mixing board and an analog compressor. Further, some engineers like to use tube gear for this purpose although leaving the digital domain just to use a tube compressor should be carefully considered. Remember that your music has been mixed to a digital audio tape (DAT) and that normally the music will be transfered to the hard drive without any changes to the 16 bit resolution. So if the mastering engineer runs it through a tube compressor just because it seems to be the popular move these days, he could be weakening one aspect of the mix (re-quantizing your sound file) while adding the tube sound. Mastering engineers should look at each album on a case by case basis to decide whether the overall sound of the project would be enhanced by leaving the digital domain. We have no hesitation in running masters through an analog chain if the overall result can be noticeably improved.

Backing Up Your Album Sequence

Another major benefit of hard drive editing systems is the ability to back-up your work. Most programs allow precision back-ups to DAT or modular digital multi-tracks. This allows exact restoration of all files so you have the option of making changes later, without having to do any of the original prep work. Fine tuning of the album sequence can start exactly from where you left off from the previous session.

Resist the urge to save money by not making a back-up of your running master DAT. The old studio adage is that if you don't make a safety copy, your master will get lost or damaged.

CD-Rs vs the Sony 1630

Once your album sequence is finished, there is still a process that must be done before the CD glass master can be made. Glass masters can be manufactured from either a "write once CD" (CD-R) or by running a master DAT and converting it to a Sony 1630 format.

Earlier article I mentioned that the 1630 format has the potential for uncorrectable errors because it is not a particularly hardy platform and because of that and other reasons, using the 1630 can sometimes degrade your masters. All veteran producers have had the experience of sending out a master to the manufacturer and later wondered what went wrong after hearing the final product! The 1630 is most often the culprit!

The CD-R technology is still relatively new and there are two methods for cutting CD-Rs. Additionally, there are really only a couple of CD-R issues you need to be concerned about.

One genre of CD recorder uses your sequenced and mastered DAT as its digital audio source. This method is sonically OK but is not precise in the manner in which it places the track numbers. If you have have cross-fades or even one song that butts up to another, this method is not recommended. Also, you may have noticed on major label CDs that when a track is finished and there is a four second space between song, there will be a reverse count-down showing on the CD counter to represent the space. The CD-R recorder mentioned above cannot account for the space in the same manner and will add any time between tracks to the total length of the song.

The easier and more professional way to cut CD-Rs is using a CD recorder that gets its audio from the computer's hard drive. In a studio that uses a digital workstation, the songs are already stored in the hard drive and so it is relatively easy to create a list of all the songs, with the track numbers in exactly the right place, and cut a CD-R directly from the computer. An additional bonus is that most CD recorders of this type can make CDs at double speed.

The other issues are well beyond the scope of this article but if you are interested in an ongoing debate, check out Stephen St. Croix's series of articles in Mix Magazine starting in October, '94 and ending March, '95.

I prefer CD-Rs to the 1630 because the results are much more consistent.

Since we started mastering at Silverbirch, most of our masters were cut on CD-Rs and sent to the two major plants in Canada and two different plants in the U.S. I don't know for certain how each plant dealt with the transfer but I have generally observed very little difference in the sound when comparing the DAT master to the CD-R. The same holds true for the occasional opportunity that I have had to compare the running DAT master, the CD-R, and the manufactured CD. With the 1630 transfer, you can never be certain what you are going to get. Incidentally, the 1630 transfer fee is usually quite a bit more expensive than having a CD-R cut.

Finally, since CD-Rs will play in most home CD players, perhaps the biggest bonus is that the producer or artist can take it home and check it to make sure quality control has been maintained and track numbering has done properly before sending out for manufacture!