Background FAQ: Making better mixes, tips on compression, EQ, monitors, CDRs.

• Core FAQ covers analog and digital mastering, DRT's custom gear, sample tracks and pricing.
• Background FAQ and Letters offer tips for making better mixes: compression, EQ, CDR issues.
• Custom Service is a quick summary for people who target the Major-label market.
• Tech News expands on selected technical topics. 
• One-Minute Tour is where most people start. 

• What is CD mastering?
• Why not just transfer my mixes to CDR?
• Tubes vs. Solid state electronics
• Digital or analog processing: More on quality
• Quality issues with CDR media
• Other audio mastering tools
• Should I mix to analog or digital tape?
• Burning your own CDRs: Tips and troubleshooting (also published in EQ magazine)
• A brief note about monitors
• The 1610/1630 Umatic tape
• My CD will be the loudest ever made. OK?
• Other options for mastering your material
• Digital editing vs. crash editing
• Choices for mastering your own mixes
• When not to master at DRT
• You need a great mix to get a great master
• Note to recording musicians: Tradeoffs
• Seven steps of the CD production process
• Short run CD replication
• Creating tighter mixes (also published in EQ magazine)
  
• More on improving the final product: Equalizer technique, TC Finalizer etc.
• A short non-technical perspective
• Thanks
• Replication info plus links to other (mostly music and audio) sites
• What's this DRT character look like, anyway. (It's not pretty, folks. Better spend your time on the text.) The insatiably curious can click here.
  
• Elektor letter about clipping on CDs
  

What is mastering?

CD mastering has a reputation as a mysterious art known only to a musical high priesthood. In fact, the right mastering can make a good recording sound excellent, and possibly turn a great one into a legend. Most discs can be made competitive for radio and in-store play.

To use an analogy - Every album has a "voice" in which the message of the artist is delivered. A strong performance and good recording technique will set the basic tone for this voice. Mastering can then profoundly affect it's impact and resonance. How? My guideline is: do what serves the music. A wide range of techniques can bring out an album's native voice. Depth, punch, sense of air and detail can all be enhanced. Vintage tube processors may work, or the latest technology could be appropriate. The solution sometimes goes against logic. Experience and a feel for the music determine the best path.

The process: Most people are familiar with the idea of recording music in a live concert or recording studio. You make tapes that store the individual performances, or takes. Ultimately these takes are assembled into a final master tape. This is sent to a replication plant, where copies are made. The process of creating the final master is called mastering. It has three steps:

1. Assembly editing: The tapes from the location recording or mixdown sessions are transferred to a digital editor. The tunes are sequenced into the order you specify and correct spacing is made between cuts. Beginnings and ends of cuts are faded to black (silence) or room tone (the natural background noise of the performing space), or the cuts are crossfaded as you wish. Pops, clicks and strange noises can often be fixed at the mastering stage, depending on their source and severity.
2. Sweetening: When engineers first began cutting master discs used to produce vinyl records, they designed signal processors such as compressors, limiters, and equalizers (EQ) to prevent overloading the cutter head. They noticed that changing the settings could also have beneficial effects on the music, especially in Pop styles. Equipment and techniques were developed to further "sweeten" the sound. Since then, this has been considered the heart of the process, where clarity, smoothness, impact and "punch" are enhanced, depending on the needs of the music. The goal is to increase the emotional intensity. If the performance, arrangement and recording quality are good to start with, then the final master sounds even better than the mix tapes, and even casual listeners can notice the difference. This is what albums need in the pop markets: big, radio-ready sound and a competitive edge.
3. Output: The finished music tracks are transferred to the media needed for mass production, usually CDR (Recordable CD). This master disc can be played on any CD player, so the client can audition it and give final approval. Occasionally cassette replicators prefer 4mm DAT tapes.

Note - To CD replicators, mastering means creating the glass master disc that is used to make stampers (which are then used to press the CDs). Almost every CD plant prefers to make it's own glass masters, for quality control reasons. The glass master should be a perfect mirror image of the CDR master disc produced by the music mastering facility. In this brochure, I use the classic definition of mastering: The final creative step in producing a CD.

Why not just transfer my mixes to CDR?

If your mixing engineer is one of the rare few who can mix and master at the same time, then a straight transfer might do fine. But there are at least three reasons why professionals send tapes out for mastering:

1. Usually a lot can be done to improve the mixes. The market is demanding. If you want the disc to compete in the radio markets, in-store play, and the homes of consumers accustomed to excellent music, it has to be right sonically. It's like very expensive shoes - without a good polish, few people will appreciate how great they are. Also, since the mixes were recorded at different times of day over a week or more, you end up with differences in level and tone. Mastering creates a seamless whole out of a collection of individual tracks.
2. The mastering engineer has fresh, experienced ears. By the time the mixes are done, everyone involved is fried. It's tough to keep perspective after you've heard a mix 50 times. A new outlook, a new set of listening skills - attuned to the complete presentation rather than the detail of the mixes - can make a huge difference.

   Mastering engineers must be fluent in both the artistic and technical areas of music-making. Good communications skills are also critical, since a lot of terms are used to describe sound. For example: "It needs to sound big, resonant, fat, warm, ambient, taut, sweet, present, smooth and live. It needs air, sparkle, depth, brilliance, impact, punch, focus, clarity and definition." The goal is to understand and accept the producer's guidance and then add or remove only what the music requires, not a bit more. It sounds simple... (Was it Michelangelo who referred to carving statues as "getting rid of the unnecessary stone"?) Often the changes are subtle, but the response from the producer or label exec should always be "That's it! Sounds like a record!"

3. The mastering facility has ultra-clean processors that are built to handle stereo signals. This may be obvious - it is one thing to run a guitar through a limiter and equalizer, and another thing to run your whole mix through it. A finished mix is a complex balance that can be made worse as easily as it can be improved. Its worth using the best equipment available.

Tube vs. solid state electronics

No one will ever settle this question, since both solutions have advantages in different areas. Either tubes or transistors can be truly excellent when the circuit is done carefully with the highest quality components available.

The best solid state circuits come close to disappearing. They tend to err in the direction of a faster, more defined, slightly less forgiving sound. The best tube designs approach perfection from the other side, with a sense of transparency and air. People consistently use the term "warmth" when describing tube sound. Tube circuits offer a wider array of options if you are trying to improve the sound, rather than just pass the signal through with no changes.

Digital designs approach the standard set by the analog solid state, but are less forgiving. A harder, closed top end feel often results from the limits set by current technology. Sense of air and soundstaging are usually somewhat compromised.

Digital & Analog processing: More on quality

1. Assume for a moment that you have a great-sounding two channel mix. Assume also that you have that mix stored to your satisfaction, either on DAT or a higher-resolution 24-bit format.

2.You need to edit the mixes into a sequenced tape, usually called a submaster or work tape. Many digital editors will do a good job if they are used correctly. This means for straight assembly editing - sequencing the tracks, crossfading between takes and cleaning up the beginnings and ends of cuts. Notice I did not say sweetening. (Ominous pause...)

3. Sweetening takes a lot of computational horsepower, using at least 24 bits of data, to maintain the integrity of the music. Many digital processing devices, including the ones built into some well-known digital editing workstations, just don't preserve enough accuracy in their computations. To be fair, many PC's and Macs are equipped with 24-bit processing cards. They should sound good, or at least neutral, but a significant number of people say the resulting master lacks the raw energy that was there in the early mixdowns. They ask what can be done to fix it. The summary? Consider using an editor to assemble your tracks, and leave the sweetening to devices that have been proven over time to be more friendly to the music.

So what are the options for processing? Some digital processors can sound excellent. Still, the better analog processors haven't been replicated digitally. This includes a number of custom tools that I use because they make the music sound right. Signal processors affect a lot of things in the music, and music is inherently an analog process. The best analog gear has a sound that a digital device cannot fully model.

To complicate matters, if a producer prefers the sound of top-shelf tube equipment, we have to go analog. While there are digital processors and software plug-ins with "tube sound" presets, its an open question as to whether they have done it right yet. Again, any two excellent engineers may disagree on this point. In the meantime, I provide actual tube processing for clients who request it, rather than a digital facsimile. All this brings us to the next point....

4. We may have to go back to analog, through some high-resolution equipment, and back again to digital to get the sounds that people want on their albums. You have added a conversion process to the signal, and theory says that the music will suffer. Technically, this is accurate. In the real world, the benefits usually outweigh the drawbacks. Test instruments will tell you one thing, and your ears may tell you another. Please remember, we are not talking about the average studio compressor or EQ, but specialized low distortion, low noise, wide bandwidth devices designed for mastering.

Also, this assumes that you use the best 24-bit outboard converters, which are capable of performance significantly higher than anything that can be stored on a CD or DAT. And when analog processing is complete, your high resolution music signal needs to be correctly dithered back to 16 bits. Dither could take up several pages of discussion. As you convert your 24-bit signal to 16 bits for CD storage, you can just truncate the data beyond 16, but quality will suffer. The better approach is to add a small amount of random noise called dither to the signal as it is converted to 16 bits. If the frequency spectrum of the dither noise is shaped so that it lies in areas where the human ear is less sensitive, the result can be a CD with an apparent dynamic range of more than 16 bits. Some techniques for dithering work better than others. Sony Super Bit Mapping and Apogee UV22 are just two of the solutions available. The topic of which unit sounds best gets technical very quickly. Every engineer has their opinion.

(Technical perspective & reality check: the average wideband RMS noise floor of mix tapes I receive is 65 - 70dB down, with variations of up to 10dB very common. This means that console noise, open midi and guitar amp channels etc. are providing dither signals well above the 16 bit noise floor, and the converters in your DAT machine may not be as terrible a compromise as you would think. 24-bit converters are worth it, but its not the end of the world if all you have is a DAT with good 16-bit converters. Make music! Let the guys in the white coats crunch the numbers.)

Every converter has a unique sound, so I keep a few proven 16-bit units on hand as well as the 24 bit hotrods. As an added measure when going to analog, I can re-clock the data stream, effectively eliminating digital jitter. Jitter is an interesting feature of digital audio that can mess up the sound of otherwise good music. The device I use has a custom manufactured ultra-stable crystal, and it can make the sound going into the analog processors actually better than the sound you heard on your mixdown tapes. The October 1993 issue of EQ Magazine has an article I wrote on the subject, if you would like more details.

So to recap, if the mixes are already stored digitally and you have digital sweetening tools that deliver the sound you want, that's great. Sometimes the detailed sound that digital is known for is perfect. It depends on your style. If you prefer the best that analog has to offer, then convert only once, carefully, since extra conversions can remove "live" feel. The most important thing is to do what serves the music.

Quality issues with CDR media

As CDR replaces Umatic tape as the common format for shipping CD masters, many mastering houses use digital editors that write the final CDR at higher than real-time speeds. This saves time for the engineer and allows the facility to generate more product. 2x speeds are common and the trend is to 4x and beyond. This practice is being mirrored at some replication plants, where glass masters are cut at faster than real time rates to maximize throughput. I have heard a number of top people discuss sound differences between CDRs recorded in real time and at higher speeds.

CDRs don't play as reliably as manufactured discs, especially on changers, car stereos and boom boxes. People have noticed variations in level during playback or crackling sounds in very quiet passages of the music, especially on high-speed copies. As manufacturers of players cut costs, the problem may even be getting worse - there's equipment out there that will barely handle a production CD. CDRs can tax this type of gear to it's limits, and strange things result. Dirty laser pickups can also aggravate the problem. (When was the last time you cleaned the pickup on your player?) So.... the bad news is that a CDR might not sound perfect on all CD players. The good news is this has never caused a problem on a master that I've produced for replication at a CD plant. If the master that I send to you for final approval sounds fine, then the plant will be able to produce discs that sound the same. The best test for a CDR disc is to play it on a high quality standalone CD player. If you want to make speaker comparisons, substitute small speakers or cheap headphones for the better speakers hooked up to the amplifier part of that system. The CDR will read accurately, and the variable will be speakers.

One possible explanation for sonic differences reminds me of the '70s trend towards mastering vinyl at half speed. When discs are mastered in real time, the media has more relative time to respond to the cutting laser. This results in a pit geometry that is more precise. When the disc is played back to cut the glass master, there is less chance for jitter and other timing-related problems to occur. A second factor is more measurable: as you increase cutting speeds, error rates sometimes go up, even though discs spinning at higher speeds are considered to be more stable. The built-in correction on the playback side should compensate for any increased errors, but not everyone is convinced. I take the conservative approach and create all CDRs at low speed, using media certified for 8x or higher cutting.

Final note: Some 80 minute discs are not as reliable to read or write as the standard 74 minute CDRs are. Consider using the 74 minute discs for your premasters.

Other mastering tools

Signal processors are like tools in a carpenter's toolbox. Spectral processors or dynamic equalizers can pull instruments out of a mix or de-ess a vocal. Exciters can boost harmonics, at the cost of making the top end grainy. There are better ways to enhance highs. I can also create surround-sound feels, using analog or digital techniques.

Vintage tube gear can "warm up" tracks, counteracting some of the effects of digital storage chains. Since vintage gear is often too noisy for mastering, I first correct the circuit and component flaws that were common 40 years ago. This keeps the clarity and the fat sound, while bringing the noise floor closer to modern standards.

When digital is appropriate, I use 24-bit hardware processors by TC Electronics, ATT/Wolfrum and others. These offer the standard suite of dynamics control, EQ and reverb. I've developed some custom DSP algorithms useful for subtle effects. I also have a full range of 24-bit software based processors, but tend to use them only where they are clearly superior - noise reduction for example.

Once you have the right tools, its a matter of deciding which ones to use. Certain combinations can make tracks sound "bigger." Which ones and in what combinations exactly? Tough to say in advance. You actually have to work with a given track. This is where mastering becomes more art than science. When you select processors, as when you select converters, the trick is to provide what the music needs, and no more.

Should I mix to analog or digital?

Given a choice, mastering facilities prefer to work with source material that is higher in resolution than the (16-bit) target media. There are two commonly available options that meet this spec: 24-bit digital files and analog tape.

Many studios now have computers equipped with CD burners, and 24-bit sound cards are available for a few hundred dollars. This makes it possible to mix directly to computer, and then burn a CDROM disc containing 24-bit stereo music files. This is the least expensive way to move your tracks into the higher-quality world of 24 bits. Common multitrack systems such as Paris, Protools, Radar, and VS1680 can also write 24-bit files.

Among 16-bit mixdown formats, DAT is the most familiar. It is convenient and sonically equal to the CD. Its worth examining the differences between this standard medium and analog tape, especially 1/2 inch stereo:

1. Dynamic range is roughly equivalent, though measured signal to noise ratio favors DAT. If you use noise reduction such as Dolby SR, analog is superior. Most real-world pop and jazz music has a dynamic range considerably less than 85 dB, so noise is not often the issue. In fact, the noisier 1/4" stereo tape is just fine for many styles of music. Most people feel that 1/2" tape preserves more detail in low level signals. When you go lower than 16 bits of signal in a digital system, the sound is gone. (A point in favor of 24-bit...) This is not true with analog, where you can still track musical information below the noise floor.
2. Measured distortion favors digital, by a factor of at least ten. To many this means cleaner sound overall. The flip side is that as you record hotter signals onto analog tape, you get a natural compressor/limiter action as the tape begins to saturate. This can be a big plus for certain styles of pop, rock and rap music.
3. DAT has a flatter frequency response within the normal 20 Hz to 20 kHz band. Above 20 kHz, response drops like a stone due to the very sharp anti-aliasing filters needed for digital. These filters can also cause strange things to happen to the high end due to their impulse and phase response. Many 15 ips analog machines have useable response from below 20 Hz out to almost 30 kHz. At 30 ips, finer machines go from near 20 Hz to way past 40 kHz, providing the extra octave of high end as well as a smoother rolloff. This contributes to the quality of "air" that people associate with analog tape. The noise spectrum is also different at 30 ips. Its still there, but has a "silkier" quality. All analog machines have slight irregularities in lower bass response caused by "head bumps". These bumps, related to the design of the playback head, are usually 2dB or less in size and do not cause phase response problems. They are often compensated for naturally when mixing.

The verdict? DAT is generally considered better than 1/4" tape at 15 ips, though if you want the fat in-your-face sound of tape, analog has the advantage, especially if you have good noise reduction. If you have a 1/4" machine available, run it in parallel with the DAT for a couple of mixes, then sit back and listen to both. I have clients that prefer the analog, especially if its running at 30 ips.

If you compare DAT to 1/2" analog, the picture changes. Even when expensive outboard A/D converters are used, most folks choose the analog. My opinion is that 1/2" is the format of choice for most styles of Pop and Rock music. Many hit records are being made in this format. Once you hear it, you'll know why. While its more trouble and expense to work with and the media costs more, the results are worth it. Definitely check it out if you have the option. If you decide to mix to analog, I will master the tapes on machines that have custom electronics. (I also maintain a complete machine shop, so there are custom aspects to the tape transport as well, but that's another story...) This equipment can probably extract sound from the tape that surpasses what you heard while mixing.

If you can create and store 24-bit mixes, the comparison with 1/2" is closer. Many people still prefer analog, but 24-bit can also be excellent, with a greater sense of air than DAT, and enough resolution to make even long reverb tails sound convincing. 24-bit storage on inexpensive CDROM discs has a lot of appeal for studios who don't want to buy and maintain an older 1/2" deck. An increasing part of my business arrives in this format. Compare them side by side if you have the chance.

Burning CDRs from computer: Tips & Troubleshooting

This section was published in the March, 2000 issue of EQ Magazine.

Many studios now mix to hard disk, storing the music with 24-bit resolution. They want greater-than-CD quality without proprietary tape formats or bit-splitting, so they burn their hi-res mixes to CDROM as sound files. I often see albums arrive for mastering on two or even four recordable discs.

Everyone looks for ways to save time. When there are two or more discs to make, increasing the the burn speed is tempting, providing that reliability doesn't drop. Over the years, I developed a list of suggestions for creating cleaner burns. This checklist goes from the basic to the more technical, but it might help improve the quality of your final product. Many suggestions came from technical people at companies that produce burners, bus controllers, and disc writing software.

Three main rules: (broken by many - at their own peril)

• Don't let the bottom of the disc touch anything other than a CD player or the case it came in. I'm amazed at the scratched-up, greasy-fingerprinted premasters that I've seen. Handle CDRs like vinyl records - edges only.
• Use quality media. In many burners, silver or gold media performs better than dark green. Asian clone brands can cost well under $1.00 per disc, but for 25 to 50 cents more, you can buy the best blanks on the planet. Is your music worth the difference? Even if your burner can sense the medium and recalibrate the laser to match, that's still no excuse to use inferior discs. My favorite? Taiyo Yuden. (Call APDC. 800-522-2732)
• Paper labels are fine on the copies you make for friends. Leave them off your premaster, because they increase the error rate! You may not hear any degradation on your player, but low error rates are critical to premasters. Label the CDR with magic marker, preferably outside the area where data is written on the bottom of the disc.

• Make sure you have enough memory. Many recording/editing/burning programs are specified to work with 16 or 32mB of RAM. 64mB is a more realistic minimum. More is better.
• 80 minute discs are not as reliable to read or write as the standard 74 minute CDRs are.
• Make sure that at least 10% of the capacity is always free on the hard drive that you use for storing mixes. The drive has to work much harder if it has to pack data into small left-over sections of disk space.
• Defragment your disk drives often. I use two large "working" drives, and eight others for archival storage. When a working drive fills up with mastered projects, I transfer the projects to an archive drive, then I erase the working drive and defragment it. (This takes only a couple of seconds.) Now the data can flow onto and off of the drive in large contiguous blocks, and the chances of interruptions in the data stream decrease.
• Log off your network. If there is any chance that someone else can access your machine through a Local Area Network, make sure to disable that option. One manufacturer goes so far as to recommend unplugging the network cable and rebooting, so that the network card is not accessing the data bus looking for network traffic.
• If you have a SCSI CD writer, make sure it is terminated properly. The manufacturer's documentation should have clear instructions for checking this. Many controllers can terminate the controller end of the SCSI cable automatically, but its always worth checking.
• While you are at it, check that you have the latest versions of drivers for the SCSI controller, and firmware for the SCSI CD writer. These can usually be found at the manufacturer's web site, or you can call their tech support line.
• Burn CDR discs from a hard disk image, rather than on the fly. Most CD burning programs allow you to adjust the levels of individual tracks, dither the fades and crossfades, then burn the CD immediately. This is convenient, but it demands that you have enough CPU and memory capacity to do the job in real time, while the CD is being burned. If you experience problems, select the option of creating a CD image. This intermediate step will perform all the computation for levels and fades off-line, then write the image to hard disk. (You will need enough spare hard disk space to store all the data that will be written to the final CDR.) As the final step, burn that hard disk image to CDR. The computer now has less work to do, and music data will flow more smoothly to the CDR drive.
• This principle also applies to making copies of CDs. Reliability will increase if you read the source CD onto hard disk as an image, then burn the image to CDR in one pass. The hard drive supplying the music data is much faster than the fastest CD reader, so error rates will drop.
• Windows only: (thought the principles apply to Macs)
  • Log off the Internet. Don't multitask. Close all applications other than the software that burns the disc. To check your situation, try the "three fingered salute," (CTRL-ALT-DELETE). This will display all the programs that are using resources. Anything other than EXPLORER and SYSTRAY may be unnecessary to the burn. Consider removing schedulers, alarm clocks, pop-up messagers and resource monitors such as Norton Utilities. These can unexpectedly take control of the system data bus, creating a glitch on the CDR.
  • The "Sleep" and "Screensave" modes on some PCs can be quirky. If you burn a long CD, some machines will try to enter standby mode after a predetermined length of time. This shouldn't happen at all, but is worth a look if you are tracking an elusive problem. This type of problem can also happen when older ISA-standard SCSI controllers are mated with newer motherboards containing only one ISA slot. PCI bus controllers have fewer problems. Try disabling the Sleep and Screensaver options, both in BIOS settings and in Windows itself.
  • Configure your machine as a network file server. Go to My Computer/Control Panel/System/Performance/File System. Most PCs work fine when set to the normal DESKTOP COMPUTER. If you are experiencing problems, choose NETWORK FILE SERVER. This gives priority to disk activity over screen activity. You might notice an occasional hesitation when the screen redraws, but that is a good tradeoff - the computer is making sure that music data is flowing smoothly.
  • Disable auto insert notification. Go to My Computer/Control Panel/System/Device Manager. Double click the CDROM icon. Highlight the CDROM that is installed in your computer and click the SETTINGS tab. Uncheck the AUTO INSERT NOTIFICATION box. Now the machine will stop needlessly interrogating the CD drive every two seconds, wondering if the disc was changed.
  • Many tech support people advise that Windows 95 is more stable than Windows 98. I've found that Win95 rev. C, the last version, works well and is the least likely to produce the infamous Blue Screen of Death. Win98 brought with it a whole new set of ...ummmmm... features that many prefer to do without when music is the primary task of the computer.
• Last resort: Try slowing down the burn speed! If your drive is rated at 8x, there is a very good chance that it will be more reliable at 4x, possibly better yet at 2X. For a premaster, doubling the write time is worth it.

A note on monitors

Many problems that mastering engineers encounter come from mixdowns made on less than perfect monitors. Its easier to fix the problems when you know that you can trust the monitors. Every engineer needs a reliable reference.

Some prefer smaller nearfield monitors placed close to the listener, since these can minimize problems associated with the acoustics of average rooms. These monitors can't deliver deep bass, but since this is also true for most speakers that consumers use, it's not a bad compromise. Also, speakers in this size range can often create a very accurate stereo image, in depth as well as width.

Some engineers prefer larger wideband monitors, correctly placed in a room that is designed for listening. These systems will give the most accurate picture of the music, assuming the speaker/room match is done right. Of course, most consumers don't have systems this good. Still, there's a lot to be said for speakers that are just plain accurate, regardless of what the rest of the world uses. If your speakers tell the truth, then you can compensate based on experience.

The 1610/1630 Umatic tape

The Umatic (1610/1630) format was the worldwide standard for storing and interchanging CD masters until recently. A strong point of this format is that compared to the common AES/EBU or SPDIF interface that connects most modern audio devices, the SDIF-2 (3-wire) Umatic interface is more stable. Umatic machines offered consistent performance even though the tapes were variable in quality and degraded slightly each time they were played. Sony no longer produces Umatic machines, and I've recently had replicators request that I ship CDR or other formats for long programs, where tape tracking is an issue. This situation was much less common during the heyday of these units. (Not coincidentally, you can buy half a dozen CD drives for the cost of routine maintenance on a single Umatic deck. If I were a replicator running multiple decks, I'd want to phase them out also!)

The net result is that Umatic is becoming less reliable and universal for data interchange. For this reason, after doing over eight hundred albums in Umatic format, I discontinued support of Umatic as of mid 1998.

Fortunately, CDR decks driven at low speeds, from low-jitter sources, are very reliable. Combine this with the ability to audition the actual master that will be used to cut glass, and its not surprising that CDRs have become the standard way of storing 16-bit masters.

My CD has to sound louder than any other disc ever made. Okay?

All kidding aside, this is a concern of record labels whose product is headed for radio play. I've also done a lot of sampler discs, and each artist on a sampler wants their sound to stack up well against the competition. Everyone is looking for an edge, and sheer volume is an obvious way to get that edge. Years ago it was shown that if you play the same piece of music twice for listeners, one time just a bit (1/2 to 3 dB) louder, most folks will prefer the louder playback. Stereo loudspeaker salesmen have used this trick for years to sell "house" speakers that carry the highest profit margin.

If you want maximum level, there is one good reason to go for it at the mastering stage. Like it or not, radio stations will jam your music through their own limiters prior to being transmitted over the airwaves. The sonic quality of these processors is usually not perfect, to say the least. The processors in a good mastering facility will produce high levels and better maintain the sound quality that you have worked so hard to create. I keep highly modified multiband compressors on hand just for this purpose, and the Halo processor also excels in this area. The limiters at the radio station will still be in the signal path, but they won't affect the sound as much.

What are the other considerations? If you master an album super-hot, for airplay only, all the tracks will sound punchy, but you run the risk that people at home will experience listening fatigue after 15 to 30 minutes. Music buffs generally don't enjoy hot sound for more than a little while, even if they like the material itself.

Looking at this problem from a music industry perspective, record labels are getting locked into a no-win contest to be 1/2 dB louder than the next guy. What gets sacrificed is clarity. You can always push the level up a bit, but the live feel and sense of air will suffer. A good home listening system can reproduce over 90dB of dynamic range. This is wasted when the music is crammed into the top 10 dB. Some related information about this can be found in an open letter that I wrote to the editor of an industry magazine on the subject of clipping.

The solution? Its usually possible to settle on a fairly hot compromise level that keeps everyone happy. For those who want to push a particular track very hard, one possible scenario is to release promotional EP's or singles, equalized and compressed for airplay. The record store version can then be mastered for a hi-fi home environment.

Technical tip if radio play is important to your release... be sure to check your mixes in mono as well as stereo. If there are mono compatibility problems, large listening areas may receive a signal that is radically different from the sound you hear in stereo. I pick these problems up quickly when I analyze your tape, but by then its too late. See the section on Creating Tighter Mixes for more info.

To summarize, if you want your whole album as loud as possible, I'll work to make it that way. You just need to be aware of the tradeoffs.

What are the other options for getting my material to sound great?

A handful of top mastering facilities offer engineers with decades of experience plus an intangible: marquis value. Some folks want that Cadillac name to be in their liner notes, and this is not a bad idea - just make sure that the ears that made the facility famous will actually work on your music. Great ears and vast experience command everone's respect, including mine. If marquis value is important, its a good idea to plan six months ahead. Expect album mastering costs to be in the multiple thousands.

At the other end of the scale are small studios that buy software and a CDR drive for their Mac or PC and get into mastering. You can expect rock-bottom rates for transfers from DAT mixdown tapes to CDR. If your multitrack has been mixed carefully, the sweetening and editing needs are basic, and the sound quality meets your standards, then you may get a real bargain.

On the other hand, if the producer decides that the finished CDR needs a little something to make it more competitive or, worse yet, that the software actually pulled some of the life out of the music, then you might need to work with a local facility where mastering is the only business, not a sideline. Many studios can't justify the very high-end toys you need to do the job.

Nevertheless, this option can be a winner if the engineer has great ears, a lot of serious technical and musical experience and top-notch tools. If you have a studio like this near you, please use it. The operator of that facility has paid his dues and deserves your business. You will get a good product and receive great value for your dollar.

Digital editing vs. Crash editing

Before a master CD can be produced, songs need to be put into the correct sequence, and the proper spaces placed between each track. This can be done before sending the tape for sweetening, or I can do it after the music is sweetened.

Digital editing: Most major and indie labels demand that sequencing and spacing be done using a digital editor, for quality reasons. Digital editing allows you to remove countdowns, stick clicks, guitar amp hum and noises between tracks. Fadeins and fadeouts can be "chased" to digital silence so that console or analog tape noise disappears. This gives a polished result that people notice.

Crash editing: Albums can also be sequenced and spaced by "crash editing" - using a digital connection (preferably coaxial, not optical) between two DAT recorders. This method does not allow you to chase the fades, adjust song spacing precisely or remove noise. The target DAT machine should be in record mode for one solid second before hitting PLAY on the playback deck. If you punch in tighter than this, your machine may play it perfectly, but others, including the ones at DRT, may not. Crash edits can produce low-level noises that may become more audible as a result of the mastering process. Crash edits also can insert up to 1 3/4 second delays into the ABS time code that you see on the counter as the tape plays. While this is not audible, it may affect the accuracy with which start IDs can be placed on the master CD. For these reasons I don't offer crash editing as a service.

An open letter to a studio operator: Choices

An experienced mixdown engineer emailed me to ask about mastering his own material and what the specific equipment options might be. My response expands on issues brought up earlier. For what its worth, here it is....

Of course, if you need the best possible sound in the known universe you have to send your tapes to me. I will perform the magic rituals required to make us all wealthy and happy. Life will be good. Barring that........

The TC M-5000 and Weiss units may still be the top of the heap in standalone digital Swiss Army knives. (I was an early M-5000 user.) Many facilities have also been drawn by the low cost of computers. They use EQ and limiter software plugins for ProTools or Sound Forge, such as the L1 Maximizer. Others prefer the features of Sonic Solutions. Studios on a tighter budget often use the TC Finalizer. (Tell it the style, and it masters the music for you! Well, almost.)

Fortunately for people with analog gear that outperforms the digital stuff, converters have improved to 24-bit resolution, and you can make one conversion in and out of either domain without losing the musical feel. It might still cost unreasonably large bucks to make that conversion, but the result is that you can choose digital or analog depending on the sound you need.

If you mix to 1/2" tape and/or prefer the wider palette available in the analog domain, (which I do), there are single band compressors, from the LA2A/3A/Joe Meek optoelectric flavor (slow attack, warm sound), to dozens of all-purpose VCA units (variable attack, neutral sound), to traditional expensive Fairchild/Manley variable-mu tube designs (fairly fast attack, smooth sound). If you're trying to avoid the intermod distortion inherent in all single-band compressors (think low-end mud), Aphex and others make brick wall multiband limiters in the $1K-$2K range. A handful of brands including Sontec, GML, Avalon and Focusrite have pretty much staked out the high end of the analog toy market. They deliver the best performance available from off-the-shelf equipment. Prices run from $2K to $12K per unit for limiters, compressors or EQ. I can't tell you which to buy, since I use different units for different tunes. You need a variety to cover the situations you meet. I tend to buy devices of the type I need, sell the ones that don't make the grade or can't be modified to be useful, and then upgrade the keepers to sound the way I want. I collect and create tools to develop the subtle control that you don't get with a Swiss Army setup. You have to mix colors to get the right tones, then blend them again on the fly as the painting evolves.

Regarding developing an in-house capability: I do albums for studios that have the financial strength and talent to get into mastering. You might ponder two of the reasons why...

1. They don't want to go through the gear evaluation process. Their profit margin comes from their known expertise - session work. The learning curve for mastering - at the quality level they need - can be steep and time consuming. Plus most folks are not fanatic enough to design units from scratch or carve up expensive gear to make it sound the best it can. They need the results without the rest of the nonsense.

2. Conventional wisdom in the high end of the business says to master the music somewhere other than the mixdown facility. Change the venue - change the perspective - the music benefits. It takes at least a couple of days away from a project to regain the perspective you need for mastering. Most engineers I know are well into their next project by then. The result is that experienced recording clients will send work out for mastering even if they LOVE your mixes. Its not a question of talent, its about letting specialists do what they do best. Another facility may actually make your mixes sound better than you can. On the other hand, if you develop an in-house capability, you can work on material from other studios, and there's a business opportunity. You know your market and what makes sense for you. The music business is so large, there's always room for another person doing quality work.

When to master locally

Experienced mastering engineers offer a perspective based on working with a lot of albums. They can (and should) provide creative input, building on the musical vision of the production team. The result can be a new and different view into the music - an album that sounds "in the pocket".

In working with labels all over the world, I've had success with phone, e-mail, fax and clear written notes to make sure I understand the producer's direction. Still, there have been a few times where the process has not worked. The problem was not that the finished master didn't meet high quality standards, but that the producer's vision was so specific that, in the end, they decided to master it themselves.

Fortunately, we parted friends, but there was a common thread in most of these cases: the producer was also the writer and arranger, and often the principal performer and recording engineer as well. They were so close to the project that it became difficult for them to assign artistic control, even for fine tuning.

This is a fairly rare occurrence, since most music people appreciate the results of a high-quality collaboration. Still, if your vision is extremely precise, you may prefer to work "over the shoulder" of an engineer at a local facility, or even run the controls yourself. You can make the subtle calls on the spot.

You need a great mix to get a great master: Avoiding major surgery

Most people who send albums to DRT have created a competitive product. There are no major technical problems, and the overall level, balance and EQ are in the ballpark. The producer just wants a big, radio-ready sound. At the other extreme, I had one client whose mixes were seriously flawed. After we discussed it, the producer decided to replace some weak tracks and remix the entire album. (You have to admire the commitment!) The mastered CD got chart action in Billboard, so everybody was happy... but the release date had to be pushed back six weeks.

This brings out a key point: mastering can only improve good product. It cannot fix bad mixes, mushy tracks, poor arrangements or sloppy playing. If the kick drum sounds great and the bass guitar is terrible, there will be problems getting both to sound good. If you are at the mixing stage, I welcome you to send an early reference mix for free evaluation. I may hear a problem that you can fix before it spreads across 15 cuts.

Note to musicians: Tradeoffs

Its easy to underestimate what it takes to make a truly fine sounding record. Learning to record music well is like becoming the master of an instrument. Pros need natural talent and a few thousand hours of hard practice, often under the guidance of an expert, to become competent at laying good tracks and mixing. If your group has an aggressive timetable for producing a tight, fully professional CD, you're much better off going to a good studio, rather than buying five to ten thousand worth of recording gear and learning on the fly. You'll get quicker results from someone else's years of experience, heavy investment in gear and excellent room. You can focus on playing and you won't be distracted by technical problems, which can wreck a creative groove quickly.

The flip side of this argument, of course, is that if you have time, dollars and the drive, then making records can provide a lot of long-term satisfaction. Just be realistic about what it takes.

Seven steps in the CD production process.

This is a general overview that describes how CD projects go from start to finish. Details may vary somewhat depending on the replicator you use.

1. Follow the guidelines in this brochure to produce a master CDR, ready for replication.
2. Create the artwork. You need to provide the photograph and text that will be printed in the booklet and onto the CD itself. These materials are assembled into camera-ready artwork. The assembly process can be done physically, which is called paste-up, or electronically, in a desktop publishing computer program. The result is that text and graphics are positioned exactly where they are going to be on the final product. Artwork must meet a tight set of specifications, which your replicator can supply to you if you wish. If you have the assembly process done in your home town, your designer will create physical artwork or a computer file that contains the right information. If you do not have local design resources, your replicator can usually refer you to excellent designers who will produce finished art at reasonable prices.

   Three things typically cause the most delays and added costs: very complex designs, artwork not produced according to spec, and folks who revise their design after it is approved. Any of these three can result in a lot of phone calls, faxes and overnight packages before things are right.

   If you want very tight control of the artwork design from start to finish, it's probably best to find a talented local designer and interact with them daily as they produce what you need. If their work meets the specs exactly, it should be smooth sailing. If you choose to use the graphics service recommended by your replicator, they can usually fax a black and white proof to you for initial approval of the artwork.

   Replicators and design firms often ask for payment to arrive with the artwork. This gives personal checks time to clear. Any additional expenses for special shipping or design services are usually paid for before the discs ship.

3. Film is produced from the finished artwork. The film will be used in the printing process. Plan on three to five working days for producing film, assuming that the process goes smoothly and the artwork meets spec. Some film houses work faster, but its a good idea to allow the time.
4. If you need to see a proof of the film, called a color key, one can be shipped overnight to you. If it looks fine, you can usually give approval by phone to get things rolling quickly, then mail or fax a final written approval.
5. If your replicator does not have in-house printing capability, the approved film is shipped to the printing plant that you choose. Printing generally takes ten to fourteen business days. Booklets and tray cards are shipped to the CD replicator so that he can package them with the discs as they come off the production line.
6. If your replicator has an in-house printing plant, film and the audio master are generally shipped there at the same time. The replicator creates a glass master disc from your CDR or Umatic tape. Most plants produce glass masters in-house for quality control reasons. Stampers are then made and the CDs are pressed and packaged. You can plan on about fourteen working days for the CDs to be ready. If you are providing the finished booklets, plan on ten days. Your replicator can give you more precise information. The pre-Christmas rush, for example, often causes delays.
7. Discs are shipped from the plant to you, usually via UPS Ground service. If you have a release party or other deadline to meet, some or all of the discs can be shipped by overnight carrier.

Short run replication

If you need to get advance copies out to agents, labels, or radio stations, I can make short runs of CDRs. Copies are made on the same 8x-certified silver media used for your production master. I can also produce basic single-sided liner note inserts with artist name, album title and track listing - adequate for getting the buzz started about your release, etc. - until your large CD run is ready. If you need retail-ready graphics, I can refer you to sources that provide replication and very sophisticated inserts. You'll need artwork and typesetting if you go this route. The discs themselves can be labeled with indelible markers, circular paper labels, or directly with special color printers.

Back to Basics: Four steps to tighter mixes - from the mastering perspective.

The Stereo Field

• Panning tips: Keep instruments that convey power and solidity in the center, unless you want a retro ping-pong stereo feel. Kick, snare, bass, lead vocal and most instrumental solos all can benefit by staying in the center. Doubled crunch guitar tracks or keyboard parts, plus background vocals, can often be panned way out to create a sense of space. The Fender Rhodes piano is also a natural for this treatment. Toms and drum overheads can be placed at least halfway out with good effect.
• Monaural compatibility: One hallmark of pro mixes is that they sound great in mono. This is important for songs that will be played on FM radio. In fringe listening areas, most stereo receivers automatically revert to monaural reproduction to reduce noise. You can't control when this will happen, but you can make sure that you don't aggravate the problem. Frequently hit the MONO button on your console when you record and mix. This combines the left and right channels into one signal that is sent to both speakers. When you switch to mono, the image will collapse to the center, but both the tones of the instruments and their relative volumes should stay the same. If the sound of an instrument changes drastically, the likely suspect is time delay.

  Digital delay lines, guitar processors, synths and stereo room mics can generate wide stereo output. Check all of these channel pairs closely for mono compatibility. Short time delays can produce huge stereo effects, but when the two channels recombine, all out-of-phase information will disappear, along with the sense of size that it created. In extreme cases, an instrument can change tone completely or vanish from the mix in mono!

  (Note: If you have a stereo instrument track that sums poorly to mono, try adding a few milliseconds of delay to one of the channels. You may be able to "time-slip" the track back into mono compatibility. This is easy if you have a digital editor such as ProTools or SAW. Just drag and drop one side of the pair. This may save you having to re-track the part!)

Mix Balance

The tone part is harder. For discussion, tone can be broken down into two components, spectral content - bass, mids, highs - and attack, the leading edge of a note. On a wide-range instrument, the fundamental tone conveys power, the lower harmonics (midrange) convey sense of fullness or presence, and the higher harmonics, attack and definition. A few examples:

• Midrange Guitar and keyboard leads need to be forward in the mix. To make this happen, many people reach for the midrange EQ control. The 1 kHz to 2 kHz frequency area is critical to the body of the sound. Players doing Metal styles tend to cut this range out to create an aggressive sound, but if you pull it too far back, the presence will disappear. If your EQ includes a BANDWIDTH control, start with a one-octave setting. Use narrowband settings to tune out annoying tones such as the low-frequency resonance of a guitar body or snare drum.
• Enhancing Attack Many engineers assume that treble EQ is the best way to produce attack. While its common to chase clarity this way, you can end up with the treble cranked on every channel, and things will start to sound tinny. Try using compression, which can control the attack and smooth out volume variations at the same time. Adjusting the compressor's ATTACK control to slow setting will let the sound's leading edge pass through, while the body of the sound is reduced in level. This gives increased definition. A faster ATTACK setting will clamp the leading edge, so the body of the tone will dominate. Set the RELEASE control so that the compression cycle is finished by the time the next note is played. With the right settings, you'll have a clear tone without the added grit that sometimes comes with treble boost.
• Tighter Low end Compression can also do a lot to give the sound of the bottom end more weight. The attack of a kick drum or bass guitar will get your attention, while the body of the sound moves the air that you feel. Its worth spending whatever time you need to get solid lows. You can build anything on that foundation. If you fail, the track will never have power.
• Sub-Bass Watch out for frequencies below 40Hz. They eat up airspace in the mix and are too low to be heard on a many systems. An exception might be 5-string basses or HipHop drops that are designed to rattle the windows in the car next to you. Even in those cases, much power comes across in the second harmonic, from 80Hz to 100Hz. Many engineers roll off everything below 40 Hz to make the mix sound cleaner.
• Snare level Most tunes benefit from a strong backbeat. You can mix snares hot because they are staccato, but they'll sound weak without enough low/mid body. Try stepping on the initial peak of the sound with fast-attack compression, and allow the body to come through as the gain control circuit releases.
• Vocals Everyone loves air and vocal clarity, so 10 kHz gets boosted. Consonants become more understandable, and you hear throat sounds. If you love this feel, add a de-esser to your equipment rack. De-essers can control just the excess sibilant sounds that lie in the 8kHz to 13kHz range, while leaving the detail and body of the sound untouched. Add some low-ratio compression to help the consistency of the performance, and you're 90% of the way to solid vocals. The mastering engineer can de-ess the entire mix, but its always better to fix a single track when possible.

Your Other Monitors

What's the solution? You could spend next year's mortgage payments on speakers plus room treatment.... Or maybe not... So let's talk headphones! You already have seven pairs in the studio for overdubbing, right? Forget them - we need reference quality. I'll stick my neck out and recommend three fine examples. Sennheiser HD600's, Grado RS2s, or my favorite, Stax Lambda Novas with tube amp. Street price of maybe $330, $450 and $1500 respectively. Ouch! Not cheap, but worth every nickel.

Listen to five minutes of good commercial music on any of them. Then one of your mixes. The differences will be easier to spot with headphones. This is because you have eliminated the room from the equation, and your ears are hearing only direct sound. Plus all of these units are pretty accurate to start with, down into the deep bass where mini-monitors never go. Think of them as near-field monitors on steroids. You'll still need speakers to check panning, since phones exaggerate the stereo image. And phones won't hit you in the chest with your massive new bass tones the way that speakers can. But they can be a magnifying glass for questions of balance and tone. You will hear stuff that was lost through your speakers. And you'll have one more tool for achieving the holy grail - mixes that translate well across many speaker systems. By the way, HD-600s and RS2s have dynamic drivers, so they need a strong, clean driver amp.

Mixing Technique

• Create a short mix. Focus your attention on a section of the tune no more than sixty seconds long. Include a piece of the main hook, part of one verse and some of the bridge if possible. If you have a digital machine, set it up to play back the section as a loop. Now work on the tones. Auditory memory is short. There are times when you might want to loop ten seconds of music to concentrate on the sound of one instrument.
• Start with the basic tracks. Bass, drums, one rhythmn instrument, and lead vocal. Go for clarity and impact first. Work until your short section sounds killer, then document the EQ, level and compression settings. You'll be able to use most of what you learned so far on the remainder of the tune.
• Add parts one at a time. Resist the temptation to dial in all nineteen awesome groove tracks and cool licks that the guitar and keyboard players wrote. That is the way of the Dark Side. If you give in, you'll use up your resources too soon, the mix will become dense and lose it's focus. There will be less tension and release, which is what involves people emotionally in the music.
• When the mix gets away from you, remember the silence. (Deep wisdom, Grasshoppa.) All kidding aside, you want to hear the instruments clearly, so become sensitive to that transition point where too many things are going on at once. When you can't hear any silence behind the notes, cut away instruments until the energy of the mix returns. In extreme cases, return to the settings you saved before your ears and judgement became toasted. Take a break and try again.
• Maintain sane levels. Don't record super-hot signals onto a digital mixdown deck. With DAT and CDR, any level above -6dB means you have a 16-bit recording. There is no advantage to pushing it harder. If the clipping indicator comes on, you have already lost music. Try for maximum peak levels of -3 to -1, and let mastering bring the volume up to commercial standards. If you use a Finalizer/ Quantum/ Masterlink/ Maximizer /Ultramizer during mixdown, try a very few dB of low-ratio program compression, to tighen up the sound a bit. If you use the Crush-O-Matic preset and change your mind later, you may have to remix the album to undo the damage. Fair warning.

Recap

More On Improving Your Final Product:

I've had quite a few conversations with artists and owners of smaller studios on ways to improve recordings. Here are a few more tips, expanding on the info in the article above. They might be of interest if you are engineering your own album: