I have recorded a number of community bands and orchestras in the Portland, Oregon, area using relatively inexpensive equipment and produced CD’s (technically, CD-R’s). I’ve gotten many compliments on the results. The technology now exists at a price that puts high-quality recording within the reach of any group. I would like to share my experience in the hope of bringing recording capability to every group that wants it.
I record performances using two inexpensive microphones and a portable mini-disk recorder. I do some minor editing on the mini-disk recorder, then I transfer the mini-disk recording to a computer using a mini-disk deck and a digital I/O card on the computer. I do more editing on the computer, then burn CD-R’s. I also sometimes go directly from mini-disk to CD-R using a stand-alone audio CD burner, but I use this path only to get a “draft” CD out quickly.
While I believe the techniques and equipment I use produce excellent results (my fellow musicians seem to agree), there are a lot of subjective aspects to music reproduction, so I must warn that my scheme may not produce results deemed excellent by every listener. One can get into almost religious arguments here. To quote Peter Schickele (who was quoting Duke Ellington), “If it sounds good, it is good”.
The complete process is described, divided in to the following subject areas: Microphones, Microphone Placement, Recording, Master Creation, Master Post-Processing and CD-R Writing. In addition, a couple of appendices delve into the more subjective or detailed aspects.
Microphones can be a problem because good ones are expensive, and there so many models in the marketplace that choice is complicated. However, I’ve found an inexpensive and excellent-performing solution, namely, Panasonic electret condenser microphone cartridges. These are available for approximately $25 for 10 from Digi-Key (www.digi-key.com). The two models I have used are the WM-60AY and WM-61A. These cartridges are commonly used in cell-phones and telephone answering machines, which are not very demanding applications, but they are specified as having the wide frequency response (20-20,000 Hz) necessary for music recording. I suspect that some expensive microphones on the market have these cartridges at their hearts. The cartridges are small, about a quarter of an inch in diameter (6 mm) and about that high (WM-60AY) or about half as high (WM-61A).
One problem of using these cartridges is assembling them into usable form, which requires the ability to solder minute connections, and find a suitable enclosure. I have used small insulated covers normally used to cover alligator clips, with the microphone projecting out of the larger end. I would welcome other suggestions of how to package these cartridges.
Another problem with these cartridges is that they’re not well matched. Any two cartridges can have several decibels (10-20%) different sensitivity. One way to solve this is to buy cartridges in lots of 10 or more, pick two, make a recording of a sound source with the two cartridges next to each other, and listen to the result, which should be a well-centered, monophonic sound.
For those who are soldering-iron challenged, Sound Professionals (www.soundprofessionals.com) offers similar microphones in ready-to-use form. (I have not used their equipment, so this should not be considered an endorsement of their products).
The Panasonic cartridges require an external power source, but can be plugged directly into the microphone inputs of a number of recording devices, specifically those which provide so-called “plug-in power”. However, these provide only about one and a half volts to the microphone, which is at the lower limit of its specifications. Microphones plugged directly into the recorder can be overloaded by loud sources (e.g. concert bands). Therefore, I have constructed a separate “battery box” to provide higher voltage. The schematic of this is shown in Appendix C. Again, for those who are soldering-iron-challenged, Sound Professionals offers similar ready-to-use Battery Modules, with and without built-in attenuation. As I will describe further below, I used both the Sony MZ-R50 and MZ-R37 mini-disk recorders; the MZ-R37 requires attenuation in the battery box.
For microphone cable I use RG-174 coax, which is of small diameter (1/8 of an inch) and light weight, appropriate to the small size of the microphones. It also has good shield coverage and relatively low capacitance-per-foot, compared to some common types of microphone cable. The former helps to prevent interference, such as from AM radio stations, intruding upon the recording. The latter assures preservation of high frequencies. I have obtained RG-174 coax from Jameco Electronics. (www.jameco.com).
The Panasonic cartridges are nominally omni-directional that is, they pick up sounds equally in all directions, however I find that as I use them and mount them, they are somewhat directional at high frequencies, with the on-axis being favored over off-axis. This can effect how one would position the microphones for recording. More on this under “Microphone Placement”.
Of all the topics I cover in this document, microphone placement is the one that will elicit the most “religious”-type argument. There are many placements used in professional recording, some a favorite of particular companies, others the favorites of particular recording engineers. James Boyk, formerly of Caltech, has tried to introduce some science into this art, and his work is documented on a CD "Demonstration of Stereo Microphone Technique” which is described at http://www.performancerecordings.com/albums.html (look about half-way down the page).
I place each of the two microphones on stands (which are available for about $40 each) from 6 to 8 feet apart, and from 0 to 8 feet from the group being recorded, which in my case, is usually a symphony orchestra or concert band. The stands have a six-foot main section and a two-foot boom, so the mikes are about 8 feet in the air. I generally angle the mikes to point toward the group. As I mentioned above, the Panasonic cartridges seem to be slightly directional for high-frequencies, so this tends to brighten the sound a bit.
The “industry standard” (or maybe “dogma”, to continue the religious argument analogy) for this mike configuration, commonly called "spaced omnis", is to place the mikes away from the sound source one third of the distance between them, so for an eight-foot spacing, the mikes should be 2 and 2/3 feet from the source. However, when the source is large, as is a symphony orchestra or concert band, the “distance to source” is hard to define. Do you use the distance to the nearest player, or to the center of the group?
The distance from the mikes to the group affects the quality of the sound because it affects the mix of direct versus indirect sound (i.e. reverberation). Too far away, and the group will sound like it is at the bottom of a well. Too close and the sound has a quality I can best describe as “in your face”. Listen carefully to different commercial CD’s and you will hear a variation between CD’s (sometimes even between tracks on the same CD) in the apparent distance from the mikes to the group. Obviously different recording engineers have different preferences in this regard. Many (most, I suspect) commercial CD’s are “sweetened” with reverberation added digitally, and the mix of reverb versus original sound determines the perceived distance.
The acoustic properties of the venue have a large affect on the resulting recording. Some halls have too much reverberation, which tends to produce a “muddy” sound. Close miking would be preferred in this case. Other halls have little reverberation. One group I play in and record regularly plays in a college lecture hall, which has been “tuned” for voice, therefore, has little reverberation and a peculiar frequency response (in the words of our conductor, the room “eats bass”). No mike placement seems to compensate for this hall’s shortcomings.
The most challenging recording I have done is of a concert band outdoors, where there is no reverberation at all. A most “dead” sound. In this case, I reprocessed the master through a digital reverb unit.
Here is a picture of a sample mike placement. This is the Tualatin Valley Community Band performing at the Evergreen Air Museum (current home of the Spruce Goose), in McMinnville, Oregon. I have the mike stands aligned with the chairs of the outside players. In retrospect, I should have had the stands somewhat farther back to get more reverb and less direct sound. This venue is a building with a volume of several million cubic feet and a reverb “hang time” of several seconds.
You should experiment with different mike placements and choose the one that sounds best to you. Rehearsals are the best time for this, so you can have your make placement chosen before the performance you are recording, if the rehearsal is in the same venue where the recording will take place.
I use Mini-Disk as my recording medium. Mini-disk uses a combination of magnetic and optical techniques to record 150 megabytes on a disk in a 7x7 cm cartridge. It is not CD quality, because it utilizes a 5-to-1 compression technique called “ATRAC” to fit 74 or 80 minutes of digital audio in a space that would normally hold only 15 or 16 minutes, if recorded in true CD quality (44,100 16-bit stereo samples per second). It is, in the opinion of most listeners, “near CD quality”, that is, audibly indistinguishable from true CD-format recording. To my admittedly aging, but well-trained ears, it is indistinguishable. It has the other advantages of digital recording, namely, excellent signal-to-noise ratio, excellent frequency response, and the ability to post-process audio data in the digital domain without introducing analog distortion or unintentional frequency-response changes. Mini-disks are editable, which eases the post-production task somewhat.
Mini-disk recorders are small and portable, which is a convenience at the recording venue. Besides my mike stands, I carry the rest of my recording studio in a bag 4 x 11 x 15 inches. This is what the entire “kit” looks like:
The recorder is battery powered, eliminating the need to string power cords (finding nearby outlets is a problem at some venues). Because of its small size, the recorder can usually be placed near the microphones, which allows for shorter cables, with less chance for interference and frequency response degradation, and shorter cables mean less to carry.
I use a Sony MZ-R50, which is no longer made, but may be available used or refurbished. I have also used a Sony MZ-R37, which may also be no longer available new. However, a number of other units that should work as well are available. For more information on mini-disk, as well as information on specific models, see www.minidisc.org.
As mentioned above, I connect the microphones through a separate battery box which allows them to handle louder sounds (i.e. gives them better dynamic range). A battery box is further described in Appendix C.
The MZ-R50 has a “high” and “low” mike input sensitivity switch. The Panasonic cartridges will overload the mike input on the “high” setting when recording the groups I record, so I always use the “low” setting.
The MZ-R37 has no mike sensitivity switch; its mike input sensitivity seems to match the MZ-R50’s “high” setting, therefore, the MZ-R37 must be used with external attenuation (most conveniently provided in the “battery box”), at least with the large groups I record. A battery box with attenuation is described in Appendix C.
Headphones are used to monitor the recording. They’re used mostly to verify that the equipment is working. If the recorder and headphones are near the sound source, the sound from the headphones will be overridden by the sound from the source, so headphones can’t be used to assess the quality of the recording while it is taking place. I use Sony MDR-V6 headphones.
The trickiest part of recording is setting the recording level. Some recorders allow changing the recording level while recording. The MZ-R50 and MZ-R37 do not. (a source of minor complaint from owners); recording must be paused to change the level.
You could use the automatic recording level provided on most recorders, but then you would lose the dynamic range of the source, and sudden changes in loudness might not be compensated for quickly enough to avoid momentary overload. Therefore, proper setting of a manual recording level is mandatory. The recording level must not be so high that the loudest sounds overload the recorder, causing clipping distortion. The recording level must not be so low that the softest sound is buried in the noise.
Too high a recording level cannot be compensated for in post-production; distortion can be “softened” somewhat but not eliminated. Too low a recording level can be compensated for merely by multiplying all the digital values by a constant. This will also multiply any noise present, thereby decreasing the so-called “signal-to-noise ratio”. However, mini-disk has inherently so little noise (the medium itself, unlike analog tape, is absolutely quiet, as are all digital media) that the increase in noise level is usually inaudible. Therefore it is better to err on the conservative side, i.e., set a lower recording level than may be necessary, and plan to compensate in post-production. In my experience, musicians tend to play louder in the actual performance than in a warm-up or rehearsal, so it may be wise to set a little lower recording level than the warm-up suggests.
Again, two important principles of recording are 1): make sure the microphones used don’t overload the input; 2): set the recording level appropriately.
When the recording is complete, place track marks at the start and end of each piece. Place track marks to define the start and end of each movement of multi-movement works, like symphonies. Commercial CD’s usually record separate movements on separate tracks. It is more time-efficient, in the long run, to work with a larger number of shorter tracks. Label each track containing music, so it can be located easily during mastering.
You could also delete the tracks that don’t contain music, but you risk deleting the wrong track by mistake. It’s better to merely add track marks and label the music tracks.
I usually leave about 10 seconds of applause (assuming there was at least that much!) after single-movement pieces or after the last movement of a multi-movement work.
Mastering, or post-production, as I have sometime called it above, is the process of converting the recording on mini-disk into .WAV files (which I call the masters), on a computer, suitable for writing CD-R’s. I have found that carrying out the various steps of the procedure I describe takes about 4 hours per hour of recording to get ready to produce CD-R’s.
When I began making recordings, I did not use the computer at all. The original equipment I purchased for CD-R creation, in addition to the MZ-R50, was a Sony MDS-JE520 Mini-disk deck, and a Philips CDR-760 audio CD recorder. I made these purchases because the MDS-JE520 has a digital output which can feed the CDR-760’s digital input, allowing CD creation without any intervening analog-to-digital or digital-to-analog conversions, thereby yielding the best quality. However, this scheme does not allow any manipulation of the mini-disk data, so a mini-disk recording at a low level produces a low-level CD. Also, the stand-alone recorder requires the more expensive audio-coded CD-R blanks. I still use this setup occasionally to produce single “draft” CD’s when I need to send someone a copy before I can go through the complete mastering process.
As the volume of CD-R’s I was producing rose, the computer, with its high-speed CD writer, was a necessary tool, plus I wanted to produce higher-quality CD’s. The computer method traded the added mastering time for the reduced reproduction time and increased quality.
There are two requirements to produce masters. The first is suitable sound-recording software. Most computer sound cards come with recording software. I use DART CD Recorder 4 (which I also use later in the process to write CD-R’s). Dart’s web site is http://www.dartpro.com/.
The second requirement is a connection, either analog or digital, from the mini-disk recorder (acting as a player here) and the computer. There are two ways to make this connection: through an analog path and a digital path. The less expensive method is analog, connecting the mini-disk recorder’s line output to the line input of the computer’s sound card. A stereo mini-plug-to-mini-plug cable is appropriate if the sound card has a mini-plug line input. Some sound cards have dual RCA-type jacks as their line input, requiring a different type of patch cord.
The recording software may offer several output format options. The preferred output format is WAVE (.WAV). The transfer to the computer is initiated by starting the recording software and simultaneously starting playback of the desired track on the mini-disk recorder.
Two things are necessary to get quality masters. First, use the sound card’s mixer application to set all input sources except the line input to minimum. These sources may include microphone, CD, and synthesizer inputs. If not set to zero, they can contribute noise, particularly the microphone input.
Second, the recording level must be set correctly by setting the line input or master input level, or both, appropriately. Usually your recording software will provide some kind of level indicator, and may have its own level-setting control. Just as you previously set the mini-disk’s recording level based on the loudest sound to be recorded, you must set the computer’s recording level based on the loudest sound. As in the original recording process, too high a level result in clipping and distortion. Too low, and the result will be noisy. Once you’ve determined the correct level, you can use that level over and over again because the mini-disk output will never exceed a certain maximum, assuming you’ve experienced that maximum level. You can set the computer recording level by making a dummy recording on the mini-disk at an intentionally high level, one that drives the recording indicator on the mini-disk to its maximum, and even drives the mini-disk recorder into overload, then playing that into the computer recording software.
A mini-disk recording that was made at too low a level can be compensated for by increasing the computer’s recording level. It is preferable to do this at this time because a better signal-to-noise ratio will result than from using a lower recording level and compensating later.
Shut down every application on the computer except the recording software. This includes internet access, e-mail, virus protection, and even screen-savers. All of these can interfere with recording and causes glitches in the masters. I have DSL on my computer and when I determined I needed some kind of internet firewall, I chose a separate hardware router/firewall box instead of a resident software firewall, in order not to introduce another piece of memory-resident software that might conflict with the audio recording software.
Recording software vendors also suggest keeping your hard drive de-fragmented.
Make a short computer recording of the loudest portion of the mini-disk, and use the Windows Media Player, or a playback function if available within the recording software, to listen to the result. Use headphones connected to the sound card’s audio output, rather than the computer’s speakers, to do the best job of evaluating the master. Also record a quiet section of the mini-disk to check the master for noise.
Once you have established the correct recording level, record entire tracks from the mini-disk onto the computer. After a track is recorded, you need to verify that the recording was glitch-free. This is done by playing back the entire track, using Windows Media Player, or the playback function of the recording software. Glitches can be clicks, pops, gaps or jumps in the music, or short (fraction of a second) repeated sections.
If you detect a glitch, verify that it is a real glitch by rewinding the playback over the glitch and repeating the playback. Just as glitches can be randomly created while recording, temporary glitches can appear during playback, but not be actually present in the master. If the glitch in the master is real, you might want to go back to the mini-disk original to verify that the glitch did not come from the mini-disk itself. In several dozen recording sessions, I did find one glitch in one track on a mini-disk original, which is a tribute to the reliability of mini-disk.
If you detect a real glitch in the master, the only effective remedy is to delete the bad master file, and re-record the track. This is one reason to break multi-movement works into multiple tracks. The probability of a glitch occurring is proportional to the length of a track, so shorter tracks will have fewer or no glitches, and if a glitch is found, it takes less time to re-record a shorter track.
The analog method described above can produce good results and requires no additional equipment besides a patch cord. However, there is some loss of quality because the mini-disk recorder is doing a digital-to-analog conversion and the computer sound card is doing an analog-to-digital conversion and neither of these conversions is perfect. Also, there is the possibility of noise sneaking into the analog signal path.
I produced a couple of CD’s using the above method, but I opted to go an all-digital route instead. I already had a mini-disk deck, the Sony MDS-JE520, which has optical digital outputs. I added to my computer a Midiman DiO 2448 PCI Digital I/O Card (http://www.midiman.com/), which has optical and coaxial digital inputs and outputs. The mini-disk deck digital output is recorded onto wave files without any redundant conversions, and with no possibility of introducing analog noise.
Perfect transfer though it may be, the digital method is still susceptible to glitches during the recording on the PC, no more or less so than the analog transfer method, so it is still necessary to listen to each track carefully after recording it. Note that, unlike the analog method, the digital method transfers the exact signal level the mini-disk was recorded at; there’s no control in the path to boost the level, although some recording software might provide a digital gain control. Also, some mini-disk decks might provide a digital output level control.
Another advantage of the all-digital setup is that with the DiO 2448 digital output connected to the digital input of the MDS-JE520, the latter can be used as a high-quality digital-to-analog converter (when set to Record with no mini-disk loaded) to drive headphones for checking the masters and monitoring during post-processing.
Once you have created clean masters, you could proceed to burn CD-R’s, but additional work on the masters may be preferred or required. For example, I like to clean up extraneous noise at the beginning of a track (coughing, chair squeaks, etc.) and taper the applause at the end of a track. Also, I usually adjust overall levels, and occasionally left-to-right channel balance. For this I use Voyetra Digital Orchestrator Plus, Version 2.11. Voyetra’s web site is http://www.voyetra-turtle-beach.com/.
This software does an adequate job, but has some deficiencies and problems. Cleaning up the start of a track and tapering applause at the end must be done separately for the left and right channels, which requires care to keep the two channels matched. The software does allow adjusting overall levels for both channels at once, and it has some rudimentary signal-processing functions that can be used on both tracks at once. The major problem I’ve had with this software is that its playback function is glitchy. I find that after doing some processing, I must export a .WAV file from it, and then use Windows Media Player to evaluate the .WAV file.
Digital Orchestrator and other similar software packages call the level-setting function scaling. Digital Orchestrator has a “suggest” option in its scaling function, which will analyze a track and suggest a value that will make the loudest sound reach the maximum possible level. This is called normalization. However, since the loudest sounds are usually short transients (drum beats or cymbal crashes), I’ve found it’s possible to use a higher scaling value without introducing audible distortion. In fact, it’s necessary to use a higher scaling value in order to get a track to play with the “right” volume when a CD is finally created. Experiment with different scaling values until you find the value that sounds right to you.
The software I use for master recording, DART CD-Recorder 4, has a “Producer” function with several signal-processing functions, including normalizing, fading, de-click and de-hiss, and equalization however, these are controlled by uncalibrated sliders, and there are no user-savable presets, so I have found these difficult to use.
Once you determine a scaling value for one track, you should use the same scale values for all tracks in a session, unless some tracks deserve special adjustment (different instrumentation, for example). In order to determine the scaling value, I load each track in turn into Digital Orchestrator and do a “suggest” function on each, write down the result, and choose a single value. I then test that value on the loudest track to make sure that I haven’t introduced audible distortion by using too high a value.
Once you’ve done the level-setting and cleanup, and written new .WAV files, you’re ready to burn CD-R’s. I use DART CD-Recorder 4 (which I also used for the master transfer to the computer). There’s not a lot to say about this process. All the hard work is done in the earlier steps.
I started out using commercially-available jewel case inserts, but found that, as prices for blank CD-R’s fell, the cost of insert was more than half the total cost of a packaged CD. I went to Kinko’s and had them cut a ream of 70-lb stock down to 4 ¾ by 9 ½ inches, and that reduced the cost of inserts to about 3 cents each. These feed through and print on my laser printer just fine, and although they’re more difficult to fold (their not pre-scored like the commercial inserts) the low cost more than makes up for the minor inconvenience.
This ignores the issue of the so-called “tray-liner”, but lately I’ve been packaging CD-R’s in slim cases, which don’t take a tray liner, so the issue is conveniently rendered moot.
The MZ-R50 and MZ-R37 have a nine-bar recording-level indicators. The instruction manual says to set the recording level such that the loudest sound activates the seventh bar on the MZ-R50, and the fifth bar on the MZ-R37. During a warm-up prior to the concert, I ask the conductor to play the loudest section of the works on the program, and I set the recording level so that the indicator activates at the least the seventh bar, sometimes higher if I feel confident.
The MZ-R50 has a 30-bar record-level-setting indicator. The MZ-R37 displays the recording level setting as a number from 0-30. The MZ-R37 is easier to use because the numerically-stated level is easier to interpret and reset later if necessary. The bar-type setting indicator on the MZ-R50 is harder to reset. I find that I have to count the number of button presses necessary to turn all the bars off, and remember the number, or just use the bars as rough analog indicator. Resetting the correct recording level is important if you are recording an event that requires more than one disk. When the second disk is loaded, the recorder reverts to its default recording level, which is 13, and must be reset to the same level used with the first disk.
On the MZ-R50, I always use the low-mike-sensitivity setting when recording large groups. On the MZ-R37, a separate attenuator (such as in a separate battery box as described elsewhere) must be used.
I’m sure there are recording professionals who would react in horror to my setup with only two microphones. Their approach is to use multiple microphones, multi-track recording, and mix-down to two tracks. Most commercial recordings are done with multiple mikes, although I suspect some are done with only two mikes, judging from how they sound to me.
The main problem with multiple mikes, in addition to the extra equipment required, is that the mixing must be done with great skill, and in a lot of recordings I’ve heard, it hasn’t been done well. Remember that the brain places a sound in the horizontal plane, or azimuth, by the its determination of the relative delay of the sound in reaching each ear. This is known, by coincidence, as the Haas Effect. The brain also uses the relative loudness (amplitude) of sounds reaching each ear to derive placement cues. When the delay and amplitude information conflict, the result is, to my ears at least, uncomfortable, if not painful. Sort of the audio equivalent of looking at something with your eyes crossed.
When multiple mikes are used, each mike’s output is added into each final channel in a proportion depending upon to its azimuth (horizontal location). This is done with so-call “pan” pots on the mixer. The single signal is added into both channels with the same delay, but different amplitudes. Therefore the delay and amplitude values don’t correspond, giving the conflict I mentioned above. Most people aren’t bothered by this, but it bothers me. Some mixing systems can insert appropriate delays but these are the most expensive systems in use, and, again to my ears, apparently not used often.
Another problem with multiple mikes that the signals from nearby mikes can combine in such a way as to artificially boost some sounds and cancel out others.
The bottom line, in my opinion, is that the two-mike scheme is more than adequate for most recording, and that the multi-mike system does not usually produce results that are so much better that the cost and other hassles are justified.
The two-mike scheme does present problems when there are soloists performing with a large group. The soloist is usually placed to one side of the conductor. This means they are close to one mike, and far from the other. Therefore their performance comes out of only one channel. In most commercial recordings, soloists have a dedicated mike, and the output from that mike is mixed into both channels equally, placing them in the center.
As mentioned above, I prefer to power the Panasonic microphone cartridges from a source that provides higher voltage than the input jack of the mini-disk recorder. To do so, I built a “battery box” whose schematic drawing is the following:
This circuit powers the mikes from a voltage (9.0 volts) well above their minimum specification (2.0 volts) guaranteeing that loud sounds won’t overload the mikes.
C1, C2, and C3 should be high-quality aluminum electrolytic or tantalum capacitors. There is no on-off switch; there is only a significant drain on the battery (and then less than 2 milliamperes) when the mikes are plugged in. When the mikes are unplugged, the only battery drain is the leakage through C1, which, if it is a high-quality capacitor, is negligible.
This circuit should be built in a metal utility box for shielding. I used a box 1 ¾ x 2 1/8 x 3 ½ inches. One of the hardest parts of building this circuit is getting the correct connections on the jack and the plug so that the left and right channels aren’t reversed.
The Panasonic microphone cartridges have too high an output (at least when recording large groups) for the mike input of the Sony MZ-R37, so a battery box with attenuation must be used. The schematic of such a circuit follows:
Suitable values for R1, 2, 3 and 4 for the MZ-R37 are as follows:
R1 and R2 – 820
R3 and R4 – 1.5k
This gives an attenuation of about 63%, or 9db.
Last updated 7/04/2003 (Simplified html)
Back to my home page
