Recommended Procedures for Transferring Analog Audio Tape and Analog Audio Disc for Digital Output, with Participant Commentary
The following is an edited, expanded version of the proposed workflow documents created by Larry Appelbaum and Peter Alyea before the roundtable meeting. It sets forth the transfer practices agreed on by roundtable participants and incorporates comments and annotations made by individual participants during the online discussion period after the meeting. The roundtable meeting was held in January 2004 and the ensuing online discussion continued until April 1, 2004.
Many of the comments made during the online discussion were highly technical, delving into areas of dissent and controversy as well as agreement. While this document does not incorporate every comment made online, it does include most of them. The goal of this document is to stimulate further debate and to share roundtable participants’ expertise with the broader community.
ANALOG AUDIO TAPES
1. Pretransfer issues
1.1. Inspecting, preparing, and cleaning the source materials
1.1.1. Determining the composition (polyester [e.g., Mylar], acetate, paper) and the thickness
Optimal practice: Inspect the tape visually; this is usually adequate. Consider nondestructive testing methods. Tape boxes and labels can aid identification but are not always reliable. The optimal inspection technique is nonintrusive; however, the use of destructive techniques can sometimes be justified.
Occasionally, a single tape reel held in an archive or a collection may hold tapes of differing compositions. These tapes may be spliced or simply wound together. Such reels ought to be set aside and examined individually.
Thickness is important because it can help determine how a tape responds to stress and how it should be handled. In addition, tape thickness is an indication of tape length, which may be helpful in predicting program length.
Recommendation: Produce a flowchart or logic tool to identify the composition of the tape.
1.1.2. Methods or techniques used to address specific physical problems or conditions
Brittleness affects acetate and paper tape.
Optimal practice: Visually inspect the tape, then slowly wind and play it. If a scrap piece is available, test it by bending. Identify brittleness to implement appropriate handling techniques.
22.214.171.124. Splices (loose, broken [i.e., separated splices], or bleeding)
Splicing is one of the biggest problems associated with audio tape preservation.
Optimal practice: Repair all damaged splices before beginning transfer. Remove and repair bad splices if the splices pose a physical risk to the media or compromise sonic quality. Clean with naphtha-based lighter fluid or isopropyl alcohol (alcohol should not be used on acetate tape) and resplice.
Once the splice has been repaired, remove the tape from any consumer or non–preservation-quality plastic reel and rewind it onto a clean, standard, slotless NAB hub and metal reels.
Core competency: The ability to remove adhesive from old splices and to resplice tape. It is assumed that transfer engineers who clean tapes also possess basic tape-handling skills.
Recommendation: Articulate the exact steps for repairing and cleaning tape in a workflow document.
126.96.36.199. Miscellaneous tape issues
Overarching recommendation: Store tapes that cannot be transferred because of limitations in current technology until appropriate treatment procedures become available.
Cupping is a deformation in which the tape, when viewed end-on, appears curled instead of flat. Cupping is most common in acetate tape that has been stored in damp conditions.
Optimal practice: Let the tape sit as a B-wind for three to six months if possible. If that amount of time is not available, the tape may still benefit from a shorter period in the B-wind configuration. If a dub must be made immediately, either (1) adjust the tension of the playback machine or (2) use pressure pads. A tape that is severely cupped could expand and come off the reel. For this reason, it is important to wind slowly and to monitor the take-up wind.
A B-wind involves packing the tape with the oxide on the outside and the backing on the inside. (Other wind descriptions that preservationists should be aware of are A-wind, flat wind, library wind, fast wind, tail-in wind, and tail-out wind.)
188.8.131.52.2. Edge wear/edge damage
Optimal practice: Use a narrower track head for playback. Be aware, however, that this practice may sacrifice some signal-to-noise ratio and fidelity.2
There are two types of leader: plastic and paper. Plastic leaders can accumulate electrostatic charges that, if discharged during playback, could disrupt the signal.
Lack of consensus: Sprays are available to help remove electrostatic charges. Participants disagreed on whether such spraying could contaminate the tape. One engineer noted that to his knowledge, use of such sprays was not an accepted practice. He further stated that he was unaware of any research confirming their effects. Thus, the value of sprays cannot be supported or refuted at this point.
Optimal practice: When repairing splices as part of routine conservation activity, replace plastic leaders with paper leaders. Proper control of room humidity can also reduce electrostatic buildup.
184.108.40.206.4. Loss of oxide
Blocking. Blocking occurs when the oxide comes off the tape in strips. Such tapes should not be played or wound. At present, there is no remedy for blocking. Future research may result in techniques for preventing or remedying this problem.
Shedding. Some tapes shed as they degrade because of aging. Strict environmental controls can help minimize the rate of degradation; little else can be done to prevent it. Shedding differs from a condition sometimes referred to as sticky shed, which is discussed in section 220.127.116.11.
Optimal practice: Remove loose oxide from the tape and tape path whenever shedding interferes with optimal playback. Remove heavily shedding tape with Pellon cleaning tape (see section 18.104.22.168) before audio transfer. In addition, clean the machine transport as often as necessary.
Preventive measures for blocking and shedding include proper storage and cleaning.
Store tapes that cannot currently be transferred because of technical limitations until appropriate treatment measures are available.
22.214.171.124.5. Wind types: scatter/smooth, tension
Optimal practice: Do not put loosely wound tape under tension, and do not play it back in a loose wind on any servo-motor transport; hand winding is preferred for loose winds. Wind the tape onto a slotless NAB hub and metal reel. Experiment with different winding speeds and transports. To promote longevity of heads, it may be preferable to bypass them when winding.
1.1.3. Methods for removing or mitigating surface problems
126.96.36.199. Topical debris
Optimal practice: Depending on the amount of debris, address the pack first by vacuuming it before winding. Consider winding against Pellon cleaning tape. When using Pellon cleaning tape, clean both sides of original magnetic tape, i.e., the backing as well as the oxide. Feed the Pellon strip slowly and incrementally in order to present a constantly clean surface and to prevent accumulation of debris on a fixed pad of Pellon.
Optimal practice: Vacuum with a HEPA filter. Follow with Pellon cleaning on both sides of tape—backing and oxide.
Note: This procedure is a potential health hazard, and appropriate protection is advised. Cleaning should be done in an environment where it will not place other media at risk for contamination.
Two distinct conditions may cause tapes to become sticky and emit a telltale squealing noise during playback. One condition is sticky shed, also known as binder breakdown. It affects polyester tape and is an indication of binder hydrolysis. The second is lubricant loss, which primarily affects acetate tapes.
The most common practice for making sticky polyester tapes playable is extended baking at low heat in a scientific or convection oven. Other treatment methods include using environmental chambers and desiccants. Alternative methods should be explored.
Lubricant loss in acetate tapes should be treated through a relubrication process. Acetate tapes should never be baked.
In both instances, treatment results may vary, depending on the physical and chemical makeup of a specific brand and manufacturing run of tape.
Further research: Research is needed on (1) alternatives to oven baking for reversing hydrolysis in polyester tapes; (2) methods of vacuum sealing polyester tapes that have been baked in order to keep them from reabsorbing moisture; and (3) new methods for relubricating acetate tapes suffering from lubricant loss.
188.8.131.52. Water damage
Optimal practice: If the tape is dry, vacuum with a HEPA filter to remove debris, use Pellon in a slow wind, and replace the splices. If the tape is still wet, consider a distilled-water rinse. Allow the tape to dry, vacuum debris, use Pellon in a slow wind, and replace the splices.
1.1.4. Choosing cleaning methods
For spot cleaning of old splices, a naphtha-based lighter fluid (such as Ronsonol) or isopropyl alcohol is recommended. However, users are advised to consult relevant Material Safety Data Sheets (MSDS) before employing these chemicals.
For overall cleaning to remove oxide shedding or other dry, loose debris, dry cleaning using Pellon is recommended.
Optimal practice: Use judgment to determine whether tape degradation is present and whether intervention is needed. Every tape does not need to be cleaned before transfer.
1.2. Configuring and calibrating playback equipment
1.2.1. Determining playback speed
Optimal practice: Start by playing the tape at 7.5 inches per second. Listen carefully. If possible, have machines that can play back at all known speeds and head configurations that may be encountered, plus variable pitch control. Note all speed variations as documentation to accompany the digital preservation copy.
On the basis of their knowledge of program content, music or subject specialists may be able to help identify speed. Listening (or inspecting with test equipment) for low-level power-line hum at 50 to 60 cycles per second may help determine speed. This is because 50 and 60 cycles per second are the standard frequencies at which electrical current alternates (depending on which part of the world you are in). However, there may be slight fluctuations to this, i.e., rates may fall below 50 or exceed 60. Additionally, pitch may vary throughout the recording. (This is particularly true of field recordings.)
1.2.2. Modifying playback technique to maximize quality transfer of problematic sources
Adjust for azimuth. To maximize the transfer quality of the original tape, align azimuth to that of the original tape. Use an oscilloscope to adjust azimuth for optimal high frequency and phase coherence, where applicable. Over time, aligning azimuth to various source tapes may cause tape head surfaces to wear unevenly and become deformed, with a resultant loss in fidelity. This wear will require regular relapping of tape heads.
If the tape being copied is second or third generation, it can be difficult to optimize for azimuth correction. The azimuth alignment of each recording deck used to make each generational copy could differ, thus preventing optimal reproduction because cancellations will take place.
1.2.3. Monitoring aurally and with test equipment for anomalies
Both the experienced, trained human ear and carefully calibrated testing equipment should be used to monitor recorded-sound transfers.
Tools are available to assist in distinguishing transient, event-based anomalies (e.g., clicks and pops) from global anomalies that may affect the entire tape (e.g., bandwidth and dynamic-range limitations). Sources of anomalies in recordings include the original recording itself, degradation of the tape, playback-transport error, and signal-path error. Correctly identifying the source of recording anomalies is imperative for preservation of the recorded content.
1.2.4. Setting level gain throughout signal chain with or without tones
There is no industry-wide agreement on level setting. When a digital preservation copy is being created, level setting is dependent on the bit depth of the linear PCM recording. Higher bit depths provide greater dynamic range. Use peak-level metering, not volume-unit metering, when available.
1.2.5. Setting record level
Optimal practice: Use reference-set alignment tones provided at the beginning of the source tape if they are available. If the source tape has no reference tones, use an industry-accepted frequency-alignment tape as a baseline reference in setting playback levels.3 Have on hand a repertoire of alignment tapes to accommodate various flux densities.
Using a standard alignment tape to set playback level may occasionally result in very high noise in the case of a low-level source tape (e.g., some amateur recordings). It could also result in clipping the playback amp or in the subsequent analog chain in the case of an inordinately high-level source tape (e.g., early Atlantic Recording Studios rhythm and blues recordings).
For these reasons, setting the recording level for material without reference tones requires the optimization of levels throughout the analog signal chain for maximum signal-to-noise ratio without risk of overload or clipping (overloading) the signal. The engineer must determine the average level of the source tape and adjust gain stages appropriately, while observing the principles of unity gain. The optimal recording level for digitization is as close to digital zero as possible, without clipping.
2.1. Guidelines or methods for setting playback curves
Most tapes conform to standard playback curves, either NAB or CCIR. The CCIR curve is typical in Europe, and the NAB curve is used in the Americas. The playback-machine curve should match the source.
To determine the proper playback curve, the engineer should take note of documentation included on the tape or with the tape box and use EQ-alignment reference tones provided at the beginning of the tape.
If there are no EQ reference tones on the tape, the engineer should mount a standard alignment tape conforming to the presumed general EQ curve and align the playback deck’s EQ to that tape. The source tape should then be mounted, and the azimuth should be adjusted to match that tape. Determining the proper playback curve of a tape with no documentation or tones can be difficult.
If the EQ tones are included on the source tape, the playback equalization should be set to play back the tones as nearly as possible to the same level. If there is a large discrepancy in this result, the operator should suspect that the overall EQ curve is different from that in the player. Switching to another type of EQ curve may allow the tones to be adjusted to the same level.
2.2. Guidelines or methods for making slate announcements
Slate announcements are considered metadata. It is generally believed that slate announcements are not needed if it is possible to embed metadata within digital preservation copies. The need for proper identification and labeling cannot be overstated, and the methods for retaining metadata will depend on destination media.
Disagreement: One engineer cautioned that he did not think enough discussion had occurred during the roundtable to regard the substitution of embedded metadata for slate announcements as an optimal practice. When slate announcements are used, he suggested, they should be well separated from the audio content being preserved.
2.3. Playing the source tape
See 1.2.1 through 1.2.5.
2.4. Monitoring aurally and with test equipment for anomalies
Experienced listening and proper test equipment can confirm anomalies.
2.5. Monitoring physical playback mechanism
Participants concurred that playback and recording mechanisms should be monitored throughout the transfer process.
2.6. Take-up reel wind type (A/B wind, heads out, tails out)
Optimal practice: When the recording goes in only one direction, store the tape tails out. When the recording goes in two directions, store it with side-A heads out. (This also results in a tighter wind.) Experiment with A/B wind only when there is poor packing of tape for long-term storage. Some European tapes are stored using a reverse wind.
3. Post-Transfer Quality Control
3.1. Methods or techniques for real-time spot-checking
Participants suggested creation and use of a “transfer confidence index” to assist the engineer in designating the quality of the transfer in the metadata file. A “transfer confidence index,” refers to an assessment of the quality of the transfer to guide future use of the transfer and disposition of the original media (though it is very rarely recommended that original materials be disposed of).
3.1.1. Aural monitoring
Monitoring systems should achieve accurate reproduction to allow for proper evaluation of quality. They should be capable of accurately monitoring the highest-resolution source employed in the preservation signal chain. In some cases this will be the playback medium; in others, it is equipment farther downstream in the signal chain.
Recommendation: One engineer suggested that one characteristic of the monitoring system (including the room itself) should be a flat frequency response up to 20 kHz in the listening position.
3.1.2. Peak/average level meter
Optimal practice: Use peak metering for all level measurements.
Disagreement: Participants failed to reach a consensus on how best to set levels in a disc with high-amplitude transients—namely, they wondered whether “overs” are ever acceptable. One view holds that they are not. Some believe that decisions concerning whether it is easier to remove these with software filtering or to raise low-level signals must be made in conjunction with the rest of the music and available processing capabilities.
3.1.3. Waveform display (amplitude over time)
Use a waveform display to confirm what you hear. Predigitization test equipment should not be in the recording chain but parallel to it. Postdigitization tools are acceptable.
The following instruments may be useful in the post-transfer listening suite; in this case, their purpose should be to confirm an anomaly that has already been noted in the documentation accompanying the archived program. These instruments are also recommended for inclusion in the pretransfer suite to identify problems such as azimuth drift, dropouts, and unusual frequency-response characteristics. The transfer technician should note such problems in the metadata.
3.1.4. X-Y scope
The X-Y scope is essential for setting azimuth in stereo, 2T mono, or multitrack pairs. It is also useful for analyzing phase.
3.1.5. Correlation (phase) meter
This tool can be used in analyzing two different signal sources as in stereo, two-track, or multitrack pairs.
3.1.6. Frequency analyzer
A frequency analyzer can be useful in identifying and confirming the frequency information of a recording. It can help locate and identify components of hums or other repetitive anomalies.
3.1.7. Spectrum analyzer
A spectrum analyzer can be useful in identifying and confirming the spectral-frequency information of a recording. It can locate and identify components of hums or other repetitive anomalies (e.g., turntable rumble) that may be present in the source recording or the playback chain.
3.2. Methods or techniques for automated checking
Advanced error detection: Participants suggested that a software-based system could be employed as an aid for error detection in files and that a combination hardware/software solution may be used to flag errors in physical digital media. For quality-control purposes, data integrity should be checked at some point after the transfer is completed.
ANALOG AUDIO DISCS AND CYLINDERS
4. Pretransfer issues
4.1. Inspecting, preparing, and cleaning the source materials
4.1.1. Determining the composition (shellac, lacquer/acetate, aluminum, vinyl, metal parts)
Identify disc composition before cleaning or outsourcing for transfer in order to specify cleaning methods and to anticipate problems. Experienced engineers can readily identify disc composition.
Recommendation: Research on the various formulations of shellac discs is needed. Advanced materials science could be helpful in this area.
4.1.2. Methods or techniques used to address specific physical problems
184.108.40.206. Physical damage (broken or missing pieces)
Optimal practice: If a glass or shellac disc is broken, reconstruction can be attempted. If a broken disc cannot be satisfactorily restored because of deformations or missing shards, keep all the pieces for potential future signal-reconstruction techniques. Consider special housing for badly broken discs.
220.127.116.11. Broken substrate
Some discs can be played in spite of broken substrate.
18.104.22.168. Loss of lacquer (separated lacquer)
Save all pieces for possible future signal-reconstruction methods.
22.214.171.124. Cracks and scratches
Depending on severity, play through them and make notes for metadata.
Some warped discs can be played at a much lower speed than originally intended; in other cases, additional tracking force can be applied. The deciding factor is the severity of the warping. Take note of speed for metadata and note if any EQ is applied. Do not apply playback EQ curves during low-speed transfers.
A vacuum turntable works with vinyl but not shellac. It is also an expensive, highly specialized piece of equipment.
126.96.36.199. Groove abnormalities (no lead-in groove)
If the disc has no lead-in groove, record the sound of the stylus dropping. Keep trying until you get as much information or signal as possible. Document all abnormalities.
188.8.131.52. Shallow grooves
Experiment with styli of different sizes and shapes. Some standard groove geometries can be inferred from the record label, brand, or type. For example, Edison Diamond Discs are typically very uniform and are ordinarily played with the same shape and size stylus, but Pathé Sapphire Discs from the same era require a different-size stylus. Yet another size is used for pre-stereo microgroove records, and so on. Try a commonly recommended stylus first. If there are problems with noise or distortion, apply an array of alternate choices.
184.108.40.206. Nonconcentric grooves
Nonconcentric grooves are a manufacturing-related problem that causes wow. No post-transfer wow-removal tool is currently available.
220.127.116.11. Hole punched off-center
This problem is similar to nonconcentric grooves. To correct both problems, mount the record on a turntable with a smaller-than-normal spindle to allow some adjustment of the centering. Alternatively, use a turntable with an adjustable spindle. Centering of the record can be done visually to a high degree of accuracy.
18.104.22.168. Worn grooves
Optimal practice: Careful stylus selection and careful cleaning are essential. If dictated by the condition of the disc, make multiple dubs and listen to them to determine the best recording.
4.1.3. Methods for removing or mitigating surface problems such as topical debris, exudation, and presence of oils or fingerprints
Optimal practice: When transferring an audio program from an analog disc, clean the disc first, unless cleaning will damage the media.Alcohol should not be used to clean any disc records except Edison Diamond Discs and metal parts.
When cleaning, start dry: Use dusting and vacuuming. Then, in descending order, try antistatic brushing, deionized water, “pure” (i.e., fragrance- and additive-free) liquid soap, and scrubbing. Mechanical disc washers are also commercially available.
Disagreement: Some engineers advocate using a stylus as a cleaning tool. This method can work with shellac records but can subject the disc to degradation. Special caution is needed when using a stylus for vinyl records if they have not been properly cleaned first. Some engineers believe that it is advisable to use a stylus only to clean metal parts. All agreed that it is best to remove as much debris as possible before passing a stylus through the grooves.
4.1.4. Choosing cleaning solutions
Cleaning solutions must not leave residue or cause damage. Research and development must verify a solution as safe before use. Disc recordings that cannot currently be transferred because of technical limitations should be stored until appropriate treatment is available.
4.2. Configuring and calibrating playback equipment
4.2.1. Choosing a stylus
Stylus choice is one of the most important factors in transferring audio from disc. Most audio preservation engineers choose the proper stylus through a combination of experience, educated guesswork, and expert listening. Participants noted that a stereo lab-microscope and a good groove microscope with a reticle calibrated in 1-mil increments can reveal a great deal about groove condition and stylus requirement. The microscope does not, however, replace the role of expert listening in finalizing the choice of stylus.
Disagreement: Can a stylus be selected on a purely scientific basis or is it a subjective decision that is based on expert listening? Or are both required? What methodology can be developed for blending the two approaches? Time pressures or lack of equipment for examination and analysis of the groove may make it necessary for an expert to rely on a subjective method. But, as one engineer offered, one “can envision a procedure where there is some preparatory visual groove/stylus matching, using a microscope. By building up the needle profiles and storing them as data files, certain images could be taken of the disc and then correlated with the stylus types. This gives an approximate result and a range of styli to be considered.”
4.2.2. Determining playback speed
While approximately 78 rpm was the predominant speed of early discs, there was no standard disc speed until late in the 78-rpm era. Many of the same guidelines discussed under tape transfer apply to disc transfers. Careful listening is essential. Use a variable speed turntable. Note disc speed and variations as documentation to accompany the digital preservation copy. On the basis of their knowledge of program content, music or subject specialists may be able to help identify speed.
Recommendation: Compile and document available research to provide guidelines for determining playback speed.
4.2.3. Setting tone arm tracking force and height
Tracking force, also called stylus force, is the static force between a stationary record and the stylus tip, in the playing position. With traditional, stylus-based playback technology, the optimal adjustment of tracking force is a trade-off between record wear and dependable tracking and tracing of the groove. To minimize record wear, use the lowest tracking force that reliably maintains continuous contact between the stylus tip and the groove, as indicated by a minimum of audible distortion in the extracted audio signal.
Phonograph-cartridge manufacturers usually recommend a range of tracking-force values over which a given cartridge may be expected to perform within its rated specifications. However, cartridges are typically characterized for playback of stereo LPs and 45-rpm records that are in good physical condition, and the tracking-force values depend on equipment properties such as tonearm mass and damping.
For any turntable setup, the minimum practical value of tracking force is influenced by the prevailing conditions of groove-modulation level, record warp, and the motional dynamics of the overall pickup-and-tonearm system.
When using a stereo phonograph cartridge to play vertical-cut discs, 78s, or other pre-microgroove record formats, the transfer engineer should be alert for tracing or tracking problems, displayed as audible distortion. As the tracking force is reduced, the stylus eventually begins to lose contact with the groove. The signal becomes “fuzzy,” then gradually more “buzzy,” until the stylus ultimately fails to track the groove (it “skips”). Heavily modulated, warped, or locally “bumpy” records may require increased values of tracking force to improve tracking and extract a cleaner signal.
Modern, high-compliance phonograph cartridges are typically rated for a maximum tracking force of only a few grams. If loaded at significantly higher forces, the stylus-suspension system in many cartridges will “bottom out,” making the cartridge inoperative when the stylus cantilever recedes into or strikes the cartridge housing. One workaround for this limitation, when the goal is to track warped or “bumpy” records, is to increase the stiffness of the stylus-suspension elastomer in the stylus assembly. Another approach is to mount the cartridge in a “floating” (i.e., pivoted and spring-loaded) headshell, to reduce the effective mass of the pickup assembly.
It is important to set the tracking force high enough to avoid mistracking—a condition in which the stylus leaves the groove and may then return, slamming into the record surface. Mistracking can seriously damage records made of softer materials, such as vinyl, lacquer, or wax.
4.2.4. Setting antiskate compensation
Pivoted tonearms (as compared with linear-tracking or radial-tracking arms) experience a phenomenon called skating when playing a disc record. This is caused by the bend, or dogleg, that is designed into the tonearm. The bend is there to reduce the incidence of lateral tracking angle error, which is a natural consequence of using a single lateral pivot point in the tonearm. The skating force causes the stylus to bear with greater force on the inner groove wall (stereo left channel) than it does on the outer groove wall (stereo right channel). This can result in more distortion in the right channel when playing stereo records. The antiskating compensation has been added to modern turntables to apply an outward torque to the tonearm, which counteracts the skating force, resulting in nearly even tracking forces on both the inner and outer groove walls. Test records are available for use in adjusting the antiskate compensation. Alternatively, as a temporary corrective procedure, the technician can adjust the compensator to reduce groove sticking or skipping.
Some believe that this issue primarily involves stereo recordings. However, the online discussion revealed that physical problems caused by not riding in the center of the groove are the same for all records. To many, standard preservation dictates that both groove walls of a mono recording be preserved if transferred using a stereo cartridge; the right channel from a mono recording cannot be ignored. A bias toward vinyl stereo recordings might be perceived even in the mention of an S-shaped tone arm. LC engineers stated that they use straight tonearms for most of their nonvinyl transfers.
4.2.5. Monitoring aurally and with test equipment for anomalies
See section 1.2.3.
4.2.6. Setting level gain throughout signal chain
See section 1.2.4.
4.2.7. Setting record level
See sections 1.2.5 and 3.1.2.
5.1. Methods or techniques for setting playback EQ curves
If the type of recording EQ curve for a disc recording is not known or cannot be assumed with a high degree of assurance, or if no recording EQ was used (as with the early acoustic recordings), it is generally recommended that the disc be transferred “flat” with no playback EQ curve compensation. If necessary, these recordings can be adjusted post-transfer. The EQ used should be noted in the documentation. Not all phono preamplifiers allow for the EQ to be switched off, and switching off the EQ is not easy to do with most phono preamps. The Recording Industry Association of America (RIAA) playback curve is often designed into the constant-velocity to constant-amplitude compensation that is intended to be used with magnetic (dynamic) phono cartridges. If one uses a flat amplifier instead of the RIAA magnetic preamp circuit, the result is a very tinny sound from a magnetic cartridge. Although this can be useful for preservation, it is not recommended for standard playback. A separate preamp velocity-compensation equalizer is required to convert the magnetic cartridge response to constant amplitude that does not have RIAA compensation added as well. This circuit is rarely found in any commercial preamps.
5.2. Methods or techniques for recording test tones
There is no need to record a series of test tones onto the digital preservation copy unless those tones are part of the original program on the source tape. Digital recording is inherently flat in frequency response at all signal levels, and no digital EQ is used, so there is no purpose served by the inclusion of test tones.
5.3. Methods or techniques for making slate announcements
See section 2.2.
5.4. Starting the recording device
Start the recording device first, and then start the playback.
5.5. Playing the disc
The group did not want to discuss playing discs wet versus dry.
Disagreement/Further research: One engineer who opposed playing discs wet stated by e-mail, “All commercially made records are designed to perform perfectly well when played back in their dry condition. The wetting of the surface is more likely to introduce undesired artifacts such as a change in the damping of the playback stylus cantilever, which will result in an improper frequency response. Furthermore, the wetting can trap dirt on both the record and inside the playback cartridge. I do not recommend wet playback under any circumstances, with the possible exception of playing non-commercially made aluminum disc records.”
In contrast, an engineer who has had success with wet-disc transfers stated, “I am not convinced nor have I seen any evidence that wet playing of commercially made shellac 78-rpm discs or lacquer-coated discs introduces more unwanted artifacts than it removes. Personally, I have had success in wet transfer of acetate-coated discs and shellac 78s, resulting in a lessening of surface noise. I feel this is a subject in need of more research.”
5.6. Monitoring aurally and with test equipment for anomalies
See section 2.4.
5.7. Monitoring physical playback mechanism
Roundtable members agreed that ideally, all transfer equipment and processes should be monitored from beginning to end.
6. Post-Transfer Quality Control
Participants suggested use of a “confidence index” to assist engineers in designating the quality of the transfer in the metadata file.
6.1. Methods or techniques for real-time spot-checking
In addition to aurally monitoring, one should use a peak/average level meter and waveform display (amplitude over time). Software tools that incorporate an X-Y scope, correlation (phase) meter, frequency analyzer, and spectrum analyzer should also be used.
6.2. Automated checking
Advanced error detection: See section 3.2.
7. Other Issues
7.1. Choosing sampling rate and bit depth
Recommendation from George Massenburg: 96 kHz, 24 bit, linear PCM files are the minimum standard for digital audio preservation files.
Reasoning: The emerging standard is the DVD, on which audio is 96 kHz. Storage space is becoming cheaper all the time. IASA has embraced 96 kHz as its specification, and we want to be interoperable with Europe. At least one roundtable member recommended sampling at 88.2kHz in order to best produce a 44.1kHz file. It was pointed out that good digital converters can now down sample 96 kHz to 44.1 kHz much better than they used to, with negligible loss of audio fidelity. And the point was made that 44.1 kHz would be used only for access copies, not for preservation copies. A minuscule loss of audio quality in such copies would be acceptable to most attending the roundtable.
Although there was some discussion that preservation of lower-fidelity media (e.g., oral histories on cassette) might not currently benefit from 96kHz/24 bit, 96kHz/24 bit is nevertheless recommended for all media and all content whenever possible. One cannot foresee how the transferred audio may eventually be repurposed and how future restoration technologies might be used.
One engineer noted that as of early 2004, the AES had a proposal on the table recommending setting a new digital preservation standard at 192 kHz/24 bits.
To accommodate limited budgets or other resources, preservation reformatting often involves compromises between best practices and acceptable results. Roundtable participants suggested that guidelines be created to assist in making decisions that result in compromises to the quality of the product.
- Define and spell out compromises and trade-offs.
- Provide guidance to all institutions, regardless of size and budget, on how to deal with any needed compromises between best practices and acceptable results.
- Store recorded media that cannot currently be transferred because of technical limitations until appropriate treatment is available.
- Develop and define new standards for the archival community.
- Develop guidelines on how to make judgment calls.
8. General comments from the listserv discussion
Recommendations: Create the following three resource documents:
- a list of suggested equipment that should be available to perform routine digital audio archiving tasks at a 96 kHz/24 bits standard;
- “Sources of Equipment and Supplies for Audio Archives” (this document should be updated regularly); and
- advice for small archiving institutions on how to obtain grants and other funds for digital preservation.
2 For more information on problems faced when playing back open reel tapes, see D. Michael Shields, Dennis D. Rooney, and Seth B. Winner. 2003. Considerations in the Playback of Archival Analog Magnetic Recordings with Wide Recorded Tracks. ARSC Journal, 34(1): 48–53.
3 Frequency-alignment tapes may be purchased from Magnetic Reference Laboratory on the Web at http://home.flash.net/~mrltapes/ or by phone at 650-965-8187. Tapes may also be purchased from JRF Magnetics on the Web at www.JRFmagnetics.com or by phone at 973-579-5773.