Digitally Restoring Old Recordings

On the Cheap

by John M. Clement, PhD

 

Old recording restoration is a very complex issue which can cause you many headaches and cost a lot of money.  Fortunately, with the advent of fast low cost computers and software it is possible for the average person to restore many old recordings with excellent results.  I have attempted to provide a guide to how this can be done with available software at a modest cost.  In the end the quality of restoration is determined by how much work and money you are willing to invest in the project.

 

The recordings from each era need different treatment and pose many different issues.  My concern with this issue originally centered on a set of old 78 rpm dance recordings that our group was using, while beautiful showed signs of age.  The bulk of recordings belonged to an individual who had loaned them to our group.  We needed to return them to this individual, so a master copy on open reel was made of the recordings.  The original recordings are no longer available, so only the open reel copy could be processed.  This master copy was made with no processing, and only cassette copies were used on dance nights. As a result we had a good copy to start with that could be processed later.  It turns out that this is the key to being able to restore these old recordings.  Any processing in the early stages complicates or prevents restoration later.  Digital restoration needed both 78 rpm and tape restoration techniques.

 

While software can be valuable, the best restoration must begin with good quality equipment, and good quality sources.  For records this means a good quality turntable, magnetic cartridge, and preamp with proper equalization.  Records should be cleaned prior to being recorded.  This is best done by a machine which vacuums the cleaning fluid out of the grooves such as a Nitty Gritty cleaning machine.  Tapes need a good quality tape player with the proper Dolby noise processing for Dolby encoded tapes.  Software can compensate for some missing elements, but it will in general not be as good as the correct equipment.  A properly designed preamp will provide accurate equalization as well as the correct impedance match with the cartridge.  This will result in the best frequency response.  Using a microphone input will not provide the correct match to the magnetic cartridge, and even with software simulation of the RIAA curve, the frequency response will probably not be correct.  Click and pop removal software can never completely fix all such problems.  Cleaning the records first will remove many of the large clicks and will also help prevent record skipping.  The software will then do a better job on the residual noise.   Inexpensive equipment will result in wow and flutter (speed variations) as well as extraneous noise.  Speed variations can usually not be easily fixed by software, and high levels of noise reduction can remove some of the music, so good equipment is the best investment for satisfactory results.

 

Features Needed for Restoration

 

A variety of features can be used for restoration of old recordings.  These include click and crackle reduction, peak restoration, noise reduction, hiss reduction, equalization, track delay setting, time base correction, dynamics expansion, and individual track editing.  Only some of these are needed, depending on the recordings.  Time base correction, for example, can be used to correct the pitch of old 78 rpm recordings which may have been recorded at a variety of speeds, but which you may have played on a single speed player.   Time base correction may not be needed for other recordings.  Dynamics expansion can be used with all old recordings to expand excessively compressed sound.  This will also increase the effective dynamic range.  An adjustable filter can be useful in removing low frequency pops, but often this can be done with an equalizer if it has 24 or more bands. High pass, low pass, and notch filters are useful when the source contains noise at specific frequencies. A spectral editor is an invaluable asset for removing audible noises which can not be removed automatically.

 

Tape Restoration

 

First you need to start with the correct type of tape deck.  Cassettes are fairly easy because they are all manufactured to the same standards, but open reel is more difficult.  Open reel players generally came in two kinds, half track and quarter track.  While half track tapes can be played on a quarter track machine, the alignment will not be good and the tape will not have good frequency response or the lowest noise.

 

Tapes can generally be restored with just four features, hiss or noise reduction, equalization, track delay setting, and peak unlimiting.  Most tapes have hiss which can be obtrusive, so a hiss or noise reduction can make older tapes sound much better.  Sometimes tapes may skew, or the recording machine and playback machine may have a slight misalignment.  This can cause one track to be slightly delayed compared to the other.  This is obvious when you examine a part of the tape with only sound in the center.  By expanding the scale you can see that the peaks do not line up.  This can cause the image to be slightly off center and can cause problems with playback in mono.  It can only be corrected by slightly delaying one track until the two tracks line up perfectly or are in phase.  Tapes that have been played frequently will suffer some loss in highs.  This can also happen if the alignment of the playback machine does not match the recording machine.  This may be corrected with an equalizer.  If the tape has been recorded at too high a level, peak unlimiting may be useful in recovering the sound in the peaks.

 

Sometimes there will be momentary or even length drop outs in tapes.  Momentary dropouts can be simply cut out if they are very short.  Extended dropouts require both equalization and volume changes.  Often this will happen in only one channel so the program must be capable of performing different operations on each channel.

 

78 rpm Restoration

 

78 rpm recordings can use all of the features that are available for correction.  The first problem is that the recording machine may not actually have run at 78 rpm.  The early recordings varied quite a lot in speed, but later ones should be close to the standard speed of 78.25 rpm.  If the program allows you stretch the recording and raise or lower the pitch, this can be used to correct for this problem.  It is even possible to skip buying a 78 rpm turntable, and instead play the 78s at 45 rpm and then compress the time base to bring them up to the proper speed.  If you do this the re-equalization curves must be recomputed.  Of course you could buy an expensive variable speed player instead of using a program. 

 

The second problem with 78 rpm recordings is that they all had different equalizations.  The recording would have been recorded with pre-emphasis or boosted treble and reduced bass.  This allowed the music to remain louder than the noise and prevented grooves from being too wide.  The treble would then be reduced and bass boosted on playback according to a de-emphasis curve.  The actual curve could vary according to the record company or even the producer.  The modern LP recording uses only one standard equalization called RIAA.  If you play back a 78 through a modern amplifier, the de-emphasis curve will not be correct.  With a multiband equalizer you can easily re-equalize the 78.  To accomplish this you may need a 10 band equalizer that has a wider range than the usual 12 db.  I have calculated a series of curves that you can use in Appendix B.  The data for these curves were taken from various sources on the web.

 

All 78 rpm records have severe pop, crackle, and hiss which may detract from the listeners pleasure.  Since the pops are often due to something located on one groove wall, it is possible to have the program remove just noise that appears in one of the stereo channels.  Unfortunately such programs are not available currently at a modest price.  One person ran his recordings through a Dolby logic processor and then sent only the center channel to his computer.  Effectively the processor put the noise in the right and left channels, but the music in the center channel.  If you have a program which can edit channels individually it is possible to locate the large pops and then just clear the channel for the duration of that pop.  It is always wise to avoid the problem to begin with by washing your records in mild detergent.  Do not get the label wet, and never use alcohol on a lacquer record.  Directions for this process are available at http://www.tdl-tech.com/restore.htm Finally the remaining pops, ticks, crackles, and hiss can be removed by good noise reduction programs.  Popular music is fairly easy to restore because it remains at a high volume, but classical recordings can pose severe difficulties.  Peak unlimiting may be somewhat useful when restoring very loud recordings.

 

LP Restoration

 

Vinyl LPs manufactured after 1955 generally only require pop, tick and hiss removal.  Earlier recordings may have used different equalization curves.  This can be solved using a multiband equalizer.  Of course remember to clean your LPs before using them.  A Nitty Gritty machine does an excellent job.  Sometimes LPs have defects in the master tapes.  This includes dropouts, skewing, and phase shift.  So being able to handle tape restoration might be important.

 

FM broadcast restoration

 

FM stereo broadcasts may be difficult to restore because the standard noise reduction routines are designed around removing clicks and pops or steady hiss or hum.  The noise in FM is mainly in the stereo or L-R part of the signal and can generally be reduced by reducing the separation or by making the signal mono.  One way of reducing noise would be to change the FM signal phase from stereo to L+R and L-R.  This can be accomplished by a stereo processor such as Virtos.  It must be set for max L and reversed R on one channel and equal L+R on the other.  Once this is done, noise reduction can be applied to just the L-R channel. Another possibility would be to filter the L-R channel.  The signal can be returned to stereo by setting  max L and reversed R on one channel, but reversed L and max R on the other.  The noise reduction programs could be easily designed to operate on the L-R channel only with some optimization for FM broadcasts, but I have not seen any programs designed to do this.  Noise reduction in L-R channel should have little effect on the overall tonal balance, but the stereo separation will be reduced somewhat.

 

Mono and near mono Recordings

 

Sometimes mono recordings have been processed to simulate stereo.  While this may be more pleasing to some people, others may find it to be poorer in quality than strictly mono.  Often just combining the two channels will not undo this processing.  Some recordings have simulated stereo by just phase shifting one channel relative to the other.  When this type of processing has been done, combining the two channels to get strict mono will cause cancellation at various frequencies and the sound will be very poor.  I have encountered recordings with as much as 0.35 ms phase shift between the channels.  This may be easily fixed by phase shifting one channel relative to the other so that the peaks exactly line up.  Then when the channels are combined you get a good mono recording with slightly reduced noise.  In addition residual noise is much more audible in stereo when it comes from a different location than the music, so restoring mono can improve the quality of reproduction markedly.  Sometimes phase shifts can also happen with material which has been recorded using just two mikes.   By phase aligning the two channels you may achieve better reproduction.  If the music is to be played in mono, then aligning the phase will improve the reproduction markedly.  Older professional stereo recordings were always engineered to sound good in mono, but some foreign, small label, and recent recordings may not have been adjusted for good playback in both mono and stereo.  If the recording is to be played at a large public event it is quite likely that it will be played in mono, so you would be well advised to pay attention to this problem.

 

An extreme phase problem sometimes happens with tapes in which one channel is reversed compared to the other.  You can recognize this because the stereo channels are mirror images of each other.  For this problem you only need to invert the phase of one channel.  When this type of recording is played in mono, the sound will essentially vanish.  The Virtos stereo processor can be used to take care of these problems, and some of the available software has both phase shifting and reversal ability.

 

Software

 

There are a number of programs available that are very good at removing noise, ticks and pops from old recordings.  Each has different features that make them desirable, and there may not be a single program that suits all needs.  Professional products are available that may do a better job for many thousands of dollars, but only consumer oriented products have been considered under $1000.  The reviewed programs are arranged approximately in order of price, and only programs that have been personally previewed by the author are considered.  Any program you buy should be previewed before buying it, because it may not be capable of handling your recordings.  Most software can be downloaded as a trial version before buying, but some notable software such as Pinnacle were not available for preview.  Sometimes stores will sell these programs at a deep discount, so it pays to watch the sales.  For this review I took an old 78 rpm recording with numerous ticks and tried to clean it on all products using default and automatic settings only.  If these settings did not work, I tried advancing the sensitivity until there was an effect.  While it may be possible to get much better results by carefully tailoring the settings, many consumers will only use the defaults.  The results are in Appendix A.  All of the noise reduction packages can produce undesirable sounds or dulling of highs when set to maximum strength, so caution is needed in using them.  Since a large number of products were reviewed, it is possible that I missed some features or controls, so you should always try them yourself.  You may also need a good CD writing program.  While there are many good ones, I can recommend Stomp inc. Record Now Max which is capable of burning and ripping music with CD-text.  In addition it can create mixed mode CDs such as CD+ which can be played on a normal audio CD player, but can also have data that you can display on your computer.  Record Now Max does not handle mono wave files correctly, but it works well for stereo wave or MP3 files.

 

As of this writing Dec, 2003, there seems to be a convergence as more products are adding features from other products.  Notably missing or scarce features are DC-text, and multimode CD support with combined sound and text.  In addition simple time delay of one channel usually requires wading through other controls.

 

Some of the cited articles on the web have good instructions for performing noise reduction.  However, a few cautions are in order.  Usually the most noise reduction introduces some audible problems that may be worse than the original noise.  You must always carefully adjust the noise reduction until it is satisfactory, but be very careful not to go beyond that point.  Always listen to the result and the original to see if there are any audible problems.  Sometimes format conversions can cause severe problems.   In particular on Windows machines conversion from stereo to mono may raise the overall level and cause the result to be overloaded.  To compensate for this reduce the output level by 6 dB before creating the new file.  This is a problem when creating a mono wave file from a stereo file, but it did not occur when creating a mono MP3 file.

 

All plug-ins were evaluated using Magix Audio Cleaning Lab.  It has facilities for either processing the file or for using the plug-in in realtime so you can A/B the results.  Some plug-ins would not work in realtime, and this was noted.  I can not tell if this is a plug-in problem or a Magix problem.

 

Acona Acoustica 3.0

 

This audio editor and recorder sells for $30, and at that price is probably one of the cheaper editors available.  It has click and noise suppression.  The noise suppression uses a noise print.  The click suppression, tested for version 2.25, has two adjustments, but after trying maximum reduction, it seemed to make little difference to my test file.  While it is capable of introducing a delay, there is no way to do this for small enough times.   It has the usual array of reverberation effects and also includes the ability to reverse a file.  It is possible to select just one channel for editing.  It also allows time base changes.    It has a high frequency enhancement module, and a 6 band parametric equalizer.  With only 6 bands the equalizer may not be adequate for re-equalization.  There is no peak unlimiter.  Acoustica has both dynamics compression and expansion.  The zoom feature in Acoustica only works horizontally, so you may have difficulty finding individual pops to delete.  Acoustica does include facilities for handling plug-ins, so it can be improved by purchasing plug-ins.

 

Magix Audio Cleaning Lab 10 (ACL)

 

ACL is my choice for the best consumer audio restoration programs. The new 2007 product is out with a few small enhancements and fixes, but the 2006 version is reviewed here.  The 2007 version includes a parametric equalizer. In 2005 the most obvious change was the interface which makes more features available as buttons on a top row, and a low contrast color scheme with much harder to read text.  For version 10 marker marker declicking has been added to the denoiser and declicker, but the declicker does not approach the click suppression of Sony Noise Reduction.  DVD, chorus effect, simulated tape sound, spectral cleaning to remove coughs, and database support (commercial CD identification) have been added.  The chorus and tape simulation are both available as freeware from the web.  In addition ACL can now invert the phase of one channel to correct for improper tape recorder or phono connections, but mono still raises the overall volume by 3dB.  Version 10 has added a bug, that the individual track settings are no longer independent from the overall settings.  This is breaks older files, and is very inconvenient.  Magix Audio Cleaning Lab 10 is available for about $30 to $40, but I have seen old versions free after a rebate.  At $20 you may wish to buy it without trying it.  For the novice, it has a large set of very usable features that can be used to acquire, enhance, and then save the results on CD.   It is the only reviewed product that can save tracks in CD-text and CD extra format. ACL does not store the mandatory performer field CD-text, so some programs might not read the CD-text  It has a pop and tick reducer, crackle reduction, noise reduction, peak restoration, hiss reduction, filters and an equalizer.  In addition it contains ambience enhancement and a highs restoration enhancement.  You may also add DirectX and VST plug-ins which are available from other manufacturers.  Time base conversion can be used to correct for recordings which were not made at standard speed.  It allows separate editing on each channel, and time delay is possible.  When doing different delays for different sections, you must edit from left to right, as any changes affect all sections to the right.  Finally it allows conversion to MP3.  It has dynamic compression, but limited expansion capability.  In addition to timed recording,  ACL can automatically shut off after 6 seconds of silence, and automatically divide the recording into regions or tracks.  This is extremely useful for converting LP recordings of popular music.  It sometimes has problems with classical music with long pauses between movements or very quiet passages.  The noise reduction features have apparently been designed by Algorithmix, so ACL and Sound Laundry should both have similar performance.  It is possible to easily slow down playback by right clicking on the playback button.  This is useful in locating pops to remove manually.

 

The new spectral cleaner is a good spectral editor.  It allows you to remove noises by selecting a section of the spectrum which contains the noise.  Then you can monitor the results and reselect until the noise is gone or much reduced.  It allowed me to remove coughs from a live piano concert without affecting the sound of the piano.  The final result was much more listenable with the coughs either totally suppressed or vastly reduced.  The program figures out what part of the sound is performance, and what is unwanted noise. The equalizer also has a filter called the Sound Cloner. The Sound Cloner can generate almost any filter shape you wish by drawing the characteristics. This supplies both low, high, and notch filters as needed between 40-20kHz. However, it can be inconvenient when you are trying to suppress a single frequency that you can hear. The free BlueLine plug-ins have a more convenient filter which can also suppress subsonic noise which robs amplifier power.

 

The tick and crackle reduction are very simple and have a single strength adjustment.  There is an additional marker declicker which applies maximum strength only to identified clicks.  Since there is only a sensitivity contol it may not be possible to adjust for specific recordings.  It worked fairly well on the Algorithmix demo file.  The marker declicker leaves audible low frequency pops when it removes large ticks. A workaround is to use the markers to find the clicks, and then use the spectral cleaner to suppress them.  ACL did as good a job of completely removing many large ticks on some test files as Sony Noise Reduction, but it was inconsistent.  Either it almost completely removed a tick, or it left a very audible tick.  Sony Noise Reduction reduced all large ticks uniformly, but some were softened to be an almost inaudible low frequency pop.  The marker feature should allow maximum declicking without affecting other regions of the recording.  ACL does not remove low level ticks and crackle well.  It seems to work best with records which have been directly digitized, but is not very good for recordings that have gone through an intermediate medium such as tape.  The hiss reduction and noise reduction are extremely effective when adjusted properly, and they have a number of hidden adjustments.  The novice can be content to only adjust the strength, while the advanced user can use the hidden adjustments to precisely remove noise.  The equalizer has 10 bands which do not appear to be equally spaced.  It is possible, but extremely difficult to re-equalize old recordings as you have to estimate the actual corrections, the gain value is not displayed, and the displayed equalization curve is extremely inaccurate.  The equalization settings can be saved along with other settings for mastering such as compression, stereo image, and real time DirectX plug-in settings.  The highs restoration is included with the equalizer.  All of these adjustments are made in real time and the original wave file is not changed.  As a result it is easy to do A/B comparisons of the original and modified sound by turning the processing on and off.

 

The stereo channels can be split into separate parallel channels, and then edited separately.  The time delay may be adjusted by just grabbing and shifting one channel.  Old monophonic recordings can have a pop removed in one channel by using the ability to divide the recording into sections and then by adjusting the gain for each channel and then recombining them into mono.  The Apply all realtime effects option can be used to recombine the two channels into a single stereo file.  When used with the Virtos Stereo Processor this allows you to use ACL for quieting the stereo signal in an FM broadcast.  The split channel feature is valuable for fixing problems in just one channel such as a drop off in frequency response on one channel of a tape.  You can quickly adjust just the one channel many times, and monitor various portions of the recording.

 

ACL has a very neat feature that can make editing easy, but also poses other difficulties.  Recordings can be sliced into individual sections and then each section can be treated separately, or one treatment can be applied to all sections.  You could create a variety of wave files for your old 78 rpm recordings and then bundle them into several projects according to the brand and age of recording.  Then all of them could be equalized and turned into mono using the same settings.  Each recording could also be individually denoised without affecting the others.  A pop or any extraneous noise can be removed manually by cutting on either side of the pop and then by reducing the volume, increasing the pop reduction, or cutting the bass for that section of the recording.  However, cutting tracks at very precise points is difficult because the cursor is positioned in the middle of a fade.  While this produces completely inaudible splices, it makes cutting difficult at high magnifications if you do not know how to do it.  You must estimate the cut location, reduce the length of time for the fade, and then slide the cut to the proper location.  ACL would be much easier to use if the fade length were shortened when viewing the wave at high magnification.  Finally the results can be saved as either one large wave file or individual track files.  Unfortunately there is no way to recombine individual sections once they have been split.  You can recombine all sections and create a new project with all changes made permanently by clicking on the poorly documented apply all realtime effects option.

 

DirectX plug-ins such as Noise Reduction 2.0 are usable by ACL.   One plug-in can be used for the whole project in real time or multiple plug ins can be used for individual sections.  Since Sound Forge used DirectX plug-ins for most functions, you can also utilize these features in ACL.  To use a plug-in to reduce a single noise peak it is possible to cut the section, and then use the plug-in on just the small section.  Unfortunately, ACL can not save plug in settings.

 

If you record a whole LP or tape, ACL can find the individual tracks and mark them.  Tracks can also be marked by just setting markers in long files and they can be rearranged by just changing the marker order.  The track markers and sections are completely independent.  The ability to easily rearrange tracks can be useful once you have decided which tracks to burn.  This feature would be even more useful if tracks could be declared dummy.  With this one feature you could easily make different CDs with different arrangements without having to create separate wave files for each track.  ACL automatically remembers the settings for each section and stores each noise print so you can easily set each section quickly and individually.  This feature makes ACL the easiest program to use for recording LPs and burning them onto CDs.  Unfortunately ACL does not give you the option of setting the CD label.  The CD label can be supplied by another CD burning program if ACL writes a CD+, but otherwise there is no labeling facility.

 

 

A severe problem is that ACL had was difficulty handling audio with a large number of long track labels.  A project with 19 tracks and an average label size of 40 gave an access violation and said the internal data structures may have been destroyed.  The workaround is to save the project as a wave file by applying all realtime effects.  Then reload the audio file and relabel the tracks.  Because Magix does not supply a file detailing the fixed bugs it was not possible to verify that this bug has been fixed.  *Another severe problem is that it gives a cryptic error message when you try to burn CD-text and the drive does not support that feature.  ACL 2005 will occasionally give an error message Cannot set recording speed -> drive not ready.  This seems to be associated with a problem in the .VIP file, and the supposed damaged CD will record with other projects.  Other projects will burn just fine and only just a particular project will consistently give this message.  If you exit from the project and click yes to save the project after trying to burn, the .VIP file is destroyed and you have to remake the project starting with the wave file.  This also seems to only happen with projects where you label the tracks.  The CD-text feature is very basic, and does not include the ability to specify artists, and it can not handle all foreign characters.    ACL can import tracks from CD-text CDs with correct track names.  ACL has difficulty when a large number of noise prints have been saved.  The noise prints at the end of the list become inaccessible, and there seems to be a limit on the number of noise prints.  If you accidentally give two tracks the same name ACL will then give them arbitrary names when you export them to MP3 files.  It would be much better if ACL would warn the user of the problem and ask for a solution, or just put (2) next to the second occurrence of a name.   Several bugs come from interactions between section cuts and other features.  *If you cut a file and move the right hand portion over to cover the left side, ACL has a problem with noise prints.  *ACL uses the concealed portion of the wave to make a noise print in the overlap region rather than the visible wave.  As a workaround take the noise print before overlapping the waves, or cut the undesired potion out.  If you make many cuts in your file and then set track marker automatically,  ACL often changes the cuts and destroys your previous editing.  The workaround is to add track markers manually after making cuts, or to only use the automatic feature before making cuts.  The automatic track marker recognition is often fooled by noise and will miss points in a noisy recording.  The workaround is to first denoise, declick, and depop the recording, Apply all realtime effects, then use the automatic track recognition.  ACL does not officially support external CD-RW drives, but I have successfully used it to burn on an external drive, but it was not possible to read from the external drive.  Not all drives are supported so you should check if your drive is one that is supported.  The mono function on the STEREO FX adjustment doubles the volume, which could overload the output.  This must be fixed by decreasing the overall volume by 6 db, or by decreasing each channel volume by 3 db.  If you combine a wave file from a previous project into the current project ACL will sometimes (when???) obligingly include the track divisions from the previous file.  Unfortunately it also sometimes adds a track marker at the beginning, underneath the first track maker from the previous project, and this prevents the new project from being burned.  After apply a DirectX plug in, ACL will usually not allow you to normalize a section of the wave file until you exit and restart ACL   *The equalizer bands are too wide with large interactions between them.  The equalization curve shown by the equalizer bears little resemblance to the actual response.  The frequency scale on the spectrum display is offset by one bin.  The first bin marked 40 is actually 28 to 40 Hz.  Finally it is not possible to adjust sound levels at the frequency extremes of the spectrum.  In all fairness many of these problems may not bother the average user.  While ACL does have the capability of stretching or compressing a recording, my experience is that it does not work well.  When a tape recording played back at 3  ips was stretched by a factor of 2 to simulate playback at 1 7/8 ips, the sound quality was terrible.  However ACL could playback the recording at half speed with excellent results.  The problem was fixed by playing back the original at half speed by ACL and recording it at normal speed on another copy.  The results were quite satisfactory.

* This bug was found in 3.0 and is still in 2005.

 

ACL has a number of other editing features that allow simulation of various halls, and encoding of the sound so it can be played back for a home theater using a Dolby compatible logic decoder.  This allows you to pan and expand the stereo image.  In addition it has both single and 3 band compression. These extra features have not been tested.  Finally ACL has a variety of live spectrum displays that can help the user diagnose any possible sound problems.  These displays are a great feature because the user can see exactly the frequency distribution of the sound in real time.  This was used by me to diagnose problems and to create some fairly good equalization curves.

 

Notable new features in 2005/10 are the ability to write a music CD with provision for CD extra music+data files,  presets for RIAA equalization and 45/78 rpm speed conversion, volume curve drawing, Dolby NR simulation, FM stereo bandwidth processing, and batch processing.  In addition ACL now has a wave editor, and a CD booklet/CD label editor, but can no longer write the tracks as a simple text file.  ACL does not actually write the data files for CD extra, but it can create an unclosed music CD so you can add data files in a second session using another program.  The data base support uses FreeDB.org and then gives cryptic error messages when presumably the search usually fails.  The Dolby NR processing is not as useful as it sounds because proper Dolby processing requires careful adjustment using a standard test tape, and it needs to be done first before any other processing.  The RIAA equalization is better than nothing,  but high fidelity sound can not be achieved by connecting a cartridge to a microphone input.  The 45/78 rpm speed conversion does not completely solve the problem of recording 78 RPM recordings on a modern player.  The equalization curve is not uniform and will be incorrect if the recording is not played at the proper speed.  ACL needs a re-equalization curve that matches the 45/78 RPM speed change.  The instruction manual does not give adequate guidance in using these features, so they may not be usable by the average consumer.  Many of these features are actually presets on existing features.  ACL did add some extra indicators such as a dB meter to the FX volume, but it now only has a mono level display next to the master volume, and the contrast is now so poor that it is difficult to see the level display.

 

ACL can not handle Unicode characters as track names.  Only the standard 256 character set is available.  As a result it is impossible to label tracks correctly in some central European languages or Greek  This is will not concern the average user, but folk music and dance enthusiasts may find this to be inconvenient.  In addition it can not import a wave file that has Unicode characters.  This problem is probably generally true of the competing products.

 

ACL can handle 48 kHz audio, but when opened, the file is automatically converted to 44 kHz.  So the saved file is automatically 44 kHz.  All of the processing is done at 44 kHz and if you wish to have the saved file be 48 kHz, you must specify that when saving.  Because of this method of handling 48 kHz, the advantages of the extra bandwidth are thrown away and some extra distortion is added by the double conversion.  They really should do all of the processing on the original 48 kHz file, and setup the wave output to also be at 48 kHz.  Then the final result could be saved automatically at 48 kHz.  You can turn off the 48 to 44 conversion in the FX section, but then the result is a slowed down 44 kHz output file.  This is definitely not the best way of handline the higher bitrate audio.  I could extract 48 kHz audio from a video file, but the option to put the result back into the video file did not work.

 

ACL appears to be an excellent choice for the novice, while providing many features for the more advanced user. but not as many sound editing features as Adobe Audition or Sound Forge.  ACL combined with Sony Media Noise Reduction 2.0 is a powerful combination.  ACL is also the best of the reviewed programs for writing CDs, and with a few simple enhancements and better pop and tick processing it could be the best of the lot for all uses.  MAGIX has improved the support for various drives with each release.  Since most adjustments are done in real time, it is an extremely responsive program.  The responsiveness can allow the user to deal with many more recordings in the same time than would be possible with other programs.  In exchange for easy editing features ACL has some quirks (bugs?).  The documentation is limited and aimed at the novice user.  Unfortunately it does not give enough details, and new features are often neglected or poorly documented. Notably missing features are the ability to specify the CD label, or any extra fields such as artist when burning CD-text. 

Coyote Groove Mechanic 2.5

 

This is an inexpensive shareware program designed just for noise reduction for about $35.  It advertises that music is inserted instead of pops and clicks.  It does not appear to work very well at reducing them on my test recording, but it worked fairly well on the Algorithix test file.  It produces more noise reduction than Audio Cleaning Lab, but far less than the other top products.  This same company also sells plug-ins for compression, expansion, and other purposes.

 

Ganymede Wave corrector 2.5

 

This is a simple program available for about $45.  It has noise reduction and click removal, as well as track identification.  It advertised that it blends the sound to make the correction inaudible, but it missed most of the significant clicks in the old 78 rpm recording.  Overall, it seems to be more effective than Groove Mechanic.  It has the ability to remove noise by comparing the two stereo channels, but this was not tested.  In addition it includes recording capability, track detection, CD writing, and track editing.  One notable feature is the ability to slow down playback to aid in finding ticks and pops.

 

Goldwave 5.06

 

Goldwave is a comprehensive editing program available for about $40.  It has tick and pop removal, noise removal, dynamic range expansion, equalization, time base alteration, and a number of other features.  It lacks peak unlimiting.  The tick and pop removal in V4.0 seems to be very effective, but it left some audible ticks in the music and had a noticeable dulling effect on the highs at the aggressive setting.  It has no facility for delaying one channel relative to the other, so it may not be suitable for tape restoration.  It lacks CD writing features and track layout features.  It does not have the flexibility of Audio Cleaning Lab to mix channels to mono with adjustable gain on each channel.  Goldwave can use DirectX plug-ins and can save plug-in settings so it is possible to expand its capabilities.  It includes a 7 channel equalizer that may not be adequate for re-equalization, but it has a parametric equalizer with a large number of bands.  The parametric equalizer may be much too difficult for the average user.  The noise reduction can use a noise print, but the controls do not seem to be very sophisticated.  There does not seem to be any facility for performing actions in real time so A/B comparisons may be difficult.  Goldwave includes both compression and expansion.  Goldwave has no ability to edit right and left channels individually.  Goldwave does not have the user friendly design of Audio Cleaning Lab.  For the user on a very tight budget Goldwave offers good capability.

 

Adobe Audition (AA)

 

Adobe Audition was formerly Syntrillium Cool Edit Pro.  The lower priced consumer versions are no longer available.  As such it is now firmly aimed at the semipro market, and is marketed at a higher price.  Only the old Syntrillium product was tested, but the new version is probably the same with a few small enhancements.  AA is available for about $299, but an educational discount may be available for qualified individuals.  For this price you get a very capable editing program with significant noise reduction.  The noise reduction is performed separately, and can not be done in real time so A/B comparisons are difficult.  AA is capable of delaying one stereo track with respect to another and of editing each track individually.  As a result it is possible to correct tapes when there is a delay between tracks, and pops in one channel can be easily erased without affecting the other channel.  AA includes two means of equalizing the sound.  Unfortunately the included 30 db multiband equalizer only goes down to 80 Hz, but it can save the curves.  Three very capable 18 db equalizers are also available.  AA can perform both dynamic range compression and expansion.  AA can do timed recording, but does not have an automatic shutoff.

 

The noise processor seems to be fairly capable.  It was much better than the inexpensive shareware but not quite as good as Sony Media Noise Reduction.  The click and pop removal has a large number of adjustments which could be intimidating.  The AA plug-ins are not DirectX compatible and could not be used in other programs.  AA writes CDs and it can convert wave files to MP3.  A free plug-in is available to allow AA to use available DirectX plug-ins.  AA has a number of other features which allow you to tailor the sound.  AA also has a pan/ stereo expander.

 

AA may take a long time to process the sound.  For example, it took over an hour to process a small delay in one channel with a 1.2 hour file.  By contrast the Virtos stereo processor needed just a few minutes to perform the same process.  AA is probably doing elaborate curve fitting, while Virtos is just shifting the data by integral channels.

 

Sony Media Sound Forge 9.0 (SF)

 

This is an audio editing program which is similar to Audition, and was formerly designed by Sound Forge.  The new Sony 9.0 version has not been fully tested, but the literature leads one to believe that it is the original Sound Forge with only a name change.  9.0 retails for about $300 in the full version with Sound Reduction, but with an academic discount it runs $100-200.  It includes delay, sound base changes, and CD writing.  It is one of the few programs that can handle full Unicode. The built in equalizer has up to 24 db correction in 3, 10, or 20 bands.  The equalization can be saved.  The bands must be adjusted via a slider and the values can not be entered as a number, but the slider adjustment is precise to 0.1 db.  It does have the ability to redraw a wave shape.  It has dynamic compression and expansion capability.  SF has either timed recording or automatic stop. While it has a spectral analyzer which can view level vs frequency and time, but unlike ACL it can not edit this display.

 

A notable feature in the full version is the plug-in chainer which allows multiple passes on plug-ins.  This might be a time saver when processing a large number of recordings.  SF includes a pan/expander and compression.

 

SF can burn CDs and it has the capability of dividing the wave file into regions.  This is similar to Audio Cleaning Lab.  However, each track must be saved to disk and burned separately.  Then before the CD is usable it must be closed.  This is fairly primitive compared to Audio Cleaning Lab, and will probably introduce a 2 second delay between each track.

 

SF is extremely capable, but the large number of controls might be very intimidating for the novice user.  With the Noise Reduction 2.0 plug-in it is superior to most other programs.  SF Studio includes many of the features in Audition.  Sony Media seems to be head to head competitor with Adobe for the same market.

 

Sony Media Sound Forge Studio 8.0 (SFS)

 

 

This is a cut down version of the very professional Sound Forge, with a limited selection of the full Sound Forge plug-ins.  It costs about $70, but occasionally it can be found at a low price or even free after a rebate. Since it can use plug-ins, it can serve as a cheap master program.  It can handle full Unicode. It can output files in 19 formats including Ogg Vorbis, MP3, MP2, Wave,   It can both extract tracks from CDs and burn CDs.   It does not handle CD-text.  It has simple tick or pop reduction, and noise reduction but no noise print.  For these features you must acquire Sony Noise Reduction.  It does have a method for adjusting the time delay between right and left channels.  It has a good 10 band equalizer, and the ability to phase reverse a channel.  It has compression, but no expansion.  It also boasts a simple 10 band equalizer which might be capable of re-equalization.  In addition it can normalize files either by average level, or by peak level.  This is a fairly capable program, but the cost is much higher if you need good noise, and tick/pop reduction.

 

Virtos Noise Wizard

 

Virtos Audio Restoration Technology offers their Noise Wizard DirectX plug-in for about $120.  This is the same technology included in Cakewalk Pyro.  It is a suite of applications that include a declicker, denoiser, filter toolbox, stereo processor and band extrapolation.  The declicker includes tick reduction and crackle reduction with two adjustments for each.  It could reduce some of the pops in my 78 rpm test, but could not eliminate them.  When used aggressively it turned my test file into mush.  It did a much better job on the other test files, but left in some residual soft pops.  The denoiser can use a noise print.  The band extrapolation is designed to synthesize lost frequencies.  The stereo processor is designed to modify the stereo image, but it also includes a control which allows setting of time delays in 0.01 ms increments.  This is perfect for correcting time delays for tapes that have skewed.  To produce a time delay, turn the stereo enhancement to zero and then just dial the delay within a 0.18 ms range.  The algorithm only shifts the existing data over and it is possible to undo the delay by shifting it back.  If both the delay and enhancement are set to zero, then the input sliders can be used to reduce alter the level or even reverse the phase of the processed section.  This can be useful for programs such as Audio Cleaning Lab which lack the facility for editing data.  Each plug-in may be purchased separately.  The plug-ins apparently can not be used in real time so A/B comparisons are difficult and adjustments may take longer.

 

Arboretum Systems Ray Gun 2.0

 

This is a DirectX plug-in for about $100 which is designed to remove ticks, pops, and other noise.  It ran under Audio Cleaning Lab, or as a standalone.  It was very easy to use, but it had to be set to 90% to reduce the pops in the 78 rpm test file to almost inaudibility.  100% produced some distorted sound.  It also produced excellent results on all of the other tested wave files.  Ray Gun includes a rumble filter, notch filters for 50 and 60 Hz hum, and a noise removal section.  It is not possible to use a noise print, so it is not as flexible as other tools and the pop and tick removal has just one control, so it may not be possible to adjust it for optimum results.  It does not include peak unlimiting.  Ray Gun can be used in real time, so A/B comparisons are easy to make.  Ray Gun has one very annoying quirk.  When you adjust a level, it doesnt automatically turn on the corresponding feature.  The feature has to be separately enabled, and as a result it is possible that the user might think that the feature doesnt work.

 

Sony Media Noise Reduction 2.0

 

This is not a full program, and requires a program that accepts DirectX plug-ins such as Audio Cleaning Lab,  Sound Forge, Vegas, Cakewalk Pro Audio, Pinnacle WaveLab, IQS SawPRo, Adobe Audition,  or MS Plus for XP.  It includes pop and crackle removal, peak unlimiting, and noise removal.  This is a very capable program, and has a large number of adjustments.  Overall it produced excellent results on all of the tested wave files using default settings, but left in some soft residual nose.  It has a good set of defaults for various types of recordings, but the large number of adjustments may confuse novices.  It sells for $280 but with an Academic discount it is competitive in price with the Adobe equivalent plug-in.  It processes a long audio file fairly quickly.  The noise reduction is performed on the file, and can not be done in real time so A/B comparisons are difficult. I have used this program extensively, and it can remove all of the extremely intrusive pops and ticks.  Generally a few are left behind after using the default settings, but they are very low level or are low frequency pops.  About of the residual noises can be removed by hand using stronger settings on just that location.  Some of the remaining ones can be muted, but the low frequency pops are usually impossible to remove except by cutting them out.  Sometimes they can be removed by a low cut filter, but this strategy is only available when there are no bass instruments playing during the pop.  Overall the remaining noises are very unobtrusive.  The large number of adjustments allows fairly intelligent pop removal when other products fail.

 

Diamond Cut DC 3.5

 

This is a complete audio editor with noise reduction facilities for about $200.  It is the only one that advertises that its equalizer can be set for re-equalization; but it does not provide any settings for doing this in the demo version.  Unfortunately, when I used it on a sample file with heavy ticks and crackle the default settings, it either achieved little audible noise reduction or produced garbage.  It did a much better job on another test file.  I have included the results from the EZ pop and tick removal set for aggressive removal.   DC includes a 12 db 10 band graphic equalizer which can save settings.  In addition, it is possible to edit individual channels, but it does not seem to be possible to delay one channel relative to the other.  There is no provision for CD burning.  It does have the ability to redraw a wave shape.    There is no peak unlimiter.  DC lacks any dynamics compression or expansion.

 

SEKD Declicker

 

This is a DirectX plug-in for $200.  The declicker has only a strength adjustment and it does not do a very good job on the 78 rpm sample file, even when set to the maximum, but it did much better on other test files.  It also has a decrackler.   SEKD also produces a descratcher plug-in for about $1000 which should work better.  This module worked with Magix as a real time plug-in.

 

Algorithmix Sound Laundry 2.5

 

This program is a noise reduction suite bundled into one interface for about $230 or $300 in the advanced version.  It consists of a series of plug-ins which are not DirectX compatible.  It has a descratcher, DC-removal, denoiser, derumble, high and low cut, and an equalizer.  The descratcher muted the ticks in my test file, but did not remove them as well as Adobe Audition, Noise Reduction, or Ray Gun.  Sound Laundry did a much better job on other test files.  The demo version can not produce output and the display does not show the cleaned sound so there is no sample in this review.  The shellac default did nothing on my test file, but the level had to be advanced to 100% and the shape to 75 to achieve some muting but no removal of ticks. The denoiser can use a noise print, so it should be capable of good noise removal.  There is no provision for manual editing, a delay, or peak unlimiter.  The equalizer has only 6 bands and may not be suitable for re-equalization.  The suite works in real time so A/B comparisons are easy to make.  Manually removing single ticks is not possible.  The wave form display is extremely small, which is adequate since there are no manual editing features.  Many of the features were not documented in the help file.   Since Algorithmix has a corporate tie with Magix both of their products often produce similar results.

 

Dart Pro 32

 

This is a complete audio editor with noise reduction facilities for about $260.  This program has pop and tick detection which appears to work much better than most other programs that I tried.  However it does not remove crackle.  It is possible to set markers on ticks so that they can be removed.  It includes a 15 db 10 band graphic equalizer which can save settings.  It has a peak unlimiter.  It does not seem possible to delay one channel relative to the other, nor can it change the time base.    It also lacks CD burning facilities.  It has dynamic compression, but no expansion.

 

It is possible to make A/B comparisons of the original and the cleaned audio.  This is done by displaying both the sound and the processed sound.  This neat feature allows the user to do A/B comparisons even when the processing is not done in real time.  This is a feature that should be available in all audio editors.

 

Programs and plug-ins that lack most needed features

 

Sound Edit Pro 1.1 is a basic sound editor lacking effective noise reduction or delay features.  BlueLine plug-in suite has delay, but only in 0.1 ms increments which is not adequate for correcting tape delay; but their filter might be useful.  The Blueline parametric equalizer does not have enough bands for effective equalization.  DB has a dynamic range compressor/expander and also a plug-in to reduce sibilance, as well as other effects.  Stomp My Sound Studio for about $70 is a decent sound editor,  but it lacks good noise reduction.  It does have time base correction and a good equalizer with +20-60 db range but no ability to save equalization curves.  A good listing of available DirectX plug-ins can be found at http://www.thedirectxfiles.com/plugins.htm .

 

Programs of Interest, not Reviewed

 

Pinnacle Clean 4.0   From the manufacturers information and a review there is noise reduction, and CD writing, and channel delay.  There is no mention of automatic track splitting.  It has pop and tick suppression as well as noise reduction using a noise print.  It has 3 band compression similar to Magix but no expansion.  It includes time base modification and brilliance enhancement.  Overall it looks similar to Magix Audio Cleaning Lab, but with possibly fewer features.  The quality of the noise reduction could not be assessed because they do not offer a trial version.  I have been told by someone who owns both products that Pinnacle does a much poorer job of pop and tick suppression compared to Magix, but I have not personally verified this.  Magix and Pinnacle seem to be head to head competitors.  Pinnacle Clean was formerly developed by Steinberg.

 

Cakewalk Pyro 2003  Pyro costs about $25 as a download.  From the reviews and the manufacturer supplied comments; it has tick, crackle, and noise reduction using a noise print.  There is no compression/expansion or channel delay.  It has a 6 band parametric equalizer which may not be adequate for re-equalization.  CD writing and automatic track splitting are included.  It probably does not have time base modification.  The true quality of the noise reduction could not be assessed, as only the older version 1.5 is available as a demo.  Since Virtos declicker and denoiser are now incorporated into Pyro, the Virtos review should also apply to Pyro.  The older version has delay, but it could not be adjusted to correct for tape problems.

 

More testing on pops and ticks

 

The Sound Laundry demo came with a demonstration wave file which had many severe ticks, pops and crackle.   Indeed they were so severe that the music was very nearly inaudible.  I used this file as a test for several of the tick and pop reducing programs.  Not surprisingly the best reduction was from Sound Laundry which removed the noise without leaving audible artifacts.  Excellent reduction was produced by Sony Media Noise reduction using its default for 78 rpm recordings.  Virtos declicker was almost as good. Goldwave also removed all ticks using its aggressive setting, but the aggressive setting may have undesirable effects on highs and transients.  Ray Gun was almost as good, but left in a couple of loud ticks and may not be as good at removing crackle.  They all left some soft pops instead of the clicks.  Surprisingly Diamond Cut produced good results but left in a few clicks.  SEKD removed many of the clicks, but left some soft squeaks in the result.  Dart Pro was similar to SEKD.  Acoustica removed all of the large clicks, but left some small pops and the cleaned track sounded unstable with a bubbling sound in the background.  Magix produced slightly worse results even at the maximum setting.  Adobe did achieve good removal, but left in distinctly more ticks, pops, and crackles.  Groove Mechanic and Wave Corrector did not remove as many clicks as Adobe and not as much crackle.  It is quite evident that various manufacturers have optimized their products for particular types of files.  Sony Media was one of the best, which should not be surprising in view of the price, but Ray Gun at 1/3 the price made an excellent showing.  The following picture shows the original file, then how it is processed by Magix, SEKD, and finally by Sony Media.

 

A test on a simple file supplied by Groove Mechanic with 3 large pops and a few small ticks again showed that Goldwave, Ray Gun, Sony Media, and Virtos were superior, with SEKD matching them, while Sound Laundry left in a large pop.  Finally on a test file supplied by Dart Pro, I had similar results, with Goldwave, Sound Laundry, SEKD, Ray Gun, Virtos and Sony Media all doing an excellent job.  Sony Media has more adjustments, is more adaptable to different audio files, and eliminated all pops and ticks, but left in some artifacts.  Sony Media also has very adjustable noise reduction and a peak unlimiter.  Goldwave was also excellent and was very easy to use.  Ray Gun is also much easier to use, and it can process files in real time at a lower cost, but failed a couple of times.  Virtos, SEKD and Sound Laundry are also very good at an intermediate price, but could not handle my test recording.  Both Virtos and Dart Pro seemed to be more likely to mangle the sound.  No single pop and tick reduction program produced absolutely the best results on all of the samples.  I have not attempted to assess how badly various programs handle percussion instruments, so careful listening is necessary whenever tick and pop suppression is used.

 

For best tick and pop suppression you might consider owning more than one program, to be able to take advantage of slightly different implementations.  A very good example of this is a file that I worked on.  Sony Media Noise Reduction successfully found most ticks, pops and crackles, and removed them.  However, several series of repeating crackles still remained.  Noise Reduction was unable to remove some of them, or to even substantially reduce them.  Magix Audio Cleaning Lab was able to completely remove most of these resistant crackles, despite the fact that it is not as good a performer overall.  Audio Cleaning Lab is apparently optimized to remove ticks and pops on LP albums.  I have an album which had several very large ticks which were inaudible after ACL, but were only converted to soft pops by Noise Reduction.  Noise reduction seems to work better on ticks and pops when recordings have been transferred to other media.  For example, Noise Reductions works better on records transferred to tape and then digitized.

 

Software Summary

 

The choice of program depends on what you wish to accomplish.  The Magix Audio Cleaning Lab is the clear winner for ease of use and CD burning capabilities.  It is the only product clearly designed to be easily usable by novices for recording, cleaning, enhancing, and storing old recordings.  It has good to excellent click and pop reduction, but this can be supplemented by a plug-in. The spectral cleaner an indispensable asset.  It is also the clearly a best buy ($40) and includes a large number of features, but lacks some advanced editing capability.  Pinnacle Clean may also be a best buy, but I could not get an evaluation version, so it is unrated.  Goldwave is also a best buy with good pop and tick reduction, but is not as easy to use.  Sound Forge is clearly more capable than all other programs, but the cost including the Sony Media Noise Reduction plug-in is fairly high ($300).  For a full feature editor, Adobe Audition is a slightly better bargain, but I could not obtain excellent tick and pop reduction with it.  Dart Pro 32 may be the best solution if pop and tick reduction is paramount, but it lacks some features.  Diamond Cut DC 35 is comparable to Dart Pro, but the tick and pop reduction seems to be difficult to use.   From my experience inexpensive shareware products are not very useful for heavily scratched 78 rpm recordings.  However both of the shareware products allow you to mark ticks for removal, and may be suitable for LP recordings.  Some combinations might be a good buy such as Audio Cleaning Lab plus Ray Gun for $110 if you need very good pop and tick reduction with ease of use.   For the ultimate in noise reduction at the lowest price Audio Cleaning Lab plus Sound Forge Noise Reduction is only $320, but it costs much less if you qualify for an academic discount.  The well heeled expert would obviously go for the full Sound Forge suite.  The expert on a budget might buy Adobe Audition, but the average user might find Magix to be the perfect choice.  Anyone who needs to do the job quickly might also find Audio Cleaning Lab a good buy.  Remember that you should always test a program before you buy it.

 

The perfect software would include the ease of use and features of Audio Cleaning Lab with the editing ability of Sound Forge combined with excellent pop and tick suppression.  The ideal program should make more than one version of the sound available to facilitate easy audio and visual A/B comparisons, similar to Dart Pro.  Such a package should allow the user to mark a pop to be automatically removed. Exact time delay of either left or right channel generally takes a long time to process, but a simpler scheme could round off the delays to the nearest time bin.  Then it could be accomplished instantly by just keeping track of the delay for output and the display could simply be shifted.  A simple algorithm could be included to compare 2 mono tracks or even two stereo tracks and automatically align them.  Equalizers should have at least 18 db range in 0.1 db steps over at least 10 bands with the ability to specify the value rather than just using a slider.  Standard re-equalization curves could be available for older recordings at no added cost.  Finally left/right comparison software could be included to significantly reduce noise in monophonic recordings.  A separate left right comparison could also be used to automatically adjust the balance of either mono or stereo recordings.  For mono recordings this would enable an exact cancellation of difference noise which can be 3dB to a larger figure for old LPs and 78s.  Many of these features could be added to existing programs at minimal or no cost.

 

Appendix A Sample tick and crackle reduction

 

Click on the link to hear the results.  The original was a 78 rpm dance recording which had been put onto open reel tape before being digitized.  This combination seems to pose more severe difficulties than other sample files.  The samples are approximately in order of quality with the highest pop reduction last.

1.      Original recording

2.      Magix pop and crackle reduction at maximum

3.      Wave Corrector at maximum

4.      SEKD declicker

5.      Groove Mechanic at maximum

 

6.      Diamond cut No sample audio file available EZ removal - agressive

7.      Virtos Noise Wizard (threshold = 20%)

8.      Adobe (Hiss + lots of clicks) Auto find levels

9.      Goldwave aggressive preset

10.  Ray Gun set to 90%

11.  Sony Media Noise Reduction 78 rpm default

12.  DartPro No sample audio file available  - Heavy scratch light crackle removal.

 

Appendix B Equalization Curves

 

These curves were taken from two sources on the web.  First a set of equalization curves were given at http://www.vadlyd.dk/English/RIAA_and_78_RPM_preamp.html for a commercially available preamp.  The second set was given at http://shellac.org/wams/wequal.html which were taken from an issue of High Fidelity published in the early 1950s.  Both of these articles can be used to help you restore old recordings.

 

The table below is a set of curves for re-equalizing recordings.  The RIAA column shows the actual RIAA de-emphasis curve.  The other columns show the curves necessary to get good reproduction with the specified recordings.  The top column shows the numbers used to calculate the curves.  The Lo Bass is the 3 db point at which the bass is no longer rolled off.  The Bass frequency is the 3 db point at which the bass is rolled off.  And the High frequency is the 3 db point for the high frequency pre-emphasis.  The curves have been normalized to either be almost zero dB at 1000 Hz, or zero over an extended range.  Acoustical recordings would generally need the inverse of the RIAA curve.  Bear in mind that acoustical recordings have severe resonances which can be reduced by a good equalizer.  In all cases record producers often deviated from the published record label specifications, so try alternatives until you find that one that sounds the most natural.

 

78 rpm Equalization curves

 

 

Lo Bass

50.05

1

50

50

40

50.5

1

50

50.5

50

50

50

Bass

500.5

200

250

500

250

500

150

353

500

400

300

300

High

2122

50000

50000

50000

6360

5000

3400

3180

2600

2600

2050

1600

Norm

1

 

1

 

 

 

 

 

 

 

-0.3

-1.16

Labels

RIAA#

Westrex#

Blumlein

Early 78s

ffrr 78 '49

Victor38-47

Decca 78

BSI 78

Victor 47-52

AES

ffrr '51

Columbia 38

Frequency

Hz

|

V

 

HMV (tri) Col (W)

EMI 31  HMV 31  Col (Engl)

Brunswick Parlophone Kismet

Decca London

 

 

All 78s after 1978--

 

Decca 34 Mercury Capitol 42

 

 

30

18.68

-2.11

-6.98

0.00

-4.64

-0.07

-4.54

-3.01

-0.07

-1.93

-4.11

-3.25

50

17.03

-4.73

-6.90

0.00

-5.03

-0.05

-7.04

-2.98

-0.05

-1.92

-4.07

-3.21

60

16.19

-5.36

-6.84

0.00

-5.15

-0.04

-7.59

-2.97

-0.04

-1.91

-4.03

-3.18

87.5

14.04

-6.10

-6.65

0.00

-5.25

-0.02

-8.09

-2.90

-0.03

-1.87

-3.92

-3.07

125

11.65

-6.14

-6.31

0.01

-5.09

-0.01

-7.78

-2.77

-0.01

-1.80

-3.71

-2.86

175

9.26

-5.63

-5.77

0.02

-4.65

0.01

-6.88

-2.56

0.00

-1.68

-3.38

-2.54

250

6.77

-4.62

-4.93

0.05

-3.87

0.04

-5.45

-2.20

0.01

-1.46

-2.83

-2.01

350

4.63

-3.40

-3.93

0.11

-2.91

0.09

-3.94

-1.72

0.03

-1.17

-2.15

-1.37

500

2.74

-2.09

-2.81

0.23

-1.82

0.19

-2.46

-1.13

0.07

-0.79

-1.40

-0.69

700

1.32

-0.98

-1.82

0.44

-0.87

0.36

-1.31

-0.57

0.14

-0.42

-0.79

-0.21

1000

0.09

0.08

-0.84

0.87

0.06

0.70

-0.35

0.00

0.27

-0.05

-0.35

0.00

1400

-1.05

1.14

0.18

1.56

0.98

1.24

0.42

0.55

0.46

0.28

-0.12

-0.07

2000

-2.50

2.54

1.56

2.75

2.16

2.12

1.23

1.18

0.74

0.65

-0.01

-0.33

2800

-4.24

4.25

3.26

4.37

3.51

3.19

2.01

1.82

1.03

0.99

0.02

-0.64

4000

-6.52

6.50

5.51

6.56

5.09

4.43

2.75

2.43

1.31

1.29

0.02

-0.90

5600

-8.98

8.93

7.93

8.96

6.49

5.48

3.28

2.87

1.50

1.49

0.01

-1.07

8000

-11.81

11.70

10.70

11.71

7.69

6.31

3.65

3.16

1.62

1.62

0.01

-1.18

10000

-13.65

13.48

12.48

13.49

8.24

6.67

3.80

3.28

1.67

1.67

0.01

-1.22

12000

-15.17

14.93

13.93

14.94

8.59

6.88

3.89

3.35

1.70

1.70

0.00

-1.24

16000

-17.62

17.20

16.20

17.20

8.97

7.12

3.97

3.42

1.73

1.73

0.00

-1.26

 

The following graph shows the curves that were calculated in the table.

 

LP Equalization curves

 

All LP recordings made before 1955 used proprietary equalizations.  This would be true for virtually all monophonic LPs which are not modern reissues.  Information available at http://www.tdl-tech.com/restore.htm was used to generate the following table. The Lo Bass point was assumed to be 50 Hz. The asterisk indicates more than one curve is possible.

 

Lo Bass

50.05

50

50

50

50

50

50

50

50

50

Bass

500.5

400

450

500

375

400

500

629

300

500

High

2122

2700

2700

2700

2700

1600

1600

1600

2122

5500

Norm

1

 

 

 

 

 

 

 

 

 

Labels

RIAA#

Capitol

Decca -53

Angel

EMS

alternate

Decca-55

Bartok

Decca -51

Oiseau-Lyre

Frequency

Hz

|

V

 

Capitol-Cetra  Cetra-Soria* Colosseum* ConcertHall* Esoteric  Mercury->54  Westminster*

Ducretet-Thomson London->LL-846              London Intl.

 MGM      RCA->8/52

 

Lyrachord*

AudioFidelity Bach Guild 501-529 Columbia

Cetra-Soria*  Colosseum* ConcertHall* Epic ->54    Folkways   Haydn Soc HMV         Lyrichord* Oceanic Overtone Polymusic Remington Urania Vanguard Westminster* VOX

Bartok 301-304, 309, 906-920           Caedmon 1001-1022

 

ConcertHall* -1954

30

18.68

-1.93

-0.91

0.00

-2.49

-1.93

0.00

1.98

-4.41

0.00

50

17.03

-1.92

-0.91

0.00

-2.47

-1.92

-0.01

1.97

-4.37

0.00

60

16.19

-1.91

-0.90

0.00

-2.45

-1.91

-0.01

1.96

-4.33

0.00

87.5

14.04

-1.87

-0.89

0.00

-2.40

-1.88

-0.01

1.93

-4.22

0.00

125

11.65

-1.80

-0.86

0.00

-2.31

-1.82

-0.02

1.88

-4.01

0.01

175

9.26

-1.68

-0.80

0.00

-2.14

-1.71

-0.03

1.79

-3.67

0.02

250

6.77

-1.46

-0.70

0.02

-1.85

-1.53

-0.05

1.61

-3.12

0.04

350

4.63

-1.16

-0.55

0.04

-1.47

-1.29

-0.09

1.34

-2.44

0.09

500

2.74

-0.78

-0.35

0.08

-0.99

-1.04

-0.17

0.94

-1.68

0.19

700

1.32

-0.40

-0.12

0.16

-0.53

-0.88

-0.31

0.47

-1.06

0.38

1000

0.09

-0.01

0.14

0.31

-0.09

-0.89

-0.56

-0.08

-0.60

0.73

1400

-1.05

0.35

0.44

0.53

0.31

-1.08

-0.90

-0.62

-0.33

1.30

2000

-2.50

0.77

0.81

0.86

0.75

-1.42

-1.33

-1.18

-0.17

2.22

2800

-4.24

1.16

1.18

1.21

1.15

-1.76

-1.71

-1.63

-0.09

3.38

4000

-6.52

1.51

1.53

1.54

1.51

-2.04

-2.02

-1.98

-0.04

4.74

5600

-8.98

1.75

1.76

1.77

1.75

-2.22

-2.21

-2.19

-0.02

5.92

8000

-11.81

1.91

1.92

1.92

1.91

-2.33

-2.33

-2.32

-0.01

6.89

10000

-13.65

1.97

1.98

1.98

1.97

-2.37

-2.37

-2.36

-0.01

7.32

12000

-15.17

2.01

2.01

2.01

2.01

-2.40

-2.40

-2.39

0.00

7.58

16000

-17.62

2.04

2.05

2.05

2.04

-2.42

-2.42

-2.42

0.00

7.86

 

The MS Excel files, and equalization curves for Magix Audio Cleaning Lab 3.0 are available in equalizations.zip.  The accuracy of the Magix files is not guaranteed because of the problems with the equalizer.  Most of them are within .5 dB of the correct curve except below 75 Hz.  The .mfx files must be put into the fx-preset folder.

 

Appendix C Table of features

* review copy is not available for testing.

 

 

Audio Cleaning Lab*

Sony Sound Forge Studio

Sony Sound Forge

Noise Reduction

Audition

Pinnacle Clean*

Goldwave

Nose Wizard

Diamond Cut

Sound Laundry

Dart Pro

Pyro*

Acoustica

Wave Corrector

Record

 

 

Auto Stop

?

 

 

 

 

 

 

 

 

 

Auto region

 

 

 

 

 

 

 

 

 

Remove DC

 

 

 

 

 

 

 

 

De-Pop

 

De-crackle

 

 

De-noise

 

 

De-Rumble

 

 

 

 

 

 

 

De-clip

 

 

 

 

 

 

 

 

Spectral edit

 

 

 

 

 

 

 

 

 

 

 

 

 

Time-base

 

 

 

 

 

 

 

Filters

 

 

 

 

 

Time delay

 

 

 

 

 

?

 

 

Stereo control

 

 

 

 

 

 

 

 

Compress

 

 

 

 

 

Expand

limited

 

 

 

 

 

 

 

 

 

Equalize

bands, range

10, 12

inaccurate

10, 24

20, 24

 

24, 18

8, 12

7, 12

 

10, 12

6, 12

10, 15

6

 

 

CD burn

 

 

 

 

 

 

CD rip

 

 

 

 

 

 

 

 

 

CD-text

 

 

 

 

 

 

 

 

 

 

 

 

 

MP3

 

 

 

 

 

 

 

 

 

Use Plug-ins

DirectX, VST

DirectX

DirectX

 

 

VST

DirectX

 

 

 

 

 

DirectX

 

Has plug-ins

DirectX

DirectX

DirectX

DirectX

 

 

 

DirectX

 

DirectX

 

DirectX

 

 

$

10-40

0-70

100-300

70-280

150-300

40

50

120

230

200

260

30

30

45