| Larry's profileLarry Doe's Scanner BlogPhotosBlog |  Tools  Help |
|
March 29 Never leave a glass of wine on nightstand next to laptop.Since our last episode, a rare unissued Levitzki Ampico notesheet has been scanned, perforator components were tested, more perforator components were ordered, leather for Chickering restoration arrived and mass extermination of poison ivy was carried out.
On May 24, 1926, Mischa Levitzki returned to the Ampico recording piano and recorded the 2nd. movement of Beethoven's Waldstein sonata Op. 53. The notesheet seen in the accompanying photos is the actual original master made on the line marker in the Ampico recording studio nearly 83 years ago. You can clearly see every perforation is hand-punched. There are 3 punches at the start of every note and a single punch at the end. A pencil mark connects these perforations on sustained notes. The numbers written next to the start of every note are probably the intensity values and might be proof the spark chronograph was in use at this early date. Some lighter vertical pencil marks show the locations of the expression tracks and the stack division. There are some horizontal lines from beginning to end with numbers written in on the treble side which might correspond to measure or beat numbers. The very first chord is numbered '1'. There are areas where evenly spaced horizontal lines are drawn in and areas of notes are encircled with a comment to even them up. There are many thousands of small yellow and white bits of paper pasted over note holes. Many have been punched through. This lends credence to the stories of many of the classical performances needing a month or more of editing. There may have been a second go-round of editing which could explain why two colors of editing tape were used. The note sheet would have to be run on the (now nonexistent) stencil machine in order to create a playable roll. It's interesting to note, the editors had little control over which exact punch row an event would fall into. The automated stencil machine had to make that decision. Perhaps the many bits of paper tape were an attempt by the editors (Miss Volavy in this case) to move an event to a different punch row. I don't know of any surviving trial rolls of this performance and we will have to manually insert all of the chaining into the Scanimage MIDI file. Won't it be interesting to hear how polished this performance was? The photos show entire sections of paper having been spliced in. Could this be be some of Miss Volavys 'assisting'?
By the time I arrive at Keystone April 4th. to install the stepper advance unit, I will know just how effective my protective clothing was for preventing contact with the many poison ivy vines which were cut down yesterday. My face was covered with Ivy Block lotion and a face shield. There was no other exposed skin and all gloves and protective clothing were put into the trash. No tools were handled after the gloves came off. I devised this experimental way of killing the roots by wrapping the ends of the vines with a paper towel and dipping it into herbicide. The ends were then encapsulated with urethane foam. This will contain the poison and prevent it from being washed away by rain. It also keeps the poison from killing the host Arbor Vitae bushes. Time will tell whether this method is effective.
The Z-tronics organ rank driver board set has been bench tested successfully and a second set of stepper drive components has been ordered for the late Aeolian perforator.
One night while in dreamland, I thrashed around enough to knock half a glass of wine off my nightstand and onto the keyboard of the new Toshiba laptop computer. The wine poured through the computer and out the CD ROM slot into a pool on the floor. A bath towel came to the immediate rescue but did little to improve the condition of the computer. I avoided powering it up and proceeded to disassemble the entire laptop the next morning. Luckily, only a few drops made it into the hard drive and CD ROM and did not touch the motherboard at all! The worst case scenario was avoided and only the keyboard would need replacing. I decided to try my luck disassembling the keyboard. There were enough eyelashes in there to reconstruct Marilyn Monroe! The trick will be in discovering how to unsnap the keytops from the tiny scissor hinges without breaking anything. There are a few decorative divider strips which must be broken off before the sheet of flexible rubber buttons can be lifted off the circuit board membrane. There are no active ICs in the keyboard assembly except a few LEDs and all switch contacts are hermetically sealed within the membrane. After washing and drying all parts, the keyboard was carefully reassembled and the laptop has been working reliably for more than a week. March 08 Ampico stepper drive system finishedFor the past three hours, I have been exercising the newly completed stepper drive system for Keystone's original Ampico Taft Pierce perforating machines. The last week was spent designing, fabricating, assembling, wiring and programming the new device. See accompanying photos. It will now be possible to perforate piano rolls with virtually any punch row spacing. The immediate goal is to be able to punch Ampico rolls made during the 'merger' period when the rolls were manufactured on Aeolian perforators with .030" punch row spacing. Instead of physically changing gear ratios, it will now be possible to select the desired punch row spacing with a simple thumbwheel switch. The number of microsteps per punch row has been calculated using an Excel spreadsheet designed by the author. This spreadsheet contains equations which model the physical characteristics of the perforator mechanism along with the gear reduction ratio of the drive motor. All cells enclosed in heavy borders are the input parameters. The most important calculated result is the number of microsteps per punch row which is programmed into the motion controller and selected by the thumbwheel switch. The rest is a matter of calculating the proper acceleration and deceleration rates which will produce smooth indexing movements without any undesirable harmonics.
One crescendo pneumatic has been restored and a few progress photos are included. I plan to make another visit to Keystone in a few weeks to install the new motion controller and will check on the progress of the Chickering piano restoration at the John Zeiner & Sons shop during that trip.
February 15 The great summitThe Keystone crew was in Frederick, Maryland Feb. 6th to attend a screening of the lost (and now found) silent film "Beyond The Rocks" starring Rudolph Valentino and Gloria Swanson. Accompanying on the mighty Wurlitzer was Ray Brubacher. We arrived early for Richard Groman to assist Doug Miller in tuning the pipes numbering over 600. We were fortunate to be joined by Spencer Chase, Warren Trachtman and Joe and Lauren Orens. There was time for us all to have dinner at the local brew pub before showtime.
Warrren and Spencer followed us back to Bethlehem, PA after the show and were guests of Richard Groman for most of the weekend during which many of the nuances of roll mastering, editing and technological advancement were discussed. Warren gave us a demo of his new USB scanning software and was present while several 88 note rolls were edited, perforated and played on the Lauter foot pumper. Spencer showed us many of the utilities he has written relating to scanning and rendering. He is now testing a new perforator control program which could allow a PC to replace the older MIDI sequencer which depends on (sometimes unreliable) floppy disk technology. There was considerable discussion on paper characteristics, lyric printing and web hosting. On the latter subject, I would expect to see some progress soon since some new people are becoming involved. The response to the latest printed catalog has been excellent and we're hearing very positive feedback about the product quality.
The next technological advancement will be the addition of a modern programmable motion controller and stepper motor with gearhead to the now computerized Ampico perforator. Classical enthusiasts need not worry. The number 2 perforator is untouched and still operates from the traditional paper stencil masters and is making beautiful rolls with its brand new die set. Today, I finished the bench testing of the motion controller. It is able to read a BCD thumbwheel switch and execute any of 10 different pre-programmed step rates. We have effectively surmounted the limitations of the available change gears and should now be able to perforate many more piano roll formats including the Aeolian/Ampico titles issued during the merger period. With microstepping, the overall length of a remastered roll can be matched to the original roll with a precision of one sixteenth inch. The holy grail of hole-to-hole roll copying is now within our grasp!
On the home front, the Chickering Ampico restoration is moving forward. The piano was picked up today by Modern Piano Moving. The 40 foot trailer was filled with 20 pianos and one Ford Ranger which will be employed on a challenging delivery in Maine. Russ, Joe, Ken and (dog) Maggie will be making stops in Rhode Island, Massachussetts and Maine before heading back to NY and Allentown, PA where my Chickering will be delivered to John Zeiner and Sons for full restoration and refinishing. That will take at least until early Summer to complete. So far, I've rehinged, covered and sealed four of the larger pneumatics and will start on the crescendos next. Will try and get some progress photos for the next blog entry. December 01 Scanner and perforator progressThe much anticipated Keystone catalog arrived in the mail last week and I'm told today the web page designer has all the information he needs to complete his work. Richard Groman tells me we should see www.keystonemusicroll.com coming online within a week. With a pile of 150 rolls in the queue to be scanned, I have made a more efficient work center with two scanners for accomplishing this task. The prototype scanner (serial No. 1) had earlier been removed from its original equipment rack and stripped of its original temporary wiring. I decided it should be made the same as the 8 production scanners and is a lot more portable in this configuration. The photos show the wiring in progress and the end result. Yesterday, I spent a little over 10 hours keeping the two scanners running and managed to scan 61 Ampico rolls, averaging about 10 minutes per roll. This mode keeps me quite busy and there is no more staring at the monitor and falling asleep. Today, the MIDI files were rendered and uploaded to www.IAMMP.org .
The Aeolian American perforator is progressing nicely. The wiring has been added for powering the drive motor. I made standoffs to place under the motor for properly tensioning the belt and after a little oiling and greasing, the machine sprang to life with the punch ram bobbing up and down at about 10 strokes per second. Before the die set could be installed, I felt it prudent to remove all the surface rust and hardened grease from the punch ram assembly and all moving parts. Nearly all of that is done now except for a few paper rollers. I did a dry-run assembly of the solenoid bank and decided the terminal boards were in too poor condition to be reused. After some searching, I found the perfect substitute. Insulated turret terminals with a 6-32 male stud will fit the two existing tapped holes in each solenoid. Made by Cambion, they should arrive in mid January. The organ rank driver and microprocessor should be shipping from the supplier in a couple weeks. I also purchased a Yamaha MDF-3 for sending MIDI data to the perforator. We did have some success testing Julie Porters Compact Flash MIDI filer in Bethlehem on the last visit. It sequenced several roll master files successfully and we think we can work out a couple of minor bugs once I have this perforator in a state where the testing can continue.
Other ongoing projects such as the Chickering restoration will be moving forward during the Winter and I've got my wheels turning devising a way of scanning or photographing the Ampico master stencils. Richard was gratious enough to lend me an Ampico-A 30 to 1 stencil and an Ampico-B 30 to 1 stencil which are different in width. He will also be sending a few older examples with the wavy left edge (pumping line) and some of the more fragile ones which can no longer be run on the perforator. I will try and find a way to preserve the information on some of these extremely delicate master stencils, some of which are literally falling apart. More on that in future articles.
October 27 Stepper motor R & DAnother week-long visit to Keystone Music has yielded significant progress. I arrived in Bethlehem on Saturday, October 18, prepared to modernize the Ampico MIDI perforator with a stepper motor drive which will allow for nearly unlimited ratchet rates to be accomodated. It was found there are no gear combinations available for accurately perforating DuoArt rolls on the Taft Pierce machine and it was decided a stepper motor drive system would be the most practical solution. Of course, a DuoArt die set will have to be fabricated, which should be quite easy now that an Ampico model exists on a CNC multi axis milling machine. With slight modification, a DuoArt die set can be made. One of my objectives on this visit was to carefully examine and measure the pull-through rate of the original Aeolian perforator which had been used to punch many of the Ampico rolls during the 'Merger' period. In the photos, you will see how a dial indicator was attached to the chain-driven paper feed mechanism and precise measurements were made for both the coarse and fine ratchet wheels. The measurement was found to be 44 and 66 ratchet respectively.
Another objective was to precisely measure the drive train of the Ampico perforator in order to install a stepper motor in place of the intermittent transmission. Some research was necessary to determine the amount of gear reduction and torque required to move the paper with a high degree of precision. I made modifications to the Excel gear calculator spreadsheet and created an equation which would display the exact distance the paper would move between each punch row. I found that by using some of the existing gears with a microstepping motor controller, the step rate could be controled with a precision of 30 millionths of an inch per punch row. It was necessary to do a speed calculation to make sure the motor would be able to move the distance required within the available time window. Since this time window was unknown, I had to make a timing diagram for the perforator. This was easily accomplished by measuring the pulley diameters and figuring the nameplate RPM of the motor. A piece of masking tape was wrapped around the large pulley on the punch ram. Marks were carefully made on the tape at every point where something was happening on the perforator. (see accompanying photo) Since the RPM of the pulley and its circumference are known, the distance between the marks can be converted into milliseconds. Now, the precise time window for moving the paper is known and the existing signal from the photointerruptor can be used with a fixed delay for triggering the stepper motor. I plan on employing a Parallax Basic Stamp microcontroller for generating a train of pulses for moving the stepper motor the proper distance for each punch row. A thumbwheel switch can be used to select values from a look-up table for each of the popular pull-through rates. All of this will allow Keystone to offer accurate recuts of many popular piano roll formats. Someone once said "You don't know what you don't know." Now I think it's safe to say we really do know!
A few of the remaining titles for the upcoming catalog were punched and we met with Jason, the web page designer on Thursday. Customers wishing to place orders will be able to do so online at www.keystonemusicroll.com . The page is still under construction but should be appearing within a few days. You will be able to place your order and pay online using Google's secure payment system. At least, this is the plan as I understand it at the present time. Richard Groman will also accept orders VIA his cell phone number which I will publish here as soon as the catalog is finalized.
On Friday, October 24, after most of the work was done, we had a bit of fun. On this 83rd. anniversary of the day Dr. Clarence Hickman conceived of the ball bleed valve, we rented a truck and visited Doug Miller in Silver Springs, Maryland to pick up Richard's newly restored 1926 Lauter Humana upright foot pumper. When we arrived, we were treated to a private demonstration of Doug's two manual Wurlitzer theater organ which had been carefully transplanted from the demolished Reading, PA Lowes theater by Doug's father into the basement of the Miller residence. Doug's friend and organist, Ray Brubacher gratiously demonstrated his skills at the keyboard. With permission, I will try and publish some of the video footage of his performance on YouTube in the near future. The highlight of the day was the demonstration of Doug's newly restored 1924 Mason & Hamlin Ampico piano. Records show the piano being delivered to a 'Mr. Mason' in Boston. This is quite a unique piano since in 1929, it was reworked at the factory, receiving a new serial number. Doug has made an excellent restoration of this piano and fitted it with a 'B' drawer. I had the pleasure of hearing some of Doug's fine recordings of this piano on CDs during the return trip home. Many thanks to you, Doug, for the sample CDs and for selling me your last set of neoprene Ampico motor mounts. (The previous two sentences corrected 2/15/09)
To top off the weekend, I had the privilege of reading Richard Groman's autographed copy of Larry Givens' book, Reenacting the Artist. It was here that I discovered the coincidental occurence of the timing of our visit to Doug with Dr. Hickman's invention of the Ampico ball bleed valve.
The last day at Keystone was spent packing several boxes with about 150 original rolls to be scanned which will be offered in future catalogs. August 10 Production resumes on Ampico 20 to 1 rollsFor the first time since Ampico began producing 30 to 1 piano rolls in 1923, Keystone has gained the ability to reissue Ampico piano roll titles mastered in the older 20 to 1 format which was originally implemented by Rythmodik. Few of the 20 to 1 masters were retained apparently because the popular dance music of the time tended to be replaced rather quickly by new material and would drop off the charts. A few holiday titles would remain in demand and continued to be produced in later years but for the most part, the majority of Ampico rolls produced after 1923 were done on the new stencil machine in the 30 to 1 format... until the merger with Aeolian.
With the last batch of rolls scanned and punch masters rendered, I traveled to Bethleham, PA with CD ROM in hand and spent the latter half of last week assisting Frank Thompson in the cutting of many hundreds of Ampico rolls for the upcoming catalog. Quite a few 30 to 1 titles were run, then I changed to the 20 to 1 gear set and the remaining time was spent cutting that format. The equipment ran well and the rolls look and sound flawless.
For a video of the perforator in operation, try this YouTube link:
We briefly installed the 21 to 1 gear set and punched the unissued Paderewski 'Etude' roll recovered from a surviving Aeolian cardboard stencil. Later, after playing it on the Knabe, I discovered I had made an error in translating the T0-T8 control tracks to the corresponding notes on the musical scale. In short, they were transposed down by two steps. A corrected master punch file has been sent to Keystone for recutting.
We took time out Thursday to go visit the John Zeiner piano shop in Allentown. It did not take long to decide this would be the shop where the full restoration of my 1926 5'-8" Chickering will be done. After looking at the facilities and several beautifully restored pianos, I spoke with a few of the technicians and the owner, John Zeiner. It will be 8 months to a year before the work will commence. This allows plenty of time to correctly restore the Ampico expression system, a subject which will undoubtedly occupy these blog pages in coming months.
July 27 Getting the lead outWhile waiting for material and supplies to arrive from Player Piano Company, I have been disassembling the Ampico drawer and sketching the tubing interconnections. As the photos show, the lead tracker bar tubing was corroded through. All tubing has been cleaned off the fanning strips and tracker bar and the lead has been melted down into several ingots. All components have been removed from the drawer and now await restoration. The transmission and play controls are full of cracks in the pot metal, as predicted, and I am thankful for having been advised by Frank Thompson to obtain replacements in a recent eBay auction.
For a brief time, the treble end of the drawer was home to a mouse and she even stocked the pantry with one small sugar cookie. The mouse must have met with an untimely spring loaded fate as she never returned home to eat the cookie and there was very little of the 'evidence' one would expect to find in such circumstances.
Frank has generously provided some scans of past issues of the AMICA bulletin with relevant historical Ampico documentation. I have been able to determine without a doubt, the configuration of this particular Ampico player mechanism fits the description of the type 8 system. A cutout in the left side of the drawer for preventing false triggering of the OFF switch is only found in type 8 systems. The lost motion pneumatics in the stack and the right-side-up secondary valves are also important clues.
I performed one tuning on the Chickering piano and, amazingly, it settled on A440 on the first attempt. After a few days of hand-playing, a couple of strings settled a bit and can be fixed up quickly. The tuning pins seem to be adequately tight. We are very pleased with the overall tone and our daughter has all but abandoned the Milton upright.
Within the next few weeks, I should be making another trip to Bethlehem, PA. to return the most recent batch of scanned rolls. (assuming I do find time to scan them) While there, I will be paying a visit to a well known piano restorer in nearby Allentown, PA. Upon returning from that trip, there should be more news relating to perforating operations at the Keystone Music Roll Company. We should be punching a trial run of the unissued Paderewski roll during that visit.
A new custom built MIDI filer with external synchronization capability for use with a piano roll perforator is being tested. Many thanks to Julie Porter for her outstanding efforts toward making this happen. The device will be able to store large numbers of MIDI punch master files on a Compact Flash memory card. July 13 Chickering Ampico arrivesWhen the phone rang Wednesday and the caller wanted to confirm my address for the Friday 'delivery', I thought he was talking about the dumpster I had ordered for Friday. The dumpster arrived on schedule and I spent the rest of the afternoon commencing to fill it. While relaxing after the days work, there was an unexpected knock at the door. "There's a piano at the front door!" Although I wasn't quite ready making a space for the piano, I was still delighted it had arrived. The movers assured me there was room to fit it into our living room and brought it in. We left it vertical on my dolly until the player mechanism could be removed. As you can see from the photos, no time was wasted getting it disassembled. It's quite remarkable how much of the rubber tubing was still intact since nearly all of it was hard and brittle.
Today, we placed the piano on its legs and it produced music for the first time in many years. In spite of the time spent in storage and the stresses of moving, the keyboard action and tone were quite satisfactory. Naturally, several tunings will be needed but it didn't sound too far out while some of us hand-played it. I've noticed most readers seem to prefer the photos. Enjoy!
July 06 Unissued Ampico Paderewski Etude ready to punchSubstantial progress has occurred in 3 areas. 1, The unissued Paderewski Ampico B cardboard stencil has been converted to a Punch MIDI file and is ready to be cut on the perforator. 2, A new paper feed mechanism for the late Aeolian perforator has been engineered to match the pull-through rate of the original Ampico perforator. 3, A late 1926 Chickering 5'-8" Ampico piano in original unrestored condition has been purchased for use in Hyde Park.
First some history. The Paderewski performance of Chopin's Etude Op. 25 No. 9 was issued as DuoArt catalog No. 6097-6 in 1919 according to the Charles Davis DuoArt book. Among the small handful of surviving Aeolian cardboard stencils in the Richard Groman collection, was this Paderewski master with Ampico expression coding. Curiously, this title was never issued as an Ampico roll. Outside of Richard's small handful of cardboard stencils, no others are known to exist. It is fairly well known that after the merger, most of the late Ampico recordings were mastered on these cardboard stencils and punched on Aeolian machines like the one in last weeks photo album with the chain-driven tractor feed pulling the paper through the die set. Since most of the masters produced after the merger were unavailable in recent times, a special one to one perforator was built (by Harold Powell in the mid to late '70s) to function as a roll copier. It was built upon one of the original Taft Pierce Ampico perforators. It had a roll reader with a special tracker bar which would look ahead at the chaining pattern and, using some digital timers and logic gates, would try and reproduce the chaining as best it could. Many in the scanning community were puzzled as to why some of the Ampico rolls cut within a certain time period did not exactly match the older original rolls. Now you know the story. These rolls were not made from original master stencils, but copied from an actual 1:1 roll. Keystone acquired the 1:1 machine and it was decommissioned a few years ago to be converted for operation with a MIDI interface. Although the parts were not discarded, its not likely they will be pressed into service again. A new die set has now been installed in the MIDI machine and we're ready for production.
Now back to the Paderewski master... An original copy of the DuoArt roll was provided for scannning. Using Warren Trachtman's Scanimage to Punch MIDI converter, a DuoArt punch master was created. An emulated MIDI file was also rendered to give an approximation of what the performance sounded like. Next, using Cakewalk, a copy of the punch MIDI file was made and all control tracks were deleted except for the sustain pedal. Comparing the cardboard stencil with the computer generated Punch MIDI file, I was astounded by the degree of accuracy which has been achieved. Congratulations Warren! I spent the entire day yesterday comparing every single perforation in the cardboard stencil with the Punch MIDI file. It really is punch-for-punch accurate. While checking and comparing the note fields, I also manually added all of the Ampico expression codes. This task was made much easier by setting the snap grid in Cakewalk to 32nd notes. Doing this makes it impossible to place the cursor on locations outside of the grid. All events will start and end on increments of exactly 24 MIDI ticks. Another labor saving device was the measuring stick I made for quickly identifying the control tracks. The coding is definitely for the Ampico B system. Until this master can be punched, I don't have any way of listening to the Ampico version to compare its dynamic expression with the DuoArt roll.
Some observations about this cardboard stencil. This doesn't necessarily apply to all other similar stencils. Most of the note field and sustain track was prepunched. The holes are perfectly aligned, cleanly cut and round. Probably done on a special copying machine reading the DuoArt master. The DuoArt expression tracks were omitted. An editor marked-up the stencil with a blue china marker. A laborer would then manually punch out all of the marked areas. He would use a small round punch and a rectangle punch which was the length of 3 punch rows. You can clearly see the difference between the hand punched holes and the machine punched holes. There were more than a few manually added holes to the note field. Several editing marks were made in error on this stencil and crossed out indicating they should not be punched. I saw one error in which the laborer failed to notice one small blue mark at the start of a note and didn't punch it out. Looking at the DuoArt scan, this hole was punched and I corrected the error on the Ampico version. There is no doubt the Ampico performance is an exact copy of the DuoArt roll. I don't believe this stencil was ever run on a perforating machine because there were still quite a few hanging chads in the rectangular hand punched holes. These stencils were read by a stylus mechanism which would try and poke a row of probes into each punch row. If the probe dropped into a hole in the stencil, the interposer would move and allow a hole to be punched in the roll being cut. These hanging chads would have blocked the probes from moving past the surface of the stencil. The large tractor feed holes on each edge are spaced 1 inch on center and there are 6 punch rows per inch. The stencil is pre-printed with a grid of squares and the musical note scale appears on every sixth row. There are 99 squares per row and for Ampico B coding with 100 tracks, the sub-intensity track would have to be punched within the narrow margin between the left edge tractor feed holes and the last marked square at the left end of the punch row. Sub intensity was not used on this stencil. (Was it ever used for Paderewski?) The amplifier track was used briefly near the end. The left edge is the treble end of the scale. The 'cardboard' is .012" thick and is wound up like a clock spring. You could not unroll it on a flat surface without damaging it. I wouldn't say its extremely fragile or near disintegrating but it is a bit yellow and would crack if folded.
June 17 Ampico perforator studyMany thanks to all of you who have helped support the IAMMP archive by contributing your roll scans. Although it might appear the author has been inactive, in fact, a great deal of important work has been going on behind the scenes. I have just returned from a very successful week-long visit to the Keystone Music Roll Company. Over the past several months, Richard Groman, Frank Thompson, myself and Larry Chase have collaborated on the implementation of a MIDI interface to one of the original Ampico perforators. The purpose of this most recent trip was to select the appropriate gears for cutting rolls having pull-through rates other than 30:1. I made measurements of the existing gear ratios, the reduction rate of the intermittent transmission and the effective capstan circumference. A digital counter was used to count the revolutions of the punch ram while measuring the distance between two marks on the paper. This yielded a precise figure of the actual pull-through rate. I then created an Excel spreadsheet which allows for repeated measurements of the pull-through rate whenever there is a change in paper thickness or the number of copies being cut. This spreadsheet allows alternative change gears to be specified and the resulting pull-through rate is displayed. Using the established pull-through rate, the length of a punch master file can be input in the form of Measures, Beats and Ticks. Repeated testing was done with several punch master files and we have found the calculator to be able to predict the actual length of the roll to be cut within an accuracy of better than 1.5" for a 30 foot roll. I added an input field for the measured length of the original roll and the deviation is displayed as well. This allows for some additional gear combinations to be tried in order to match the predicted length of the copy to the original. After some new gears arrived, we had excellent results cutting 20:1 rolls and many of these will appear in the next Keystone catalog.
We assembled the tractor feed mechanism to the original Aeolian perforator and threaded one of the surviving cardboard stencils for the photo in this weeks album. Among the surviving cardboard stencils, Richard found an unissued Paderewski recording. It appears to contain Ampico expression coding and is similar to the DuoArt roll shown in the photo. I have accepted the challenge of remastering this roll. The DuoArt roll will be scanned, a punch master will be rendered and the DuoArt expression tracks will be manually replaced with the Ampico expression coding from the cardboard stencil. No one knows if the original stencil was ever finished, but we will soon hear the results! Many people are not aware, a great many of the late Ampico popular rolls were punched on the Aeolian perforator with the cardboard stencils. Hence the need for scanning and digitally remastering the popular rolls.
As the newest member of the Keystone team, I have been helping to scan many of the titles which will appear in the upcoming catalog. I have also brought home a couple boxes of hot titles to scan for the Winter catalog. New die sets have been ordered for both Ampico perforators and full production will resume once they arrive.
In order to better control the punch-for-punch accuracy of the product, Richard has been kind enough to lend me one of the late Aeolian perforating machines which will be fitted with the same microprocessor and relay as the Keystone Ampico perforator. This perforator will be used to punch proof-copies which will be compared with the original rolls. Comparisons have already been done at the Keystone facility and we have gained a high degree of confidence that the remastering software is producing accurate results. All titles punched last week were auditioned on the Knabe Ampico-B piano with outstanding performance quality. May 14 Do we have idle scanners?I would like to take this opportunity to speak to several of my MK4 archival scanner recipients who have not yet begun sending scans to the IAMMP archive. I'm sure none of you have forgotten the promise you made to me to share the scans you are making on the scanners I built for you. You have two choices available for honoring this commitment. Either upload the scan files to the IAMMP or burn a CD and mail it to Larry Doe. It would be comforting to hear from those who have been inactive. Some kind of confirmation regarding each persons preference would be greatly appreciated. There are limits to the number of FTP accounts which can be supported by the IAMMP server and the inactive accounts should be closed if they are not going to be used. There are many benefits to being an active contributor to the IAMMP since you have full access to the files uploaded by the other contributors. As an active contributor, you are more likely to receive favorable treatment when Warren makes updates to his software. It is true that a substantial number of Warren's CIS files are not yet available to all of the contributors, but Warren has kindly provided me with an FTP account on the other server where the balance of the CIS files are stored. I tested this connection today and those files really do exist. I looked at a few of them and the quality is top notch. With Waren's permission, I could act as a liason and provide selected copies of his CIS files to other contributors on a request basis. This should satisfy any of you who were concerned about the availability of these files. Warren is as concerned as I about the absence of a few of the people who signed on. It's a small miracle Warren has been able to make the IAMMP archive function as well as it does and I think we sometimes fail to applaud the excellent work he has done. His detailed letter to me this past week has given me a new understanding of the skill level required to manipulate this huge database while keeping everything legal and secure. He has invested a huge amount of time and money into making this a success and I feel it's crucial for ALL of us to show our appreciation by sending in our scan files in a timely fashion.
If you will be choosing the CD option, please let me know. I won't be expecting your first CD until you have had reasonable time to scan enough rolls to fill up a CD. The first CD I mailed to Terry Smythe took me about a year to generate. Please don't forget my standing offer to buy back any inactive MK4 scanners.
If anyone is unfamiliar with a particular software product or if any additional training is needed with the uploading procedure, please don't hesitate to ask. I'd be glad to spend some time teaching people how to process their scan files so they may become contributing members of the IAMMP.
Regarding Keystone, a lot is happening with the debugging of perforator No. 2 which has the microprocessor and MIDI interface. In a recent test, the revised microcode enabled the MIDI port to communicate with a PC running Cakewalk and the punch rows were clocked out in sync with the pulses from the photodetector on the main cam of the perforator. A few of the note-off commands were missed and this resulted in some rather long slots in the paper. This test did produce a roll which could be played on a piano with satisfactory results. Larry Chase is having a second look at some of the timing issues and we may see another revision of the microcode within the next couple of weeks. Until perforator No. 2 is up and running, scanning operations at Keystone are on hold. After the perforator is debugged, I will be making a trip to Bethlehem to provide Richard Groman with some additional tutoring with the processing and uploading of his scan files.
I've posted some new photos of the WHUD transmitter site. These show the delivery of the first two cabinets of the new Nautel V40 transmitter. They have been positioned in the spot formerly occupied by the old Harris backup transmitter. On Thursday, these new transmitters will go on the air for the first time. Each cabinet contributes 10 kilowatts which is fed into a hybrid combiner and then to the antenna. If successful, I will then begin dismantling the old main transmitter to make room for the second half of the Nautel V40. May 07 Honor(on'er), n. 1. glory; renown; fame: The honor of inventing the telephone belongs to Alexander Graham Bell. 2. credit for acting well; good name: It was greatly to his honor that he refused the reward. To their honor, the patriot nobles did not take thought for themselves alone (Edith Thompson). 3. a source of credit; cause of honor: to be an honor to one's family, school, country, etc. 4. a nice sense of what is right and proper; sticking to action that is right or that is usual and expected: Honor is sometimes found among thieves (Scott). 5. great respect; high regard: True wealth I hold in great honor (John Ruskin). 6. an act of respect: funeral honors. The queen was received with honor. 7. high rank or position; dignity; distinction: Knighthood is an honor. Papists were admitted in crowds to offices and honors (Macaulay). 8. chastity; virtue. 9. Golf. the privilege of teeing off first, awarded to the player or side winning the previous hole. 10. Poetic. a decoration or adornment: Geranium boasts Her crimson honors (William Cowper).
The World Book Encyclopedia Dictionary 1964.
April 30 Richard collection movedAll of the 'keeper' rolls from the Richard collection have been scanned and the CD recording project is now complete. Saturday, the CDs were delivered to the owner and the remainder of the purchased rolls were boxed and transported back to Hyde Park. Sunday, I repacked the rented storage unit and cleared one entire shelf for the Richard collection. The photos do not show the 4th tier in the loft space, but the loft is able to hold the same number of boxes as a typical shelf. The exact number of rolls in the Richard collection is known to be 1458 and the owner kept 57 of them. What you see on the shelf are the 1401 rolls sold to me. This exercise has given me a better estimate of how many rolls are actually in the storage unit since there are 12 equally sized shelves including the loft spaces and each shelf has a capacity of about 1500 rolls. The density of the unboxed rolls is a bit higher since they fit closer together. The estimate is between 18000 and 20000 in the storage unit and another 10000 or so in my home. We will probably never know the exact number. Most of these rolls were acquired before I ever thought of making a roll scanner. It's quite satisfying to know it is now possible to preserve the content of these rolls and make them available for others to enjoy.
I've included some photos of a project my day job has me occupied with. They show some views of the WHUD transmitter site in Putnam County. These photos show the old standby generator being removed and replaced with a larger 150 kilowatt unit which has the capacity to run the entire facility. Earlier that week, I had to cut up and destroy a large backup transmitter. A temporary transmitter will be installed in its place while the 30 kilowatt main transmitter is dismantled and removed. All of this is being done in preparation for the arrival of a new Nautel transmitter capable of 40 kilowatts. The licensed transmitter power output is 28 kilowatts combined with the gain of the antenna to produce an effective radiated power of 50 kilowatts. Later next month I will be performing a large part of the installation of the new transmitter. Next week, I will be in Binghamton to assist with the construction of several new studios for National Public Radio. April 08 Scanning translucent paper32 of the 57 'keeper' rolls were scanned today. Nearly all of them were of the newer translucent variety. A few weeks ago, I wrote about some methods for scanning this type of roll and had ample opportunity to put those methods to practice today. The process has been further fine-tuned today and I will share the results with you. The first photo shows a typical translucent QRS word roll from the 1960's and the capstan can be seen through the paper. The MK4 card on this scanner is the new style double-sided board produced by Keystone which has color selection jumpers for both video inputs. For this scanning mode, the holes in the paper must be the darkest objects in the video signal. In order to achieve this, the lamp holder is removed from the glass cylinder so there will be no reflection from the white phosphor of the lower lamp. The color selection jumper for the 'notes' channel is set to match the color of the ink used to print the lyrics. We want the lyrics to appear as bright objects to the 'notes' channel so they will blend in with the background. For example, BLUE ink reflects BLUE light. The color selection jumper for the 'words' channel is set to RED so that the BLUE or GREEN ink will absorb the RED component of the light source and the lyrics will appear as dark objects to the 'lyrics' channel. In this mode of scanning, the holes appear in the lyric image (see photo 2). However, the 'notes' image does not contain any of the lyrics (see photo 3).
When the clipper levels are set correctly, The words and outer margins will appear YELLOW on the display. The notes and inner margins will appear RED on the display. The background will appear GREEN with a small amount of YELLOW hash in the background which can be minimized by small adjustments to the 'lyrics' clipper pot for each roll scanned. These display colors are hard-coded into the SCAN2005 software and are not user adjustable.
The CIS files created by this method will have the notes data on the correct side of the Bicolor image. The lyric image will contain the lyrics combined with the notes.
I processed all of today's scans into MIDI files and a small amount of editing was required to clean up a few false notes. There is a tendency for some areas of the lyrics to bleed into the 'notes' image when the ink is particularly dark. I usually see this more at the beginning of the roll where the ink roller was laden with a heavier dose of ink. During the roll manufacturing process, a stencil was placed over the entire length of the roll and an ink roller was rolled from one end to the other. At the start, the ink would be darkest and as the roller was rolled down the length of the stencil, the lettering got lighter and lighter. It is in some of the areas where the ink was heaviest, the lyrics tend to come through into the notes image.
Overall, the results were surprisingly good and the editing was confined only to the area of the image where the darkest lyric printing was. The old problem of having a huge number of false notes from the transmissive scanning method has been completely eliminated. For the older rolls with opaque paper, I would still recommend the tried and true transmissive/reflective method using both lamps simultaneously.
I have been contacted by another roll collector in New Hampshire with about 1000 old rolls to be scanned. I guess I'll have to hurry through this batch of 1345 rolls from the Richard collection.
April 01 Add 1,345 rolls to the collectionA deal has been struck today, with the heir of a collection of 1,345 88 note piano rolls in New Hampshire. This collection was amassed by Robert and Theresa Richard (now deceased). I have just returned home with one pickup truck load. The round trip required eleven hours of driving. Fortunately, the rolls are all packed in large Tupperware bins which protected them from some rain encountered on the way home. One of the bins is large enough to give your St. Bernard a bath. As details of the collection emerge, I will post them on this blog. For simplicity, I will refer to it as the 'Richard Collection'.
Here's what I know so far. This collection has been in the hands of the same family for two generations. The grandmother began buying rolls in the 1920's and passed them on to her daughter (Theresa) who replaced the original piano with a Kimball spinet sometime in the 50's or 60's. After the Kimball was acquired, a great many new QRS, Klavier and Playrite rolls were added to the collection. Better than two thirds of the collection consists of these newer rolls. None of the rolls have been sold off and the collection is complete. Theresa's granddaughter (I'll call her Dee) remembers her grandmother spending up to four hours of every day playing her pianola and literally wore it out. Sometime later, a vorsetzer was acquired to play a small organ the family had. By the 1980's, the Kimball also became leaky and the suction motor burned out. The stack is now in the capable hands of Ed Bourdleau in Hooksett, New Hampshire. I have an extra suction box and will be donating it for the restoration of this piano. Dee did not sell me all of her mom's rolls. There are 57 'keepers' she will be holding on to which were titles she remembered her mom being particularly fond of. Dee has gratiously agreed to allow me to scan these 57 titles and I will be producing a set of CD recordings for her to use until the Kimball restoration is completed.
March 12 Ampico dance instruction rolls gaining in popularityIt seems my mention of Denishawn dance instruction roll recordings in a previous blog entry has attracted the attention of Thomas Gladysz, director of the Louise Brooks Society www.pandorasbox.com . On this web site, I learned that on March 11, 1924, Louise Brooks performed with the Denishawn Dance Company in Fond du Lac, Wisconsin and was best known for her role as 'Lulu' in the 1929 German film "Pandora's Box" . She also performed in the Ziegfeld Follies.
Thomas asked "Do you think these recordings will be made available commercially. Or perhaps on-line. I would love to be able to hear them."
I'm told there is a very good possibility the Keystone piano roll company will be reissuing the complete ten roll set of Denishawn dance instruction rolls. At the very least, it should be possible to make a set of MIDI files available to those who would like to hear some of the music on their computers. I have heard a few of the rolls and they do indeed contain some musically pleasing selections. I don't yet know if Terry Smythe's upcoming recording is to be for the exclusive use of the dance museum. (see previous blog entry)
On the subject of dance instruction rolls, Frank Thompson has been kind enough to send along some high quality photos of the Charleston instruction sheet which accompanied Ampico roll 207331, Played by Adam Carroll and issued in May of 1926. A few photos are included for your enjoyment.
I have made an offer to Dorothy Hickox in New Hampshire to scan her grandmother's collection of 1,500 88note rolls. We're still waiting for a final decision.
I've been reading some of the chatter on the Rollscanners forum regarding the subject of adding expression to regular 88note rolls. The response which really struck a chord with me was from Johan Liljencrants of Stockholm, Sweden. He wrote:
A music composition is normally represented with notes and supplementary performance instructions like tempo, accents etc. A human performer will not reproduce this raw material slavishly, but rather interpret it in a hopefully musical way, something he may have learned in a conservatory. No doubt this will introduce some randomization, but more important is that he will apply performance rules, which depend also on genre. Such rules have been researched for several years by Johan Sundberg, Roberto Bresin, Anders Friberg and colleagues at my former employer institution. You may familiarize with this rather intricate subject at http://www.speech.kth.se/music/performance . Here you also find the free dowloadable software 'Director Musices' to apply such rules on midi files. February 25 Automatic importation of ANN data achievedThe missing Denishawn dance lesson No. 2 arrived Thursday and has been scanned and archived. Since this Ampico roll had been punched two days earlier from the original master stencil, on the original perforating machine, I think it is safe to classify it as an 'original' roll rather than a recut. However, the new paper was very translucent and the scan had to be done by the reflection method. This is done by simply switching off the lamp under the glass and reducing the 'words' clipper level until the holes appear as red slots on a black background. A regular bicolor CIS file will be created with the holes appearing in the 'words' image. The 'notes' image is blank in this case. If you try to process this file with the Stibbons SCN33 program, you will get a SCN file containing zero bytes, as Terry Smythe discovered. The Stibbons software cannot be used to process these scans for two reasons. The notes image is negative and it appears in the 'words' half of the bicolor file. There are utilities available for splitting the bicolor file into two separate files and then converting to bitmap and back to CIS which will get the negative switched back to positive.
You can avoid all of this complexity by simply processing the scan with Warren Trachtman's ScanImage to MIDI converter. When you open a reflection scan file, you will see the notes over on the right side where you usually see the words. By entering a negative offset, the notes will move to the left. Existing software does not allow entering an offset higher in magnitude than negative 10 inches. To nudge it the rest of the way, use the left arrow button repeatedly. Warren's software doesn't care whether the image is positive or negative.
After several Emails, Warren and I succeeded in solving a few issues with settings and elevating Terry's confidence level with the ScanImage converter software. A new Beta version of the converter emerged late Friday evening which will allow offsets of plus or minus 20 inches. Now for the buried lead story!!!!!!
A new feature has been added to the ScanImage Beta version which will automatically import data from annotation files. This will be a tremendous time saver for those of us running Anthony Robinson's Scan2005 software on MK4 scanners. At the end of each scan, you can enter all of the information from the roll leader and box label while still in DOS mode and you only have to handle the roll once. A whole batch of rolls can be scanned in this way and rendered to MIDI later after moving to a Windows machine.
There are two boxes in the diagnostic settings menu which must be checked; "Embed roll info during conversion" and "Pre-populate some entries using ANN file data if found". The ANN filename must have the exact spelling of the CIS file and reside in the same folder. You should limit the keywords used in the annotation file to those used in the Embedded Keyword Dialog Window. Other keywords can exist in the ANN file but they will not migrate into the MIDI file. I divided the keywords in the dialog window into three categories. 1, keywords which you would want to fill in while scanning the roll, 2, keywords which do not change often and can be assigned default values and finally, 3, /roll_type: which is always assigned the name of the template which was used (generic_uniform, Ampico, DuoArt, etc.) . In the first photo, you will see a screen dump of the Scan2005 configuration file editor with all of the keywords from category 1 filled in. I left out /meter: but there is space to add it.
In photo 2, you will see a screen dump of the keyword dialog window with all of the category 2 keywords filled in. (fields marked with *) A checkbox at the bottom allows the category 2 keywords to be saved as defaults. In the photo, you can also see some of the category 1 keyword data which was automatically brought in from the ANN file.
I have scanned, processed and archived several DuoArt rolls this weekend including some which were Play-Rite recuts and those were done with the reflection scan method. All of the new features performed flawlessly and I'm very excited and happy to see this advancement successfully materializing. It will be a tremendous timesaver for all of us using the SCAN2005 scanning software. Many thanks to you, Warren Trachtman.
For those of you watching the IAMMP archive, you will see Warren has moved about 500 of his recent MIDI scans from the Trachtman.org site. In time, we should see the CIS files migrating over as well.
Post Script:
After reflecting on what I just wrote about the reflection scanning method described above, I realized there is an even simpler way to do it. You still switch off the music lamp under the glass, but then you would reduce the 'Music' clipper level until you see black or red slots on a green background. This places a negative image of the holes in the 'Music' side of the bicolor image where it belongs. You can also play with the 'Words' clip level to get a clean image of the holes in the 'Words' side of the image, but that is irrelevant. Using this modified method, you no longer have to use a negative offset with Warren's ScanImage to MIDI converter. I'm still not sure whether the Stibbons SCN33 software will handle a negative image, but it's easy enough to test.
February 20 Missing roll foundThe master stencil for the missing dance instruction roll has been found! The roll has been recut and a copy is now on its way to Hyde Park, NY for scanning thanks to Richard Groman. Although Keystone is now equipped with an MK4 scanner, we have not yet installed communication software or networking capability on their scanning computer.
8 of the master stencils in this series were located Monday and Richard is quite sure the remaining 2 are there also and will be found when time permits.
Many thanks to the Keystone Piano Roll Company and the dedication of Richard Groman, Frank Thompson, Terry Smythe and Anthony Robinson for their rallying efforts toward restoring the complete set of Denishawn dance instruction rolls!
It is hoped that a copy (original or photocopy) of the dance instruction booklet can be made available to Keystone. There is a very good possibility the complete set of rolls could be reissued. February 18 Dance instruction roll needed.Ampico dance instruction rolls are to be used for an upcoming live demonstration of the Denishawn technique at the National Dance Museum in New York City. The museum provided 8 original rolls to Terry Smythe for scanning and production of an audio CD for recreating this memorial event which will highlight the teaching techniques of Ruth St. Denis. Rolls 2 and 5 were missing and thankfully, a scan of roll 5 will be provided by Anthony Robinson. Roll 2 is still needed and Keystone will let us know if their collection contains this missing roll. The catalog number is 67913 and was played by Louis Horst in 1927. If anyone reading this knows of an existing copy of the missing roll, please contact Larry@DoeSigns.com if you would agree to have the roll scanned. Terry has already posted the 8 rolls he's scanned on the IAMMP archive.
Thanks to Frank Thompson for some photos of his copy of the dance advertising brochure. The museum also possesses the dance instruction booklet and notes which will aid in training the performers.
As soon as the missing dance roll is found, I will make an immediate update to this blog.
Thanks to Walter Tenten for giving a shout-out on MMD for the IAMMP archive and for supporting the concept of sharing piano roll scans. After all, what's the point of archiving the rolls if they will not be shared in some way?
A few years ago, I was looking through some of my dad's papers in the basement and came across some notes he had made while repairing an 'Ampico grand piano' for a former employer. Last week, I succeeded in contacting this gentleman and his Chickering Ampico with rolls still survives. Although the piano is not playing, the man is very interested in having his rolls scanned, so I have offered to do this for him. Stay tuned for news as it develops.
February 11 Back to the business of preservation.Scanner No. 8 is comfortably settled into its new home and the scanning of the Keystone Ampico rolls is underway. A trip to Keystone is planned tentatively for Feb. 24 to provide some tutoring and guidance with the rendering and uploading of roll scans. Batch 2 of the Ampico rolls delivered last week has been scanned and uploaded. Thanks to Frank Thompson for providing this weeks photos.
Warren Trachtman inquired about availibility of the Keystone perforating facilities for making recuts in small quantities for custom orders. Once the microprocessor on perforator No. 2 has been tested with synchronized MIDI playback, I'm told Keystone will consider accepting such jobs within the limitations of the machinery. The present setup is able to punch Ampico or regular 88 note rolls with 356 punch rows per foot. Other ratchet rates may be possible in the near future once the proper gears are obtained. Each perforator session produces 15 copies.
If you are a contributor to the IAMMP archive, please remember to avoid using characters in the filenames which are not alphanumeric. (hyphen, underscore and apostrophe are allowed) A few files appeared recently with # in the filename and this prevented the MIDI files from being available for download. The embedded (/title:) keyword may contain any characters.
I'm told Warren will soon be moving several hundred of his recent scans from his Trachtman.org server to the IAMMP site. Many thanks for the recent contributions by Michael Swanson and Robert Perry. I have begun the processing of some of my 800 earlier scans which were not annotated. 35 titles were completed today. 10 already existed in the archive, so I only uploaded the 25 that were not duplicates. Only 787 more to go!
The O'hare rolls and audio CD have arrived back at their home in Oklahoma.
Congratulations to Richard Stibbons and collaborators for the preview into the future of roll scanning. 'Play as you scan' is a feature which was high on my priority list. If the predicted scanning rate of 8 feet per minute is achieved, I might actually be able to retire someday. Best of luck to you all!
|
|
|
