Fujitone 3A Keyboard Sample Pack

Here's the latest sample pack, a collection of sounds from a Fujitone 3A keyboard. This keyboard is very lo-fi squarewave toy with lots of grainy charm. The drums in particular are quite fun.

Included in this pack are multisamples of all 12 tones dry, with sustain, and with vibrato. Samples of individual drum hits and all 8 drum patterns are also included. Normalization was used between instruments, but not individual multisamples. Each sample was recorded at 48K/24 bit sample rate.

I'm offering it free of charge again for commercial and non-commercial use, but please don't redistribute the pack.

You can find it here:

Fujitone_3A_Sample_Pack.zip via Google Documents

Click File and choose Download to get all the files. Check out the readme.txt file for more information.

Elephant Organ Sample Pack

All samples were produced using a modified toy dubbed the Elephant Organ. Each key makes a distinct animal noise using square waves. A control was added to modify the pitch of the toy. Each of the 7 keys has been recorded at 7 different pitches, from low to high. When the pitch reachs a certain threshold, the synthesis breaks down and becomes glitchy. This pack includes 20 glitch samples. All samples were recorded at a 24bit/48K sample rate and normalized.

I'm offering it free of charge again for commercial and non-commercial use, but please don't redistribute the pack.

You can find it here:

Elephant_Organ_Sample_Pack.zip via Google Documents

Click File and choose Download to get all the files. Check out the readme.txt file for more information.

Sound Surface

This video shows a sound surface made from data collected along a stretch of Congress Street in Portland, Maine.

The height of each point is determined by the ambient sound level, while the position reflects GPS coordinates.The video gives a tour of each data point, showing the ambient light level at each point. The tour reveals the data collection process. The ambient sound level at each point also introduces a degrading effect to the mesh depending on density. The mesh becomes noisy when there are higher levels of ambient noise present.

I tried to show how an invisible force like noise maps to the visible world of light. The mesh degradation shows how noise can be equally as disruptive as visible light.

Processing 1.5 was used to produce this video. Data was collected with a custom-built GPS logger which read ambient sound and light levels. This data is smoothed using linear interpolation to cover gaps in the GPS coverage. The data is also smoothed using a distance threshold algorithm which averages readings which are close together. The point data is then converted into a mesh using Delauney triangulation. Data was collected over two days around and just after sunset to provide greater ambient light contrast.

The music in the background was created using a modular synth, 4-track tape recorder and looping pedal.

FM_SK Sample Pack

The sounds in the video above are from Seaweed Factory's latest sample pack, FM_SK.

The focus is on odd, grainy FM percussive sounds. All samples were produced using a Yamaha TX81z sound module. Each percussive sample was then processed using a modified Casio SK-1 Sampling Keyboard to give the pure FM tones more grit. Over 150 samples at a 24bit/48K sample rate are included in this sample pack. No normalization was used, nor was any tuning performed.

I'm offering it free of charge again for commercial and non-commercial use, but please don't redistribute the pack.

You can find it here:

FM_SK_Sample_Pack.zip via Google Documents

Click File and choose Download to get all the files. Check out the readme.txt file for more information.

Yamaha A4000 Encoder Fix And Compact Flash Drive

I recently acquired a Yamaha A4000 rack sampler, little brother of the Yamaha A5000. The synthesis structure is deep and it sounds great. However, this machine is from the late 90s and therefore has a few annoying quirks. First and foremost, the encoders are very finicky due to age and a design flaw. Secondly, the machine uses SCSI for storage and the required equipment is very impractical. Finally, with a SCSI hard drive and the case fan, this machine runs very loud. I wanted to fix all these issues.

I should say do these modifications at your own risk. These are not simple modifications and I accept no responsibility for any damage that may be caused. That said, here are the steps I took.

The most important problem was the encoder issue. The machine uses sealed encoders with a click switch which start to act erratically after a few years. Since the encoders are sealed, they are very difficult to clean. I tried the DeOXIT route, using the tiny gap between the push switch to get the chemical inside the encoder. This resulted in a very slight improvement, but did not fix the underlying issue.

After doing some research, I discovered that the settle time for the bouncing contacts inside the encoder is very low and can result in incorrect readings due to the software routines used by Yamaha. The solution is to slow down the bouncing signals coming from each encoder. A few others have made simple lag processors out of 1uf capacitors to achieve this result. Later it was discovered that adding a 470 ohm resistor in series with each capacitor will protect the encoder contact from excessive voltage. This modification will also electrically clean the contact each time it turns.

To begin, I dismantled the unit. Many black screws hold the top of the case on; remove the proper screws on the back, top, and bottom to remove the cover. The hard drive brackets must be removed to access the encoder board. The red arrow indicates where the brackets connect. Two screws on top of the encoder board must be removed and the indicated cables must be removed.

The encoder board is also held on by hex nuts and washers on the front. To access this point, remove the faceplate by removing the three screwes on the bottom. Remove the hex nuts and washers and the board should slide loose. Do not remove the metal bracket above, it is not necessary to access the board. Be sure to save each screw and hex nut in a secure place for when everything is reassembled.

Here is the solder side of the encoder board. The blue arrows indicate the pins of interest on an encoder. This is where the capacitors and resistors will be attached. Some people choose to attach the circuit directly to the board; I worried about short circuits with the case so decided to place the circuit on a separate board. The gray arrow indicates a common ground to use for the circuit.

Here is the modified board with the additional circuit. I used electrical tape over the connections for added strength and protection from short circuits. Originally I tried to use the edge connector for all the points, but the connections were just too close together. Be sure to use a common ground to save on the amount of wires that must be routed. Also be sure to leave enough slack to mount the board elsewhere in the unit. I used 24-gauge wire for the connections to the board.

Here is a diagram of the circuit. I used 50V capacitors as I had many extra, but lower ratings may also be acceptable. Everything was mounted on a simple perfboard from Radioshack. Run the wires from behind the encoders as the encoders are flush with the faceplate.

It was difficult to find a place to mount the board. I originally thought I could fit everything above the encoder board, but I would have to make a very skinny long board to do so. There is a hard drive bracket that was used to hold SCSI drive. This serves as a very convenient place to mount additional items if there is no drive present. I had to add additional mounting holes to fit the perfboard. Insulted standoffs were used to avoid short circuits.

Also visible in this photo is the SCSI-to-Compact Flash drive replacement. I used the same bracket to mount the additional equipment, again on insulted standoffs. Electrical tape was used on the top power connector to avoid short circuits.

To create the drive replacement you will need many parts:
  • SCSI-to-IDE bridge with a female SCSI 50-pin port and male IDE port (ACARD works)
  • IDE-to-Compact Flash drive with female IDE port
  • Compact Flash card up to 8GB (Transcend 133X 8GB works)
  • 50-Pin SCSI cable (18 inch cable used)
  • 4-pin IDE molex Y power cable splitter (or whatever it is called)
  • Molex-to-floppy cable for flash drive (only if IDE-to-CF is not molex)
  • Zip ties (of course)
  • Insulted standoffs (little black plastic pieces available at Radioshack)

The SCSI-to-IDE bridge is increasingly hard to find. Prices used to hover around $35 dollars a few years ago, but have shot up to over $100 in the last year. I guess these things are no longer being produced, get one while you can. Ebay is a good source for this part. I believe you need to set the terminator on the bridge to active, I can't remember.

Get the simplest IDE-to-Compact Flash drive possible, no dual card abilities. It is best if it can plug directly into the SCSI-to-IDE bridge. The two items will hold together very well just using friction. I mount everything using the screw holes on the IDE-to-Compact Flash drive.

The Compact Flash card should be fast to avoid timing errors as many conversions are taking place. The A4000 only supports 8GB of storage, separated into 1GB partitions. Anything larger will not be used. Be sure to get a name brand card as there are many, many counterfeit junk compact flash cards out there.

The remaining items are used for wiring everything together. The power connections for each unit are surprisingly tricky and many require additional converters and splitters. Use zip ties to keep everything tidy. Use the hard drive bracket for mounting as that is what is being replaced and it is easy to remove to add additional mounting holes.

There is no way around how slow SCSI access is on the A4000. For a slight (and I do mean slight) improvement, use an IDE drive instead and put up with the noise. You will need to format the drive into 8 partitions before it can be used. Also note, it takes much longer for the unit to start up when using drives. Expect to wait a 90 seconds or more for everything to be ready to use. I was unaware of this fact and thought things were broken.

Finally, if you remove the hard drive, you can disable the case fan by unplugging the two wire connector leading out of the power supply. The power supply was designed to cope with the heat of a SCSI hard drive for hours at a time. Since this source of heat has been removed, the fan can be disabled.

I used these posts and resources for my modifications:

EX5Tech: Fixing Faulty Encoders
Flickr of A5000 with caps installed directly on board
Yamaha A3000 Hacking: Knobs Befudgery
Yamaha A4000 User Reports (compact flash drive)

After these modifications, I have a totally silent A4000 with lots of storage and smooth encoders. The encoder modification is like night and day, the machine is now a joy to use and very quick to get around.

Casio SK-1 Rhythm and Tone Sample Pack

Here is a sample pack of the rhythms and tones of an unmodified Casio SK-1 sampling keyboard. Though the sampling feature of this keyboard is wonderful, the lo-fi drums and tones are also pretty neat.

All samples were produced using a unmodified Casio SK-1 Sampling Keyboard. Normalization was used across multi-samples, but dynamics between notes are preserved. Some samples are labeled to indicate note or drum pattern. All eleven drum patterns are sampled at the default rhythm. The fill pattern for each rhythm is also available. Each of the eight built-in voices has been multi-sampled at octave intervals of the note C. Individual drum hits have been extracted from patterns.

I'm offering it free of charge again for commercial and non-commercial use, but please don't redistribute the pack.

You can find it here:

Casio_SK-1_Sample_Pack.zip via Google Documents

Click File and choose Download to get all the files. Check out the readme.txt file for more information.

Casio PT-10 Sample Pack

After a long gap, it's time for another sample pack. This packs contains sounds from a Casio PT-10 mini-keyboard. Each of the 4 voices have been multi-sampled. I'm fond of the fantasy setting in particular. Loops of the 10 different rhythms have also been provided at 2 different speeds. Enjoy this super lo-fi blip blast! I'm offering it free of charge again for commercial and non-commercial use, but please don't redistribute the pack.

You can find it here:

Casio_PT-10_Sample_Pack.zip via Google Documents

Click File and choose Download to get all the files. Click File and choose Download to get all the files. Check out the readme.txt file for more information.

Receipt Printer Scroll ASCII Art

Printing an ASCII art scroll using a receipt printer. A collaboration between Seaweed Factory and Johnathan Cook.

Some test prints.

ScrollPrint Software

I've decided to release the software used in this project. It's a graphical application written in Java and a sketch for the arduino uno. It's currently untested as the printer is in transit to Glasgow; some corrections may follow. It was tested under Windows and will soon be tested in OSX. You can find a preview here (choose download in the file menu):


ScrollPrint v1.1 uses a thermal printer and arduino uno to print scrolls from a text file. Each line of the text file should contain 32 characters, no blank lines allowed. Save the file with no formatting; use the MSDOS text format option if available. An example text file is included.

This program requires RXTX binaries to be installed. See the following for installation instructions:


A graphical interface is available by running the ScrollPrint.jar file. From the command line run the following:

java -jar ScrollPrint.jar

The source for this application is included in the src folder. The project was written using Netbeans 7.1.

The sketch labeled scrollPrint.ino must be downloaded to the arduino uno. This sketch relies upon the Thermal Printer Library from adafruit. The interface circuit and library can be found at adafruit:


If printing ASCII art, try the following converters. It helps to squish the long axis of anything you are printing by about 75 percent due to line spacing.

http://picascii.com/: Feed it images 96 pixels wide to get 32 characters back.
http://www.text-image.com/convert/ascii.html: Set character width to 32, extra contrast helps.

Send any questions or comments to seaweedfactory@gmail.com