RAM/ROM:
Samples:
http://www.defeq.com/wp/2008/09/08/analog-waveform-sampling-project-by-3ot-and-opuswerk/
http://www.freesound.org/
Recording/Playback Using the R&P Machines:
Select an R1 Machine on track M1,
MLEV to 0,
ILEV to -64,
LEN to 64,
Rate to 127.
On M2 select your P1 machine. Go back to M1 and start the seq and select the trigs where you want to start recording.
As soon as the seq has gone through your seleted trigs undo them so they’re not lit up anymore.
Now go to M2 where you have your P1 track and selet the trigs where you want the previously recorded R1 traks to begin playing.
You should now be able to play the recorded material anywhere you place a trig in the M2 sequence. Also both traks must be unmuted so make sure that you have both M1 and M2 enable.
Uploading Samples: (care of/copyright drK)
Note: throughout this article we will refer to ROM machines 1-24 as “lower ROM” machines and ROM machines 25-48 as “upper ROM” machines.
OVERALL CONSIDERATIONS
One common complaint with the MD’s sample playback is that it is prone to producing seemingly random clicking noises in some situations. It is important to understand what can cause these clicks and how best to avoid them.
The ROM and RAM sample playback machines are all monophonic. They behave identically when a note is triggered while a previous note is still sounding: the previous note will be terminated immediately and the new note will begin playing essentially at the same time.
Cutting off a previously sounding note in this fashion will likely cause a pop or click. The severity of the pop depends on the amplitude of the note at the time it is cutoff and how near zero the sample is.
Samplers have traditionally avoided this issue by doing a rapid fade on the previous playing note before starting the new one. The MD does not seem to do this. This is not an omission as much as a trade-off on the part of the designers. The quick fade-out introduces a delay before the next note will sound. For a percussive/drum sampler that emphasizes “tight” timing this is not a good side effect.
The only sure-fire way to eliminate the pops is to ensure that the ROM machine is not currently playing a sample when a new note is triggered. DECAY and HOLD work together to control the duration of a note. In particular it will probably be necessary to vary these depending on the time duration to the next note.
To understand HOLD’s role it is best to think of it as a traditional “gate” control. The higher the hold setting the longer a note’s “gate time”. Lower ROM machines are designed to offer gate times up to one whole note. If you are programming a pattern using the 16th notes then a HOLD value of 4 provides a 50% gate time. The upper ROM machines double the time per HOLD value. A HOLD value of 2 provides 50% gate time for 16th notes.
While the note’s gate is active the ROM machine’s amplitude envelope remains at maximum. Once gate goes inactive DECAY sets the rate at which the amplitude envelope approaches zero.
A sample will play through to completion once it is triggered. Whether you hear it or not depends on the setting of HOLD and DECAY.
When the sample’s natural duration is longer than two adjacent notes you must make sure that HOLD and DECAY are set so that the note is not sounding when the next note happens. Use parameter locks to decrease the gate time (HOLD) and decay rate (DECAY) when notes fire close together.
The above guidelines hold true whether using percussive or looped samples.
PITCHED INSTRUMENT SAMPLES
The MD recognizes the first loop set inside a sample file. The END value is always set to the end of the loop. There is no way to have a sample loop while the gate is active and then play the remainder of the sample. Furthermore if you change END from its maximum value the sample will not loop! This has an important ramification: you cannot reverse play a looping sample. Reverse playback happens whenever START is greater than END. Since looping is turned off if END is not at its maximum then it is not possible to reverse play and loop at the same time.
START, however, does not affect looping. This can be used to great advantage with pitched instrument samples.
A common way to sample pitched instruments (or make pitched samples) is to sample the entire attack phase and then create a short loop that will play throughout the note’s sustain and release. The MD ROM machines support this way of using samples. Furthermore by changing START (including dynamically with LFOs or parameter locks) you can usually impart timbral changes. You can even move START within the loop and the sample will still loop properly. Remember “loop start” and START are different!
This utility of START means that using a lower ROM machine is preferred for pitched, looped samples. A lower ROM machine’s START setting provides a finer degree of control over the start by using most of the 128 adjustment values to cover the initial part of the sample. The finer control over the attack portion increases the likelihood of finding a suitable setting that provides the desired timbral change as well as working with minimal clicks. The disadvantage of lower ROM machines is that it is difficult to impossible to predict where in a sample each “tick” of START lands.
The MD ROM machines do not have any ready attack control (only by using an LFO can you impart a new attack segment on the amplitude envelope). Playback amplitude is determined by the original sample until the gate goes inactive. Rapid changes in a sample’s amplitude will cause clicks! If you use START to skip over part of attack portion then you run the risk of introducing clicks unless you land on a zero crossing.
[Upper ROM machines use a linear method for START. Each change of 1 value moves the start position 1/128th of the full sample length. This is true whether at the beginning, the middle, or the end]
PACKING SAMPLES AND MULTISAMPLING
One approach to giving the MD pseudo multisampling, and to overcome its limited number of sample slots, is to pack more than one sample into a single waveform file. This works effectively provided you take care to correctly construct the composite sample, and you use it with the proper ROM machine!
The method relies on the START setting to select one of a number of samples contained within a “master sample”. To make this practical:
- Always use upper ROM machines to play these “multisamples”.
- Place the individual samples on easy to find (and remember) locations within the overall sample length. We’ll look at an easy way, and a more comprehensive way, to achieve this below.
- Make sure to include a small region of silence between each sample. This will make it easier to adjust HOLD and DECAY without encroaching the next sample, and without needing to adjust END constantly.
This method will only work for non-looping samples. It is best for percussive samples, or samples that need to sustain for a defined duration.
Since END does not facilitate moving the loop point it is unclear if there is any advantage to adjusting it in concert with START so that only the selected sample segment can play. HOLD and DECAY can achieve the same result. The one possible use would be where you do want to insure that the entire segment plays without fiddling with HOLD and DECAY. If you use it this way still be mindful of not allowing a following note to cut a playing one!
SUGGESTED PROCEDURE FOR PACKING SAMPLES
Both methods use your sequencer/DAW as a way to assemble multiple samples in one file and achieve the precise placement necessary to select between them with the START setting (using an upper ROM machine). Basically you will assemble your samples one after the other, each aligned to start on a fixed rhythmic position like a quarter note. Once the samples are laid out in this fashion you render the “song” to a new audio file that becomes the sample you load into the MD.
Quick and easy method:
1. Decide how many samples you are going to pack. Round up to the closest power of two. For example if you need six samples then round up to eight. [Why? Because upper ROM machines offer 128 sample step positions (0-127) that span the entire sample. You want the total length (128 steps) to be evenly dividable by the number of samples. Hence the rounding or padding to the closest bigger power of two]
2. Add each sample to an audio track. Use the sequencer’s “snap to” facility to lock the start of each sample to a quarter note boundary. Place an empty region to pad the time after the last sample’s time slot until the entire duration is the number of quarter notes calculated in step 1.
3. Adjust the tempo so that no samples overlap and there is at least a small silence gap between all samples. Make sure that the total duration times the sample rate you use does not exceed the number of MD samples you wish to expend on this sample set!
4. If only a few samples are much longer than the rest you will want to have them occupy multiple time slots so that you do not waste sampling time. This is a very handy use of any padding you may have calculated in #1 and #2!
5. Double check the sample starts that they are perfectly aligned to the quarter note beat. It is not unusual to find “dead time” at the beginning of a sample, even one professionally prepared.
6. Use your DAW’s “render regions to audio” or similar command to create a new audio file that includes all the samples, including any padding. Remember the selected duration must be a even power of two beats like 8, 16, 32, etc.
7. After rendering the new audio file consider doing a sample rate conversion to save on sampling memory. The MD handles audio files with a wide range of sampling rates and still plays the sample with the proper pitch and timing. You can usually save sample memory by reducing the sampling rate below 44100 and still have excellent frequency response. Sampling rates around 30kHz will work well for a lot of sounds, and even as low as 20Khz can be used for certain lower frequency sounds with limited harmonics or upper frequencies (typically low pass filtered sounds)!
A more detailed way:
1. Add up the total duration of all the samples in the set. Next count the number of samples and subtract that from 129. For example if you have 10 samples then you will end up with 119. [This calculation is accounting for a gap between samples]
2. Take the number you calculated and multiple it by 15, then divide the result by the total duration of the samples. This is the BPM that your sample set will fit in. Set the song tempo to this value.
3. Place each sample in turn in the audio track aligned to a 16th note boundary (use “snap to” set to a 16th note). Leave a single 16th note gap between samples.
4. When you have placed all samples make sure that the last sample exactly ends on the start of the 9th bar (4/4). If it is grossly short or long then you probably made a calculation mistake in steps 1 and 2.
5. If you are short you can decrease tempo until it exactly fits. If you are over then you can increase tempo to make it fit but watch that the samples never overlap. You may need to speed it up and slide things around to make it fit. Best to get the calculation right in the first place!
This method maximize the use of storage space at the expense of both ease of creating the sample and using it. You should write down the 16th note location where each sample starts. This is your sample index.
Pitch control:
The ROM machines have pitch offset calibrated as three ticks of the pitch value per semitone. This lets you create melodies over 4 octaves (and some change).
Using RAM Machine Feedback as a soundsource:
Last night I realized that with the UW, you can use the cue-feedback as an instrument and sound source. This meens that you can make music using RAM Machines only.
No there’s nothing in the RAM before, and it’s retrigged and reset every time the patterns loops. This is why it sounds different all the time, try it:
1. Add a RAM Record machine.
2. Set Mlev to max.
3. Set an LFO to Cue1, Updte: Trig, Depth: Max, Speed: whatever you like, shape: I\_ (falling sawtooth).
4. Play with the filter to soften it down.
There you go.. a really weird and unstable ocillator, but cool.
It’s just about forcing the RAM Playback Machine to different things no matter what’s inside it. Fast pitch env to do kick/snare, hipass and short decay to have some kind of hihat sound, etc. Live was mainly about pressing mutes, changing the RAM Record Machine filter (which is also affected by a lfo) and sometimes function+pitch or Samplerate Reduction. Also if you want it less chaotic you could just set the rec length to something short like 6-7, then it’s actually quite stable. You can just release the monsters by turning that up later. Oh and of course, you can easily just freeze the sound by muting the rec channel since then the ram content just stays solid until you unmute it again.
Another thing that makes it more alive is that I had the rec length set for the full pattern. On top of that, the start setting is param-locked all over the sample. This makes everything change a bit but at the same time keep fragments from the last cycle... or something.
I just gave this a shot. Try these settings for starters (in addition to the one’s Kotten mentions) - certain settings will get you absolutely nothing:
LFO Speed - Keep it on the low side - say 40.
LFO Shape - Does seem to matter - try saw or tri.
LFO Shape Mix - Start out all the way counter-clockwise (0).
Synth Cue1 - Start out @ 0 - almost seems like a decay setting as you move up Synth ILev - W/ Cue1 low, sort of soften things up like a 3/6db LP filter.
Track effects - start w/ everything @ default.
Now start pointing some of those extra LFOs towards the track effects.
Kind of sounds like a tri/saw kind of mix before filtering/amplitude mod/etc. For some reason, I was expecting a Sin. Well, I guess it is a Sin w/ some clipping/brr/srr going on.
Can it hurt anything? Start out w/ the track level low so you protect your speakers but aside from that I doubt it - it’s probably the noise floor forced into clipping.
— tIB 2009/01/12 18:25
Login Form
