All this talking about sound programming/filters/oscillators/etc..., I just had this simple question (at least seems simple)

I saw something in another thread by Bad_Mister about being able to load like 4000+ waveforms/samples into a fully loaded XF… so I began to think…

When I have a sample of a piano I have to load like 3-5 notes from each octave and then set the multisample ranges so the 3-5 notes end up covering the entire octave, etc…

So I end up using quite a bit of sample space because of all those different waves (the 3-5 notes per octave) that I have to load.

Ok…

When I go into the menu’s on my workstations so I can program a sound the base of the sound is 1 simple wave form.  I select that waveform and it spreads it across all the keys…

what is different about my one waveform/sample and the one waveform/sample that comes in the ROM of the keyboard/workstation?, why do I have to load and spread those samples across the keys, but then you don’t have to with the wave ROM?

Here’s my idea/theory (I’m just not sure if it’s correct though)

-behind each of those waves that are loaded in the ROM are there “mulit-samples” that make up that one little file?  If so how do they keep the waverom size SO small if all those samples are used to make up the preinstalled waveROM?  (is that where all this compression and stuff comes from that everyone is talking about?)

Thanks

I’m posting this on the Roland/Korg forums too just to get to hear multiple views (we all know Roland, Korg, and Yamaha people all think different!!!!  just kidding)

Behind each of the waveforms in any workstation is hours and hours of work by professional sound designers.  So those people have already done the hardest work for you.

The reason that the wave memory is small is because the sound designer has already done the toughest job which is select and loop the samples.

We have been working a lot with VST sampled instruments.  Because typically the computer can stream samples from the computer’s hard disk, no one pays attention to trimming samples or trying to reduce sample space.

Take a look at any 2 GB sample library for a computer platform and you will find in many cases that. 1 GB of what you pay for is silence at the end of samples. People can be very wasteful with sample space because

1:It doesn’t matter for the computer system
2: The bigger the file appears. ( even if it is silence at the end of the sample) , the more apparent value it has.  A 2GB library is better than a 1 GB library , right ? No, not necessarily. It depends on the capability of the sound designer.

Yamaha_US - 08 September 2010 02:36 AM

Behind each of the waveforms in any workstation is hours and hours of work by professional sound designers.  So those people have already done the hardest work for you.

The reason that the wave memory is small is because the sound designer has already done the toughest job which is select and loop the samples.

We have been working a lot with VST sampled instruments.  Because typically the computer can stream samples from the computer’s hard disk, no one pays attention to trimming samples or trying to reduce sample space.

Take a look at any 2 GB sample library for a computer platform and you will find in many cases that. 1 GB of what you pay for is silence at the end of samples. People can be very wasteful with sample space because

1:It doesn’t matter for the computer system
2: The bigger the file appears. ( even if it is silence at the end of the sample) , the more apparent value it has.  A 2GB library is better than a 1 GB library , right ? No, not necessarily. It depends on the capability of the sound designer.

so for us novice sound designers, we should get better and better at how we use/edit/apply the wavesamples that we want to use… in order to make better use of the sounds?  looking the piano samples so instead of using (for example) a piano sample that last for 6 sec… we can use the same sample, but just cut it so it only last for 2-3 seconds and then just loop the sample...etc so it takes less space?  I saw a video about using the filters to give the effect of a decay in sound (like an acoustic piano), but it’s really just a filter.  so are these the types of things that you are talking about?

so where does compression come into play?  I’ve heard people say that the workstation manufacturers are compressing the crap out of the samples and that’s how they get so much sound out of a “small” sample/wave ROM…

also during the countless hours of work that is put into the programming do they do something similar to what we have to do as far as the 3-5 notes per octave?  and all that gets turned into the 1 waverom sample that we use as the basis of our sounds?

You’ve “sort of” got the idea.

Here’s the structure - a waveform is a collection of single samples. A multi-sample is simply a “layer” of individual samples that are all assigned to velocity ranges. So a multi-sample in a piano waveform can
consist of multiple samples of the same note, each recorded at different
velocities. Each of the samples has a velocity range (for example, the soft sample is restricted to the velocity range of 1 - 70, the medium one
is restricted to the range of 71 to 108, the medium hard is in the range of 109 to 119, and the super hard is in the range of 120-127). The
note range of each multi-sample can also be set so that
it covers a particular area of the keyboard, as you have learned.

The difference between what you are doing and what Yamaha has done is
that the samples have been looped. The initial attack portion of a recorded note is kept, then at some point in the rest
of the sound, a section of the sound is set so it plays over and over
again - loops, in other words. Any part of the wave after the end point of the loop is thrown away. Getting a smooth loop in a waveform is not easy to do, and requires special software plus experience.
Keep in mind, every single sample for an instrument has to be looped (unless it’s a “one-shot” instrument such as a drum or percussion piece).

After the looping is done, then the data undergoes special compression,
similar to what happens when you take an audio track off a CD and turn it
into an mp3, which reduces the size. Yamaha has specialized compression
software that took them years to develop. Once the data is compressed and
encoded, it is then ported to the ROM. The XF has over 700MB of ROM waveforms onboard, but keep in mind, the original recordings from which
this ROM was made amounted to terabytes in size, stored on multiple huge drive arrays.

At this point you are not doing any looping of samples or data compression yourself. When you edit an existing sound on the Motif, “shortening” it (changing the decay of the amplitude EG), you are not
“shrinking” the waveform. You are changing an amplitude envelope, but
not editing the waveform itself. The waveform is simply the “color”
from the paintbox of sounds you have onboard...a voice is a “picture”
made from the colors.

If you want to get better at being a sound designer, focus on editing
the existing voices...experiment with changing the filter and amp EG’s,
the filter cutoff, resonance, filter type, LFO’s and effects. You are
thinking that you have to get better at making a better “purple” for
your sound picture, when what you have to get better at is making the
picture that uses the “purple”.

ok. so if someone downloads a soundbank to store in their flash memory and it says “this soundbank contains 200 user waveforms” are those 200 “different waveforms” (for different instruments (piano, string, flute, trumpet, bass guitar) or is it possible that 35-50 (just throwing out a number) of those waveforms are just different notes/pitches/octaves/velocities of the piano sample? and then 20 of them are from the trumpet, 30 from the bass guitar, etc....

OR is this gonna be 200 different waveforms like the different wave forms that we find in our edit screen when we make a patch from scratch…

Apex - 08 September 2010 07:52 PM

ok. so if someone downloads a soundbank to store in their flash memory and it says “this soundbank contains 200 user waveforms” are those 200 “different waveforms” (for different instruments (piano, string, flute, trumpet, bass guitar) or is it possible that 35-50 (just throwing out a number) of those waveforms are just different notes/pitches/octaves/velocities of the piano sample? and then 20 of them are from the trumpet, 30 from the bass guitar, etc....

OR is this gonna be 200 different waveforms like the different wave forms that we find in our edit screen when we make a patch from scratch…

A waveform is a collection of samples.  A waveform therefore typically represents a particular instrument or sound type (like the waveforms found in the edit screen).  A soundbank contains voices which reference waveforms.  Multiple different voices can be created which contain elements which reference the same waveforms.

Voices are stored in user memory banks—which are not located in the expansion flash memory.  Loading a soundbank containing a voice that uses new waveforms causes the waveforms to be loaded to sample memory or flash memory (by user preference) so that the newly loaded voice can access the new waveforms.

Apex - 08 September 2010 07:52 PM

ok. so if it says “this soundbank contains 200 user waveforms” are those 200 “different waveforms” is it possible that 35-50 (just throwing out a number) of those waveforms are just different notes/pitches/octaves/velocities of the piano sample? and then 20 of them are from the trumpet, 30 from the bass guitar, etc....

OR is this gonna be 200 different waveforms like the different wave forms that we find in our edit screen when we make a patch from scratch…

In the world of Yamaha and the Motif, it’s the latter...a waveform is a group of samples so 200 user waveforms means 200 different sounds.  The number of individual samples is much, much larger.

When you see sample CDs advertising thousands of samples it often means the total number of samples, not how they get used.  If they duplicate the samples in both WAV and AIFF, both 16 and 24 bit, then multiple samples per octave over multiple octaves…

Here’s two examples from sample CDs I have:

“2013 samples for 80s synthpop”. This turned out to be a total of only 25 playable instrument sounds plus an additional 1000+ loops (mainly drums, some riffs and sound FX). Do the math: take the other 1000 samples, divide by 25...that’s 40 samples per sound. Assuming 4 samples per octave and both WAV and AIFF per sound, to cover 61 keys you’d need (surprise) forty samples. So what they called 1000 samples Yamaha would call 25 waveforms.

“1GB of classic Waldorf sounds from the Blofeld”. 1GB of samples is more than the entire XF waveform ROM!!! That translated into exactly 12 different instrument sounds (no drum loops this time).

Which corroborates the earlier post warning that size of the sample set doesn’t tell you much at all about what you’re getting.

meatballfulton - 08 September 2010 10:31 PM

Apex - 08 September 2010 07:52 PM

ok. so if it says “this soundbank contains 200 user waveforms” are those 200 “different waveforms” is it possible that 35-50 (just throwing out a number) of those waveforms are just different notes/pitches/octaves/velocities of the piano sample? and then 20 of them are from the trumpet, 30 from the bass guitar, etc....

OR is this gonna be 200 different waveforms like the different wave forms that we find in our edit screen when we make a patch from scratch…

In the world of Yamaha and the Motif, it’s the latter...a waveform is a group of samples so 200 user waveforms means 200 different sounds.  The number of individual samples is much, much larger.

When you see sample CDs advertising thousands of samples it often means the total number of samples, not how they get used.  If they duplicate the samples in both WAV and AIFF, both 16 and 24 bit, then multiple samples per octave over multiple octaves…

Here’s two examples from sample CDs I have:

“2013 samples for 80s synthpop”. This turned out to be a total of only 25 playable instrument sounds plus an additional 1000+ loops (mainly drums, some riffs and sound FX). Do the math: take the other 1000 samples, divide by 25...that’s 40 samples per sound. Assuming 4 samples per octave and both WAV and AIFF per sound, to cover 61 keys you’d need (surprise) forty samples. So what they called 1000 samples Yamaha would call 25 waveforms.

“1GB of classic Waldorf sounds from the Blofeld”. 1GB of samples is more than the entire XF waveform ROM!!! That translated into exactly 12 different instrument sounds (no drum loops this time).

Which corroborates the earlier post warning that size of the sample set doesn’t tell you much at all about what you’re getting.

So which one are we loading into flash memory?  The single instruments or the many waveforms that make up the single sounds?

Do we have the ability to achieve what they do with just one single compressed waveform and make an entire keyboard full of that sound using just that one waveform...or do we have to load ALL those thousands of waveforms that make up the sound.

I hope I’m being clear… and I do have a goal I’m trying to achieve .

Apex - 08 September 2010 11:23 PM

So which one are we loading into flash memory?  The single instruments or the many waveforms that make up the single intruments

I hope I’m being clear… and I do have a goal I’m trying to achieve .

Both get loaded into flash, the many waves/samples and the waveform information which is how those samples are mapped across the keyboard to make up a complete multi sampled instrument.

Perhaps if tell us the goal , we get suggest how to get to it.