Loading WAV samples into OASYS & assigning to multisampl

[Kevin Nolan]

Hi,

I will shortlytry for the first time to build a multisample and program from a set of samples loaded into the OASYS as wav files. The samples will be already clean samples with a few seconds of the wave amplitude flat for looping well. I think the only required editing of each sample will be in finding the exact start point, to normalize the samples and to loop them.

While I've read the operation and parameter guide for Sample mode and Disk mode "sample loading", I'm not 100% clear on how the process goes. In particular, I cannot figure out how to get the samples into the sampler for editing, as it looks like the sampler only works on an already named multisample, and my samples are not in a multi sample yet. I suspect that I have to go into Sampler Recording mode and make a multisample even though I'm not actually recording, but am not quite sure.

Can you confirm/deny the following steps and add any missing steps?

- In disk mode, copy the samples into a named folder on the OASYS HD.
- In disk mode, select the wav files and do a multi load into RAM
- In Sample mode - how does the sampler see the samples? is it:
--In Sampler Recording Mode, make a new multisample
----Create the Multisample
----Name the Multisample
----In each Index, select the appropriate sample and set the other parameters
--In Sample Edit mode, Edit the range
--In Sample Edit mode, Normalize each sample
--In Sampler Loop Edit Mode, set the start point, Loop Start and End points
--In Disk mode, choose Save Sample Data and choose One Multisample
--/* The multisample is now created */
--Create a new program and assign the multisample and other parameters
-- Save the Program

I'd be grateful for any tips as I have never done this before. I'd also like to acknowledge the great help Sharp has provided to me so far by PM, but I felt at this stage it was worth posting to the forum as the answers to the questions above may be valuable to others.

For your interest, I'm using a freely available set of samples of the famed Roland VP-330 Strings Machine:

http://s3.amazonaws.com/midiwall/gear/vp330/index.html

which was an iconic instrument used in the late seventies and early eighties by many, including Vangelis fo the string sounds on Bladerunner, as well as stunning analogue human choral sounds and vocoder on his album 'See You Later' (an absolutely incredible but not well known electronic/jazz album from which the incredible blusey piano piece Memories of Green (used in Bladerunner) came from).

I aim to make five multisamples and five programs based on the five voices selectable on the VP-330 and then combine them in a Combi, along with a Vocoder program, and careful ensemble and other effectes to see if I can create an OASYS based VP-330. As said I'm new to this and time is usually tight so it'll be a while before this is ready, but will be deligthed to offer it free when complete.

But the first step is to get the multisamples right - so any and all help will be greatly appreciated.

Thanks,
Kevin.

[Daz]

That sounds about right

I am not at my O right now so I can't be specific, but the only thing I would add to what you've said is that when you create the initial empty multi-sample you'll find it already has 1 index and it's tricky changing it's settings (keyrange), so I generally just delete that default index and then start creating my own.

Daz.

[Mike Conway]

Kevin,

in DISK mode, select whatever folder that you want and pick a .wav file. Press LOAD. Do this as many times as you want.

* Press the SAMPLE button.

* Press the MULTISAMPLE tab.

This is where you should see a default multisample, with 8 Indexes (ranges). What I do is select Index 8 and CUT it. I repeat CUT, until I only have one Index.

Now, press that little sample arrow tab and you will see the .wav files that you loaded. These can be complete songs, individual sounds, etc. Select a wave. Adjust your low/high note range. Press INSERT and add another Index. Repeat as necessary.

This is the way I do it, but Sharp is the real expert and may chime in with a better way.

I think you have the DVD, right? Review the short demo, "Samples in a Program". Though only a single sample is used for the full range, it's the same mode of operation - truncate, loop, normalize, etc. Just press INSERT to add Index ranges.

My biggest initial confusion was seeing the 8 Indexes, because you can't set the high note of Index 1 above the low note of Index 2. It was driving me batty, just to set the range of the first sample, so I used CUT and INSERT. Working with one at a time made more sense to me.

[Arend Groot]

My biggest initial confusion was seeing the 8 Indexes, because you can't set the high note of Index 1 above the low note of Index 2. It was driving me batty, just to set the range of the first sample, so I used CUT and INSERT.

Good to hear that I am not the only one...it took me some time to get it. But now I know how it works I think its a easy way to assign samples in a multisample.

[Sharp]

My biggest initial confusion was seeing the 8 Indexes, because you can't set the high note of Index 1 above the low note of Index 2. It was driving me batty, just to set the range of the first sample, so I used CUT and INSERT.

Good to hear that I am not the only one...it took me some time to get it. But now I know how it works I think its a easy way to assign samples in a multisample.

If you set the template on the Multisample page, then you don't need to insert, cut, paste, worry about original notes, bottom or high ranges per sample assigned to the multisample.

If the tempate is set, you do nothing other than select the first sample and touch create on the screen. Then select the second sample, press create, third sample, press create, forth sample, press create...... and so on until you run out of samples. That it, multisample created.

You could assemble an entire multisample in sections using this method.

Regards.
Sharp.

[Kevin Nolan]

Dear Colleagues -

Thanks a million for your responses. I'll definitely take them on board. Mike - you've found me out - I had not looked at your DVD even though I had bought it some months back!! But I have now and it's absolutely excellent. Though I've lots of synthesizers (and experience), I tend to be really slow at getting into all the details of new equipment - and you have cut through all of that like a hot knife through butter.

The bass effect on the French horns is a fantastic section, as is sampling and vocoding. I had not used the OASYS Vocoder but its fantastic - especially with choir sounds - phew!

I know we've all said it time and time again, but still to this day I cannot 'take it all in' on the OASYS. Its a bottomless well of audio and musical capabilities. Despite some of my complaints of late over MOD-7 presets, in recent times I've been playing around with MOD-7 and find it brilliant; then there were Stuarts presets from a while back which only recently lead me into wavesequencing and what a phenomenon that it, and now with exposure to the OASYS's formidable sampling - where does it end. But your DVD, Mike, is just brilliant - its extremely straightforward but also very interesting because its not like a class - you are bringing you personal experience to it also which makes it extremely engaging. Splitting up the sample of your voice and applying triggering and effects variations exposes just how immediate and powerful sampling on the OASYS is. the mind boggles as to what would be possible by also incorporating wave sequencing, vector synthesis and KARMA GE's into ones own sampling! There just aren't enough hours in a human lifespan to enjoy the OASYS fully!!

Anyway, I know I rabbit on, but just felt like expressing again how incredible the OASYS is having recently immersed myself in some more of its endless list of feature for the first time.

Looking forward to some sample editing and multisample and program creating in the coming weeks; and as said, if anything useful comes from my attempt to recreated the VP-330, I'll be sure to share it here.

Cheers,
Kevin.

[Mike Conway]

I had not looked at your DVD even though I had bought it some months back!! But I have now and it's absolutely excellent.

That's what happened with Elvis - took months to get to it. Better late than never. Good to hear that it hit the spot! Thanks, Kevin.

[Kevin Nolan]

Sharp (and anyone who has thoughts on this) -

In a PM to me you indicated that when I treat the samples I'm using, I should make the start point the actual beginning of the provided samples and then decide my loop points appropriately.

As you indicated, these are quite good samples, being very flat once the Attack and Decay settle down to the flat Sustain level. However, if I make the start point the beginning of the samples, does that mean that when I eventually make a multisample from these, that the Attack time can never be faster than the Attack time in the provided samples? While the supplies Attack time is fast, it could be faster.

Hence - would it be better to crop the samples to about 2-3 seconds at the pure flat level part and remove the front 'Attack' part and any unwanted end material (given that these are so flat that loop points will not be difficult to choose). This would reduce the size of each sample and conserve memory, and allow for the OASYS envelopes to work to their maximum.

Finally - in normalizing the samples - should I normalize them to 100% in an external package (I'm using Adobe Audition for now), or is it better to let the OASYS do this?

Cheers,
Kevin.

[Sharp]

Hi Kevin

However, if I make the start point the beginning of the samples, does that mean that when I eventually make a multisample from these, that the Attack time can never be faster than the Attack time in the provided samples? While the supplies Attack time is fast, it could be faster.

For the type of sound it is, you could very well cut the attack completely out, but consider yourself very lucky here. You picked a pretty unique set of samples here.

Generally speaking you will never be able to do this for the majority of samples. This is why there are options like “2nd offset”.

In Program mode you can enable one of the 2nd offset features to force the OASYS to start further into the sample than the start loop point was set. This will in turn allow you to place a faster attack on samples that have a slow attack naturally.

So really, I wouldn't bother even cutting out the attack from the samples your working with, even though they are the perfect selection for getting away with doing this. It's better to have the best of both worlds, the natural attack, and the 2nd offset features.

Finally - in normalizing the samples - should I normalize them to 100% in an external package (I'm using Adobe Audition for now), or is it better to let the OASYS do this?

Best to use Adobe Audition. It's just faster. No other reason. 100% would be correct too.

Regards.
Sharp.

[Kevin Nolan]

Thanks a million Sharp - that's fantastic - really appreciate your expertise on this.

Cheers,
Kevin.


(If you're even in Dublin - I owe you a real pint or two !!)

[sirCombatWombat]

In Program mode you can enable one of the 2nd offset features to force the OASYS to start further into the sample than the start loop point was set. This will in turn allow you to place a faster attack on samples that have a slow attack naturally.

Does anybody know why this offset parameter is made like this? It would seem to me that it would be better to have a parameter that adjusts the offset in milliseconds or samples, rather than 1st - 8th preset starting points that can't be edited?
Or am I mistaken, is there a workaround where the start offset coud be for example tied to a modulation source?

[Sharp]

Hi sirCombatWombat .

Can't say I never found a need to have the ability to enter in a specific value into the offset, but I can see where you coming from, I'm sure that would be handy.

As far as I know, you can't tie it into any sort of modulation either. It is what it is.

Regards.
Sharp.

[Sharp]

Thanks a million Sharp - that's fantastic - really appreciate your expertise on this.

Cheers,
Kevin.


(If you're even in Dublin - I owe you a real pint or two !!)

No worries mate. Happy I could help.
Could take you up on that offer of a beer too somtime.

Regards.
Sharp.

[Kevin Nolan]

Hi Sharp (and other Sampling experts) -

I have begun to prepare individual note samples from the VP330 to bring into the OASYS. Currently, I have performed the following on 4 sampled notes (C1, Eb1, G1, C2) within Adobe Audition:

- Save 2 seconds of each sample to its own wav file.
- Convert the stereo sample to mono and from 48kHz to 44.kHz (at highest quality)
- Match the beginning and end of the sample to the beginning and end of the periodic wave (sample accurate)
- Normalize to 100%

The results of this process can be found for download as four wav files found at the end of the web page:

http://www.knect.ie/OASYS.html

However, having performed the same procedure on each sample, I have noticed that, even with 100% normalization, not all samples are at the same amplitude, nor are they equally symmetric on 0dB.

To have all at exactly the same amplitude and DC offset, I'd have to do this manually in Adobe Audition. While I have no problem doing this, I would appreciate your thoughts and opinion on:

1. The persisting difference in amplitude between each note even after normalizing in Adobe Audition seems to reflect the actual amplitude differences of the notes on the original recording (see top post on this thread). This suggests to me that the VP-330 retains this amplitude relationship between notes, and hence I wonder should I leave it as is to retain that authenticity?
2. The same is true of the DC offsets - the persisting DC offsets reflect the obvious offsets seen in the original wav file for each note; and again I wonder should I leave them as is?

However - since I only have limited sampled notes (4 per octave) and will have to spread them across the keyboard, will I hear obvious and unacceptable mismatches in amplitude and/or tone because of the amplitude and DC differences? While I realise this will never be perfect; should I (and is it normal) to manually DC shift and Normalize each sample so that all have identical amplitudes and symmetry?

Or – if I subsequently normalize within the OASYS, will it automatically sort this out?

I'd be grateful for your thoughts on this before I proceed any further - I don’t want to go through all 48 samples across the various male and female choral sounds and string sounds of the VP-330 only to find I did it wrong!

Thanks,
Kevin.

[Sharp]

- Save 2 seconds of each sample to its own wav file.

Sweet, that's a nice length considering the samples in question.

- Convert the stereo sample to mono and from 48kHz to 44.kHz (at highest quality)

There was no need to do that. The OASYS run at 48Khz.

- Match the beginning and end of the sample to the beginning and end of the periodic wave (sample accurate)

You don't need to worry about the end of the sample as you will be cropping everything after the End Loop point.

- Normalize to 100%

Correct, but Adobe would have normalised the sound based on the peak of the attack in the wave file. So consistency will be off here. Normally for sampling you use a process that levels the waveform. If you take a look at my old review of Seamless Looper you will see me talking about “Level” and flattening the waveform.

http://www.korgforums.com/forum/phpBB2/ ... hp?t=17039

To have all at exactly the same amplitude and DC offset, I'd have to do this manually in Adobe Audition. While I have no problem doing this, I would appreciate your thoughts and opinion on:

If you don't have the means to “Level” the sample as I did in the link above, you can simply just load the samples into the OASYS normalised and use the volume settings in the multisample to adjust the volume of each sample assigned to the multisample.

Regards
Sharp.