E-MU EMULATOR X2 (software sampler)

[Anderton]

My apologies. I've been in video edit mode after AES, there are 30 videos waiting to go live any day now. But I'm almost done and can get back to the X2 shortly. I even have screen shots of the sampling/editing process all ready to go...I just need a couple hours to prep them and write the text.

Then it's time to twist some loops...heh heh......

Hey, I said this was going to be a long one!

[dahkter]

Hi Craig,
Can you clarify a few basics for me regarding workflow?
1. I understand you can record right into it, is this only as a stand alone, or can it also do this when it's launched in a host (cubase/live/tracktion)
2. Once the sample is in there, how easy is it to quickly divide it into different regions, whether it's taking a fifteen minute clip, and dividing it into thirty slices, or a two second clip, dividing into ten slices. I know you're a fan of wavelab, curious if the interface is similar, easy to zoom in and out, left and right.
3. Once the samples have been tweaked (with x2's effects, filters, slices), how easy is it to export the sounds out of it? It would be nice to take the tweaked samples out of it, and load the wavs into my MPC.
4. Any chance of an EMu demo, or are there any stores in the NYC area which have the software on display so I can try before buy?

Thanks in advance for your help.

[Anderton]

>

Stand-alone only for recording or SynthSwipe. You can load or import in any mode.

>

Not quite sure I get exactly what you're driving at, but in terms of SynthSwipe, everything ends up just where you want it in terms of splits etc. Or do you mean something like recording dialog and chopping it into phrases?

>

Yes and no. The biggest limitation is that you can't resize the window -- another victim of being tied to bit-mapped graphics. Within that constraint, though, the most important factor in getting around the UI is to learn the various shortcuts for zoom in, zoom out, etc. The really helpful ones are things like go to selection end, go to selection beginning, zoom to selection, go to loop end, go to loop start, and the like. These are accessible via toolbar buttons, but you really want to take the time to learn the keyboard equivalents.

However, there is an "open in external editor" button. Say you have a waveform open in X2. You can click on the button to open it in Wavelab or Sound Forge or whatever (and stretch it across two monitors, of course!!), and if you make changes in the editor (including changing loop points or whatever), then save the file, when you go back to the X2 the waveform will reflect these changes.

[Anderton]

>

As far as I know, there's no easy way to do this. My assumption is that this is because you have the raw files and the processing, and there's no way to "render" or "apply" the effects to the raw waveform. There are no export options other than E-Mu centric ones.

The workaround for this would be to load the X2 as a plug-in within a host, and play a note to trigger the file. Bounce it to a track, then export the audio from the track.

Does anyone from E-Mu have a more elegant way to do this?



Someone from E-Mu would need to answer this. I don't know of any downloadable demos.

[dahkter]

Thanks Craig,
Much apprecaited.

[Beatsmyth20]

Do you know how easy it would be to swipe my sound module and then convert those pathces into reason refills. Reason has a sampler called the NN-XT that allows you to map wav files across the keys (whether it transposes one wave file or you map individual waves for each note to each of the keys) and you can also layer samples so that a different sample will be played if the same key is pressed at different velocities. Does the X2 create seperate wav files for each note in a patch that can be loaded into an external sampler?

Thanks for your help.

[Anderton]

See my reply two posts above regarding the question of transferring samples to an MPC. The X2 doesn't seem to have any way to export standard wav files.

However, you CAN open the samples in an editor and save as WAV or AIFF or whatever you want. But of course, these are just raw files with loop points; they won't be subject to the filtering and such that the X2 offers.

So the bottom line is this is how you'd do it:

1. Swipe the synth.
2. Open each sample in an exteranl editor and save as WAV.
3. Load in the NN-XT and apply your various splits, layers, processing, etc.

[PubeDread]

Originally posted by Anderton The X2 doesn't seem to have any way to export standard wav files.

 

 

To export a sample, R-Click in on a sample in the left pane. Select Export.

[ICHi]

Originally posted by PubeDread To export a sample, R-Click in on a sample in the left pane. Select Export.

 

***Another option you have for any output of [ASAP] Auto-Sampling And Placement is to choose the "to files" option in the save section. This is the alternative to sending the chopped files back "to sampler" for placement. Some cool ways to use this...

-Strip silence from Audio Sampling CD tracks and generate seperated/named files rapidly.

-Take stuff you recorded in your DAW or Acapella tracks and split so you can load into any sampler for more flexibilty when doing remixes.

-Chopping up long contiguous dialog or sound effects recordings.

 

Craig, thanks very much for such a thorough review. We really appreciate your support. Will be checking in from time to time and will try and answer any questions directed our way.

 

Best Regards,

ICHi

E-MU Systems

[Anderton]

>

That is brilliant!! Thanks for chiming in, I know the X2 is very, very deep and I can't catch everything. Any hints, tips, and comments will be most appreciated.

Also, thanks to PubeDread for the export tip. Much easier than opening in a waveform editor and saving.

Now, if we could just have a "render with processing and save as WAV" file...the X2 is such a sound design monster, a command like this would make it very easy to create sound libraries.

[Charm]

Craig

I'm wondering about upgrading from Emulator X to X2, but just for one specific reason: I want a VST instrument with the sound and response of an Emu Proformance/1 piano module, and I wonder whether SynthSwipe could help.

Together, Chicken Systems 'Free' and Emulator-X's sample-import features have given my Emulator-X access to my eight old Emu EIIIx CD-ROMs: Vol 1 (Emulator Standards) to Vol 8 (Vintage). Using Bank 5 in Volume 7 (Emu Classics) I've made faithful reproductions of old sequences that used to play the Proteus/2. (I played back a sequence simultaneously on a Proteus/2 and on Emulator-X (equivalent instruments), and made "A/B" comparisons whilst adjusting levels in Emulator-X.)

So, it seems my Emulator-X can stand in for my EIIIXP and my Proteus/2. But it can't (yet, anyway) simulate my Proformance/1. No piano sound on these Emu CD-ROMs is a good match with the Proformance/1. The nearest match is one of the Bosendorfers - very similar in part of the keyboard range. I don't know whether a closer match would be possible by cobbling together the close-matching samples from that Bosendorfer with samples from one of the others, but I suspect not.

SynthSwipe to the rescue?

Or not?

Ought it to be feasible to use it to get Emulator-X2 to simulate a Proformance/1? What would I do - break the keyboard range into sections and sample each section, setting a note length covering the full decay of the longest note in the range? Then set envelope release times to deal with playing shorter notes? And how fill in the gaps between the different velocity levels? - Or does the software deal with that?

Is the record level the same throughout the automatic sampling? - If so, I suppose the low-velocity samples woud be be inherrently noisier?

Would it be very time-consuming to construct a usable fake Proformance/1 this way?

Of course, the best solution would be to persuate Emu to sell a "real imitation" ("real" because it used the original samples, etc) of the Proformance/1 - whether for loading into the Emulator or as a purely-Proformance/1 VST instrument.

[ICHi]

Ought it to be feasible to use it to get Emulator-X2 to simulate a Proformance/1? What would I do - break the keyboard range into sections and sample each section, setting a note length covering the full decay of the longest note in the range?
***There are 2 approaches.
1) Sample the raw waveforms(by undoing all the programming), reloop and reprogram with envelopes etc.
2) Sample the synth as is and just make it a streaming copy.

Then set envelope release times to deal with playing shorter notes?
***Adding ENV's to a streaming copy is kind of redundant since you've already recorded it's dynamic response. For the most control, use approach 1. It's more work but you'll be able to manipulate the sample set pretty much as you wish.

And how fill in the gaps between the different velocity levels?
***It will setup velocity layer switches. You can tune these of course after the fact.

Or does the software deal with that?
***You will need to experiment a few times but SynthSwipe will sample all the levels you want.

Is the record level the same throughout the automatic sampling? - If so, I suppose the low-velocity samples woud be be inherrently noisier?
***Depending on the amount of layers you choose the MIDI device will be triggered and recorded at x velocity and thus the volume would be relative to the velocity. You can choose to normalize if you want, and yes the quiet samples will have more noise if you did this.

Would it be very time-consuming to construct a usable fake Proformance/1 this way?
***Approach 1 is definitely more work than 2 but both can deliver very good results. Approach 1 is very flexible and is probably very small and resource efficient, 2 is brute force so it will be a very big sample set.

Of course, the best solution would be to persuate Emu to sell a "real imitation" ("real" because it used the original samples, etc) of the Proformance/1 - whether for loading into the Emulator or as a purely-Proformance/1 VST instrument.
***The Stereo Grand in the Proteus Composer EXB(AKA PROTEUS 2000) is pretty much the Proformance piano. It's been tweaked, but we think for the better. This is probably the easiest way to start if you are trying to emulate the Proformance/1.

Best,
ICHi
E-MU Systems

[Anderton]

"Approach 1" is what I'm doing with my Peavey DPM3 samples. It is more work, but far more flexible. The filters in the X2 are close enough to the Peavey I can get the sounds I need; what makes the DPM3 distinctive is how it transposes sounds, and the X2 captures that essence as part of the waveform.

[dahkter]

Hi Ichi,

per craig's request:

"Now, if we could just have a "render with processing and save as WAV" file"

Is this possible yet? If I run a sample through the x2's effects, whether it be the slicer or the sp12ulator, can I export through the x2, or do I first need to render as a wav in the host?

Thanks

[Rique]

Hi Craig, I’m glad your doing such a thorough review of the X2.

I’m a Mac guy and I’m considering getting a PC just for the X2 or will
an dual core intel Mac be just as good?

Looking forward to your Twistaloops and Transform multiply review.

ICHi, since this sampler is so complex, any chance of a DVD being made
showing creative uses and EXAMPLES?

Thanks to you both.

[Anderton]

 

 

I don't have an Intel Mac, just a dual G5...so I can't really test this. I can't imagine someone from E-Mu not testing this, however...especially because they have a USB interface that the Intel Macs should get along with.

 

E-Mu is monitoring this thread, I think it's very likely you'll get an answer.

[Anderton]

Before moving on to additional features like TwistaLoop, I’d like to wrap up the sample/program creation end of things.

As you may recall, I’d been using SynthSwipe to grab a bunch of waveforms from my DPM3. At this point, I’ve “swiped” 11 complete multi-sampled sets of waves. When I did, I knew I’d be doing pretty much single-cycle loops, so I “swiped” with the loop points enclosing the last eighth of the sample.

When it’s time to actually edit samples, you look under the Samples folder in the browser tree to find the sample(s) you want to edit. I would find it more convenient if you could open up a Preset to expose the samples from there; I really can’t think of any downside to doing things this way, unless there’s some kind of problem that could occur if multiple programs point to the same samples, and you can’t include some kind of “alias” to the point to the sample you want.

Click on the attachment to see what a raw sample looks like.. On the left, in the sample browser, you can see a series of waves labeled “DPM DGW5 [key name].” There are 13 samples for this multi-sampled instrument. On the right, you can see the waveform; I sampled one second’s worth of audio at 44.1kHz, so you can see there are 44,100 samples, with loops around the last eighth of the sample. Let’s get to work and trim this into something that a) doesn’t take up as much memory, and b) has seamless looping.

[Anderton]

The first thing I did was move the loop points arbitrarily to around 3,500 and 5,000 samples. I figured that would be more than enough to cover a waveform, but would cut the memory requirements by about 90%.

Then, I hit the Autocorrelate button. This analyzes the loop, and moves the start point to a portion of the wave that most closely resembles the end point. Sometimes it’s a direct hit, sometimes a miss. If it misses, you move the end point and try again. It often seemed to be perfect except for being off by one sample. In this case, there’s a numerical field for the loop start, end, and length that you can “nudge” one sample at a time (or type in a value, if you’re so inclined). Most of the time, just nudging the start one sample later produced a perfect loop.

Click on the attachment to see the loop points selected by auto-correlation. It’s pretty impressive; this was one of those “hit the bulls-eye immediately” acts of correlation. In this screen shot I also clicked on the “Select Loop” view, which highlights the loop and sizes it to take up most of the waveform view.

The final step (other than saving, of course!) was selecting the waveform just a little past the end of loop, and selecting Truncate to cut off the excess amount of signal. Done – one sample trimmed and ready to go.

[Anderton]

That left me with 12 more samples to do. Time for an experiment: How fast could I loop and trim 12 samples? So, right now it’s 9:38 PM…let’s see what happens.

9:45 PM…12 samples looped, trimmed, and saved: Not bad at all, and that’s taking my time and making sure the loops were good before truncating. The higher the note frequency, the more likely that I had to nudge the sample start one sample later to make it all work out okay.

Of course, these are just the raw multisamples; I haven’t added any filtering, LFO, effects, etc. That’s a whole other thing.

(By the way, in case some of you have noticed that the time on these postings doesn’t really correlate with real time, here’s why: I’m writing this and posting on my Mac, while doing the X2 work on my Windows XP machine. I save the screen shots to a USB drive, and wait until I have a batch of at least two I three before I transfer them over to the Mac and actually do the posting. Yeah, I know, I should network the two…maybe next month LOL.)

[Anderton]

The toughest waveforms to loop were the SPEC waveforms, which are strange, looped digital sounds. Here, there was no chance of creating a single cycle, because that would destroy the sound's essential character.

I decided the best strategy was to find the original loop points as closely as possible, then use the Loop Processing options to help smooth things out. Click on the attachment to see the Loop Processing dialog box. This is the box that enables crossfade looping.

For those who need to get up to speed on crossfade looping, the basic idea is that a little bit of the signal prior to the loop start is crossfaded in just prior to the loop end, and a little bit of the signal just past the loop end is crossfaded in just past the loop start. That way, there’s a smoother transition when the loop end jumps back to the loop start.

The Loop Processing dialog box may not seem very deep, but there’s enough to create seamless crossfade loops. The “Compress Loop” option evens out levels throughout the loop, which promotes better looping as it minimizes big volume differences between the loop start and loop end. “Truncate After Loop” is something I never check; maybe I’m just hyper-cautious, but I always like to leave some signal after the loop just in case I need to retweak the loop someday.

Crossfade lets you specify a length in samples, but conveniently, it won’t let you specify a range longer than the maximum range that can be crossfaded (i.e., you can’t mix in signal before the start of the sample or after the end of the sample). You can also specify a linear or equal power crossfade. I just try both, undo, and use whichever one sounds better.

Getting a good crossfade is much more difficult than getting a good single-cycle loop with a constant waveform, but it wasn’t too hard to whip the waveforms into shape. I also learned to always try the Autocorrelate button first: With a surprisingly high percentage of loops lightning would strike, and I’d get a perfect loop without even having to crossfade. But I'd add a 20 ms crossfade or so, just in case

[Anderton]

Just for kicks, I thought I’d do another “timed trial” on how long it would take to do these more difficult crossfade loops. I started at 10:21 PM, and finished doing crossfades on 11 loops by 10:36 PM – much better than I expected.

[Anderton]

X2: OTHER EDITING TOOLS

So far, I’ve touched more on looping than anything else. But there are several other editing tools.

I’m not sure I mentioned there’s multiple undo, so you can try out quite a few processes on your samples and always get back to where you started (I have my history size set to 100 edits I the Preferences dialog box). You can clear this if desired, and specify automatic clearing when saving. However, note that this is not an undo history where you can arbitrarily go back to a particulary step; it’s a one-step-at-a-time “serial” undo.

As to other editing tools, here’s the story:

Fade: Click on the attachment to see the fade dialog box. This is more versatile than you might expect, as there is a choice of four curves (linear and three exponential), and you can set separate Start and End levels. So you can fade in, fade out, fade from partway up to partway down, etc. This affects only the selected portion of the waveform.

Normalize: This raises the level to maximum, but is fixed at maximum – you can’t normalize to, say, -1dB. For this, you would instead use the…

Adjust Gain: You can boost up to 96dB, and take the level down to minus infinity. You can also normalize from within the Adjust Gain dialog box.

DC Filter: Removes DC offset from the sample.

Reverse: .it desseug uoY
Swap Channels: Simply swaps the left and right channels.

Bit Reduction: OK dirtmeisters, this goes from 16 bits all the way down to 1 bit.

Time Compression: This ranges from a ratio of 50% (twice as fast) to 200% (twice as slow), with your choice of 12 different algorithms. It works about the same as other time compression algorithms – really well on some material, not so well on others. It also includes a handy calculator that shows the final time and BPM after tweaking. This is for display purposes only; you can’t actually enter numbers in there and have the ratio change automatically – all changes are done with the ratio control.

Pitch Shift: This covers –1200 to +1200 cents. You can shift pitch by nudging samples, or using something called “Gossett-Smith” tuning (which is not explained, and I have not idea what it is – other than the fact that using either of these two options doesn’t mess with whatever loop points you’ve set up). A third option, Constant Time, shift pitch while retaining duration. As with Time Compression, you have 12 possible algorithms. This option does mess with the looping, so you want to apply it before you start setting loop points.

Transform Multiply: Copy a sample to the clipboard, and Transform Multiply multiplies the current waveform with what’s in the clipboard. There’s but a single control, Intensity, and thankfully the process takes much less time than it did back in the days of the Emax – because you still don’t have a clue about what the final result will sound like until you play it back! At least know you know you’ve created a clunker or a masterpiece in a matter of seconds, not hours.

Convert to Mono: I’m sure you can figure this out.

Convert to Stereo: Ditto.

Sample Rate Conversion: Here you enter the target sample rate numerically. This also changes key, of course, and a little calculator shows you what the end frequency will be.

There’s also a beat analysis option, but we’ll get into this later.

As I worked with these various options, the exceptional sound design possibilities started to dawn on me. No law says the waveform has to be something you bring into the X2 and layer across a keyboard; you can treat the X2 sort of like Wavelab, Sound Forge, Peak, or other waveform editors. In fact, I’ll try that offline and report back if it’s either something that’s really cool or something that’s really cumbersome.

[Anderton]

I was about to wrap up for the evening, when I figured I'd import a short musical piece and do some sound tweaking. I copied a drum part from the beginning of the piece, and transform multiplied it with the entire piece.

Whoa! It produced this primitive, rhythmic Peter-Gabriel-meets-digeridoo sound, but it was sort of muffled and strange.

So I kicked it up an octave, then up another octave. It sounded even better. I didn't have to do much more to figure out that you can just throw sounds at the X2, mess around, and get some really cool sounds. And this was just a quick experiment before shutting down! Yeah, I keep getting more and more impressed...

[Anderton]

I freely admit this is addictive, and I just had to try "one more" thing.

I wanted to put together a new preset with the waveforms that I'd swiped, and started paring down the multisamples and checking out some layers and...well, suffice it to say it didn't take much enveloping and filter tweaking to come up with some really fun sounds.

[dahkter]

Good read, thanks Craig