Korg Forums

[romandesign]

I hope you find it useful, as it's not an easy task to figure out. My first experiments were a total disaster, because when I just attempted to import freshly created and perfectly fine SF2 file to Nautilus - I ended up with 2 programs with 3 oscillators and no sound at all - unusable. But eventually I solved all problems, and now I'm getting results that are almost perfect. So here's the whole process in brief:

1. SETTING UP: Download and install AKAI MPC Beats (free). Load your desired VST into the first track. You may need a free VST wrapper called Element - I needed it to load Kontakt 7 into Beats. Wire Element MIDI in to Kontakt and Kontakt's first 2 outs to Audio Output.

Load your sound and disable Delay and Reverb effects in VSTi - you will add them later in Nautilus, which will sound much more natural. Disable Glide if used, you can also add it in Nautilus - I was surprised by how well the Glide works in Nautilus, when using samples. I left the frequency at 48kHz - it has enough headroom for lowering pitch of the samples without losing quality.

2. AUTOSAMPLING: use MPC Beats Tools -> Auto Sampler. It makes things very easy and uniform.
- Set sound volume to be high but never clipping
- IMPORTANT: USE VERY SHORT BASE FILE NAME - 4 CHARACTERS OR LESS. Leaving it at default created quite long filenames that are wreaking havoc somewhere down the chain and result in incorrect sample/pitch regions. This is the first issue that was ruining the result for me and took a long time to identify and solve.
- For best results I use the following settings that I tested to get great sounding sounds:
Range = C2-C5 (or whatever is suitable for a particular sound)
Note Stride = 3 (sampling every 3 notes)
Single layer, velocity = 127 (though you can sample several layers, I didn't need it for what I was doing so far)
Note Length = 15s. Disk streaming means Nautilus won't waste more RAM for longer samples, it's only the number of samples that matters, and in my experience RAM is more precious than SSD space, and runs out faster, so longer samples are just fine. You can do with no loops at all with piano etc. but synths and pads should still have a loop.
Tail = 3s (or whatever is needed for the sound to die out between notes)
Loop Start = 5s, Loop End = 15s (end of your sample). Loops will sound bad but that will be corrected later. It's important to set the approximate loop region wide enough with this setting.
Run the Auto Sampler. It will play and record notes for you.
Save the file (you won't see WAV files until the file is saved).

3. GENERATE SF2 SOUNDFONT:
Download and install Polyphone (free)
- Click New Soundfont
- Click Add Samples icon (first one) - select and import all samples generated by Beats. They will be automatically mapped to keys, lower and higher ends expanded.
- While imported samples remain selected, click Add Instrument (2nd icon) and name it. Make sure "link selected" is checked.
- Click Add Preset (3rd icon) and name it the same.
- Go to the new instrument you created and make sure in "Global" the "Loop playback" is set to loop, unless you want it to not loop with your samples.
- IMPORTANT: SET GLOBAL VELOCITY RANGE TO "0-127"! This is the second major issue - leaving it empty as it is by default results in failure! You can also set ADSR envelope values approximately - not required, you will edit them in Nautilus anyway, but this saves time a bit. I advice setting the attack to 0 as the attack is recorded, release can be 0.5 or whatever you want.
- Select all samples, click on Toolbox icon and run Auto Loop. This will automatically find a better looping points close to the loop markers created by Beats, set the loop and add a cross-fade to the end of the file so the loop would be seamless and cut off the tail part of the file. Very neat and normally sounds great. Most loops are unnoticeable after that, and with long 15s samples loops are not an issue. Regrettably, there are no settings for how it does auto-loops and cross-fades, but so far I found the results great, if loops are large enough. It can still sound bad if loop is 3s, but sounds fine if loop is 10s or so, and the sample is 15s. This function is destructive, after you save you won't be able to get your samples back, but you always have the original WAV files.
- Save the file, copy it to a USB stick.


4. IMPORTING SOUNDFONT AND EDITING PROGRAM
- The rest is a normal procedure, i.e. clear auto-load checkboxes in Global, do auto-load to clear all RAM. If you are working on many sounds, load the current set (and only that!) into RAM.
- Load the .SF2 file from the USB stick in Nautilus into desired Program slot
- This, unlike default SF2, produced perfectly playable patch with a single oscillator that is correctly configured.
- Edit the program. I usually add a bit of Drive in Amp, EQ High +6, Low +2, map a touch of LFO1 (sine for vibrato or square for gate, which sounds great around 30% on synths, LFO synced to BPM) to amp and/or pitch mod to Joystick Y+, do parallel filters, with second filter flat/default and map first filter and its resonance to Joy Y- so the sound feels alive. Add Insert FX to mimic delay and reverb of the original VSTi. For me, BPM Stereo Delay and O-Verb with low and mid EQd out usually works best. Don't forget to Write Program.
- Go to Media and save sample memory into your set. then add the set to Autoload, but not the normal one that loads it into RAM but with "_user" in the filename, which loads it for SSD streaming, only using RAM for pointers.

Using this process, I have made several complicated granular synth moving sounds yesterday, the kind that I found were lacking in my new KORG Nautilus, and that I wanted to add to my band's sound, and it worked perfectly every time, so far. I then A/B compared them to the original instrument, and I was able to get it to sound extremely close. I couldn't tell which instrument was playing unless I was looking on the board to see which source is muted. The results are better than I expected. Even glide for an analog-based synth was working great, which is pure black magic, I think. The patches were just as good as other Nautilus patches and don't sound any rougher or less polished, which is great! I don't think I got any audible difference, and it feels like the only thing I lost compared to the original VSTi is all the controls. But on stage all I have time to control is two things: pitch and vibrato/filter/gate, and those are working fine. I don't even usually use the 6 knobs on stage. Anyway, I hope this guide helps, and saves someone a lot of time and a having very dumb feeling of not being able to import a sound into Nautilus with perfectly fine soundfont files. As a note, sounfont files are just a wrapper for original WAV samples, so while the format is considered low-end and free soundfonts often sound like garbage, it's perfectly fine for transferring high-end samples into Nautilus without any quality loss.

[voip]

Thank you for this really detailed guide.
.