Korg Forums

elvisjohndowson wrote:The results of the test is a zip file that is 313MB. Would anyone be able to give me FTP access to a server where I can upload it? It has the sequence data and audio files for resampled performances from the sequencer directly to the internal hard disk.

Why is it so huge? How about a few short WAV files that show the difference?

Hi,
Here are the list of combis, programs and preload songs that have been tested.

Programs
A114 Mega Mute Guitar

Combis
A032 Monster Power Synth
B057 Feel the Rush
B068 Dust Storm on Mars

Preload Song
Teknosys


Test Procedure Realization and Observations

1. Sony MicroVault Tiny USM1GH 1GB, when formatting with OS 1.1.0 it shows it as MS-DOS and 253MB RAM instead of 1GB. Tried both FAT 16 and FAT 32 full format options. After re-formatting it using FAT32 on a PC, it shows the available memory at 990M. However, the volume label does not get displayed correctly. It shows RDD:NO NAME, but I've named it SONYUSM1GH while formatting it.

2. Load and resample to preload.sng to the internal hard disk.

Select and load preload .sng .ksc .pcg
Resample Song S002: TEKNOSYS EXs1
Resample Combis and Programs

3. Update to OS 1.1.1
Save global setting to O111_BP.pcg (before loading preload.pcg)
Check to see if there is any difference in playback for the combis. [None detected, non MIDI playback]

Load preload.pcg
Save global settings to O111_AP.pcg (after loading preload.pcg)
Check to see if there is any difference in playback for the combis. [None detected, with MIDI playback]

Resample songs to hard disk [MIDI playback after loading preload.pcg from OS 1.1.1].

Result : No differences detected between OS 1.1.0 and OS 1.1.1 upgrade.

4. Update to OS 1.2.1
Save global setting to O121_BP.pcg (before loading preload.pcg)
Check to see if there is any difference in playback for the combis. [No differences detected ]
Load preload.pcg

Save global settings to O121_AP.pcg (after loading preload.pcg)
Check to see if differences exist.

Resample songs to hard disk.

Check to see if differences exist in the songs with just the OS 1.2.1 upgrade. [Subtle differences EXIST. More reverb type of effect detected for Program A114 Mega Mute Guitar after loading new OS 1.2.1 preload.pcg]

Load LAC-1.pcg
Save global settings to O121_AL.pcg (after loading LAC-1.pcg)
Play back programs and combis live.
Result: Differences certainly exist for Program A114 Mega Mute Guitar after loading new OS 1.2.1 LAC-1.pcg. Confirmed subtle loss of sound definition and added reverb during live playback.

Resample songs to hard disk [MIDI playback after loading LAC-1.pcg from OS 1.2.1].

Check to see if differences exist.

Result: Differences exist. Confirmed subtle loss of sound definition and added reverb. Variations detected in song Teknosys intro and alteration of the tonal charateristics of some of the sounds used in the song during live playback.

Elvis Dowson

Here's something to check out:

On CD1 of your 1.1.0 set, there's a PRELOAD.PCG file, what happens if you load or browse to this sound from that CD while your machine is running 1.2.1? Does it sound like 1.1.1 or 1.2.1?

StephenKay wrote:

elvisjohndowson wrote:The results of the test is a zip file that is 313MB. Would anyone be able to give me FTP access to a server where I can upload it? It has the sequence data and audio files for resampled performances from the sequencer directly to the internal hard disk.

Why is it so huge? How about a few short WAV files that show the difference?

Umm, yeah, you're right, I'll try to trim it down. The only problem is time alignment of the waveforms. Well, that package contains the full data set, its large because of the Teknosys song. I noticed some variations in the sounds. I've re-sampled the song thrice for each OS version. The easily discernible difference I can see is in A114 Mega Mute Guitar during playback. There certainly is more added reverb and the global settings haven't changed.

Stephen, what type of spectral analysis tool would you recommend, shall I use the one in SoundForge 8.0 and do an FFT analysis or do you have another recommendation?

Elvis Dowson

Brandon Daniel wrote:Here's something to check out:

On CD1 of your 1.1.0 set, there's a PRELOAD.PCG file, what happens if you load or browse to this sound from that CD while your machine is running 1.2.1? Does it sound like 1.1.1 or 1.2.1?

Hi Brandon,
I just checked this out. I loaded the PRELOAD.PCG file from the OS 1.1.0 CD and played A114 Mega Mute Guitar and it still sound the same as OS 1.2.1, i.e. the same ambient reverb being applied to the sound in general, compared to how it sounded when I was running OS 1.1.1. OS 1.1.1 did not have this ambient reverb applied.

If you would like and can give me a link to an FTP server, I can upload the full file set (313MB).

Elvis Dowson

elvisjohndowson wrote:The easily discernible difference I can see is in A114 Mega Mute Guitar during playback. There certainly is more added reverb and the global settings haven't changed.

Elvis, is it just more reverb that seems different? Some of what you suggested earlier was that the actual Oscillator quality was compromised. This is why I suggested doing a test with all effects (reverb and the rest) off.

I can adjust reverb and effect settings to my taste. Can you tell me if there is actually a difference in the dry output?

Thankyou, for noticing no change in Velocity curve.

elvisjohndowson wrote:Umm, yeah, you're right, I'll try to trim it down.

That would be good, because even if I gave you space on my server, I can't imagine many people are going to download a 313 mb file to see if they hear what you hear.

Stephen, what type of spectral analysis tool would you recommend, shall I use the one in SoundForge 8.0 and do an FFT analysis or do you have another recommendation?

Hmmm... I don't have any particular recommendation. I was kind of waiting to see what Dan had to say about the comparisons he had prepared. I think any tool like that would be worth looking at. Although if it's some subtle difference in the reverb that you're hearing, I'm not sure it would necessarily show up visually. But comparing the waveforms would be a first step.

Mike Conway wrote:Elvis, is it just more reverb that seems different? Some of what you suggested earlier was that the Oscillator quality was compromised. This is why I suggested doing a test with all effects (reverb and the rest) off.

Hi Mike,
Let me just try this out. But if you had told me this earlier, I could have tried removing all effects in the OS 1.1.1 tests and repeated it for the OS 1.2.1 tests.

For A114 Mega Mute Guitar, there are two reverb effects applied,
IFX3 : 106 Reverb Bright Room
MFX2: 102 Reverb Smooth Hall

Switching off the effects effectively dries the sound completely, but I should have listened to it when I had OS 1.1.0/1.1.1 running in order to compare.

The factory presets shouldn't change as far as possible between versions for the same version of the patch for OS updates.

Elvis Dowson

StephenKay wrote:Although if it's some subtle difference in the reverb that you're hearing, I'm not sure it would necessarily show up visually. But comparing the waveforms would be a first step.

I'm not so sure if something like added reverb would show up in an FFT analysis.

Elvis Dowson

StephenKay wrote:Hmmm... I don't have any particular recommendation. I was kind of waiting to see what Dan had to say about the comparisons he had prepared.

I'm still working on it...sorry for the delay. Hopefully, I'll have more soon.

StephenKay wrote:I think any tool like that would be worth looking at. Although if it's some subtle difference in the reverb that you're hearing, I'm not sure it would necessarily show up visually. But comparing the waveforms would be a first step.

Since many things might be slightly different from playback to playback (for instance, any randomized phase, randomized LFOs, un-synced effects LFOs or Program Common LFOs, etc.) I'd expect that slight variations in waveform and/or spectrum would be the norm.

- Dan

Elvis, I'm going to PM you a log-in for one of my servers. When you have your tests finished, you can upload them there. Note that it's "karma-lab.info", not ".com". I have some free disk space over there. When you are done, the file(s) can be accessed by anyone, using this URL (example only):

http://www.karma-lab.info/elvis/filename.zip (or whatever name you give it)

I'd like to help you get to the bottom of this so we can put it to bed.

Hi,
I tried listening to the WAV files that I resampled for all three OS versions, and played back using MIDI, but I cannot tell the difference between these waveforms, when played back through my laptop headphone outputs. I don't think that the S/N level ratio is good enough on my IBM Thinkpad T42p to be able to discern the difference in output.

I am however, able to clearly make the difference for A114 Mega Mute Guitar between OS 1.1.1 and OS 1.2.1 from the main headphone outputs on the OASYS.

I just thougt I'd let you guy know this. It probably has to do with the quality of the D/A converters on my laptop. It's not studio grade.

Elvis Dowson

elvisjohndowson wrote:Let me just try this out. But if you had told me this earlier, I could have tried removing all effects in the OS 1.1.1 tests and repeated it for the OS 1.2.1 tests.

I did. Page 3 and page 4, of this thread. FX or not, I'm sure Dan will have this figured out, soon. It would be interesting to hear the smaller wav files.

Hi Stephen,
I tried putting the file to the FTP server but I got an error.

230 User elvis@karma-lab.info logged in.
ftp> bin
200 Type set to I
ftp> put O121TST1.zip
200 PORT command successful
550 Access is denied.
ftp>

Elvis Dowson

Mike Conway wrote:I did. Page 3 and page 4, of this thread.

Sorry about that Mike, I missed that one.

Elvis Dowson