Korg Forums

Rob Sherratt wrote:

Hmm... The Korg Kronos do include Disk Streaming from the internal SSD. So, Korg did developed a way to streaming samples from disk to convertor. The same way with Gigastudio, Kontakt or any other Disk-Streaming capable Player on a neKo or any DAW

Sorry but I can't agree this comparison at least not based on the operation of my M3 and Pa2x.

Kronos is not M3 nor PA2X - M3 and PA2X don't do DFD! Kronos indeed does DFD in the same vein as Kontakt or GS do. And it works.

I think everyone is aware that Motif XF needs an expensive upgrade to achieve 2 GB for user samplers. Even with it's limitations, this feature is awesome. I won't cry about a few hundred dollars for an upgrade when the workstation costs several thousand.

Bottom line for me is...more sample time is always a plus. In this day, samplers should come with at least 1 GB for user samples. But even better, to have user samples at you finger tips as soon as you turn on the workstation is a game changer. That is why there is interest in having user samples streamed from the SSD.

EvilDragon wrote:

Rob Sherratt wrote:

Hmm... The Korg Kronos do include Disk Streaming from the internal SSD. So, Korg did developed a way to streaming samples from disk to convertor. The same way with Gigastudio, Kontakt or any other Disk-Streaming capable Player on a neKo or any DAW

Sorry but I can't agree this comparison at least not based on the operation of my M3 and Pa2x.

Kronos is not M3 nor PA2X - M3 and PA2X don't do DFD! Kronos indeed does DFD in the same vein as Kontakt or GS do. And it works.

So why does the limited amount of sample RAM matter in that case? Why's there any constraint on sample sizes or number of samples? Why this thread? Why not just close the matter by saying the sample ram amount is now irrelevant?

sani wrote:Thanks for the clarification. Sounds really impressive.
You just omitted the question about seamless switching. Not a deal breaker necessarily but still highly important to a lot of live players.

It does that too, but it's an option you can set to have on or off. On, it uses more system RAM, off it uses less.

Personally, I'd recommend only activating seamless switching to the actual sounds you will be using all the time and leave the ones you don't use during a gig set to off. That way you have more memory free for running VSTi's or whatever else you want to run. Even with it off, your still taking super fast loading times on most sounds.

See this video for example of a lower version of the OS running on a keyboard.
http://www.youtube.com/watch?v=FwqnVgNO4bQ
That's how fast it is with seamless switching off.

Regards
Sharp.

jahrome wrote:But even better, to have user samples at you finger tips as soon as you turn on the workstation is a game changer. That is why there is interest in having user samples streamed from the SSD.

Yep, definitely. Kronos got the first part right - you can preload user samples. Now if it could only stream them, that'd be downright killer.

Rob Sherratt wrote:Sorry but I can't agree this comparison at least not based on the operation of my M3 and Pa2x. The data from SSD is streamed into sample RAM in its entirety before samples can play back. If the sample is not in RAM and there is a patch change request requiring a sample that is still in SSD then there is a long delay while it is retrieved in entirety before it can be played back. And if the available sample RAM is full, then tough luck loading any more samples.

Sorry but you're wrong... Samples don't need to entirely be loaded in RAM before to be played... Kronos isn't M3 or Pa2x at all.

Conversely, Gigastudio, Kontakt etc fetch samples for playback from HD as and when they are needed and they start playing as soon as the first few bytes are in the playback buffer. A patch change Request results in a new sample being fetched from HD but it plays immediately after the seek-to-sector operation finds the start of the sample. A critical issue is the OS switching latency which on Windoze is no better than 20ms, hence we do not use Windoze. With real time Linux we get note on latencies of 5ms or better even when the system is fully loaded.

I don't go against your theory or facts about Linux latencies and windows latencies... however, I can tell you that it actually somehow does work on windows perfectly fine. Of course, there's some buffering or even loading time, but with a fast system, it's completely transparent for the end user.

Also, I understand you fully promote Linux since it's what you're using in the Lionstracs products... as well as the Kronos does. I don't try to say the Lionstracs isn't good at all, but you're facts are just wrong about the Kronos.

The Kronos is based on a PC Architecture, running Linux, with special GUI and Environment to run different Sound Engines and support Disk Streaming... Tell me how is it different from your Groove Series?

Maybe some major architectural change has happened in the Kronos? But with an ARMless processor I doubt it.

Of yeah, and you surely haven't enough check it... because it actually works fine, even with large samples streaming, such the 4.7 Gb pianos, with 400 voices of polyphony (SGX-1)... and can load something like 12Gb of sounds ready to be used at any moment, without any loading time or latency, on top of the other 7 Synthesis.

(Factory setup loads 11Gb of sounds and still got 270 Mb of free RAM for sampling).

In the other hand, announcing 160 multi-timbral parts is just marketing B.S... Because, if you do load some heavy-duty Analog synth such the AL-1 of the Kronos, you won't get much resources free. Same thing with the filters and anti-aliasing quality on the Sample Playback sound engine in comparison of the HD-1...

Again, I don't say Lionstracs isn't good at all, but just that all announcements/advertising on other manufacture forums (I remember them on Open Labs forum) and now here... aren't really a nice marketing approach in my honest opinion... You shouldn't have to do that. Just make a great product with great features and awesome sounds and success will come by itself. If it doesn't, it maybe just means the product isn't good enough. I tried the Mediastation (back in 2003 or 2004, I can't remind) with Dominico, and even if the concept is great, the features look interesting the overall quality wasn't really there... or at least, really far from what we can get from Korg.

However, I'm also not saying the Kronos is perfect... Of course not! We all want it to have 4 (or even Gb of RAM and custom samples streaming... on top of other few features (such improvement for Sequencer, Sampler, Editable Karma GEs, Physical Modeling for Reeds, Woodwinds, Brass, etc...).

But until then (if it ever happen), we know we'll get a nice well-designed keyboard with already great sounds and features.... and there's no need to go anywhere on other manufacture forums to let people know about it

My 2 cents,
Phil

EvilDragon wrote:

jahrome wrote:But even better, to have user samples at you finger tips as soon as you turn on the workstation is a game changer. That is why there is interest in having user samples streamed from the SSD.

Yep, definitely. Kronos got the first part right - you can preload user samples. Now if it could only stream them, that'd be downright killer.

For the record, I like what Korg is doing. As soon as the Kronos is available here in Japan, I am getting one.

However, there are a few aspects of what we get when we purchase workstations that I would like to see improved. All of these features revolve around sampling. The Big 3 (Korg, Yamaha, and Roland) continue to make advancements but also come up short:
1. Roland by far (in my opinion) has the best sampling and sample editing technology of the three companies. However, they won't put all of their technology into one product (see V-Synth, Fantom X/G, and MV-8800). I probably wouldn't consider Kronos if Roland combined the best of these synths/workstations and put it in one killer package.

2. Yamaha. I love the fact that users can install Flash Ram boards for user samples although quite expensive. But it would be nice to have 1 GB of Ram to work with for sampling instead of 128 MBs at a time. I also like the fact that Motif is the only workstation of the 3 (to my knowledge) whose projects can be opened natively on a DAW (Nuendo/Cubase). I would like to see Korg or Roland do the same. I was about to replace my Fantom G with a Motif XF until Korg announced the Kronos.

3. Korg is headed in the right direction. Oasys and M3 have good sampling and sample editing functions. The huge library of sounds is awesome. But more user sampling, streaming user samples from SSD, and and easy way to export/import Kronos projects in a DAW (such as Pro Tools) would really take it over the top in my opinion. Yes, I am buying Kronos for what it can do today. But when the first "hardware" workstation that comes along (if ever) that can accomplish many of these chores, I won't hesitate to buy it.

I agree with you Jahrome.

The feature I also like a lot (it is/was on the Fantom) is the continuous Sampling. Let's face it, even 60s of stereo Wav 16/44.1 is no more than 10 Mb. Let say we do sample separately internal sound and inputs... It will be something like 20 Mb and not so much more if we also record all midi cc...

So I could deal with 25Mb less in RAM if I could have continuous Sampling... How many time you play something, sing some lyrics or even manipulate sounds with Arps or Karma and you find something you can't reproduce even 2 mn later... Having a 60s "buffer" on the Kronos would change my life... clearly!

That's another one for the wishlist

Phil

jahrome wrote: 1. Roland by far (in my opinion) has the best sampling and sample editing technology of the three companies.

It's clear that this is your opinion but I (in my opinion) disagree completely on that. Roland might have a cool sample editing feature which is not present on a Korg, but this all highly depends on the user and what kind of samples he uses.
Regarding Roland and especially the Fantom G (which I own), there are some big shortcomings based on what I do and need:
1. No loop crossfade function - without it it's almost impossible to create clean loops on a lot of sampled material
2. No multisampling file format - there is no way to add single multisamples to your existing project or exchange them with other users. Since the Fantom S you had to redo every single multisample you use in your work.
3. The Fantom G doesn't even import the loop, just the loop points from a wave. You have to turn it manually on on every single looped sample you import.
4. A cumbersome implementation requiring the user to apply Emphasis to every imported sample (to restore some high frequency contents) and Deemphasis if you export a wave file out from the Fantom G.

sani wrote:

jahrome wrote: 1. Roland by far (in my opinion) has the best sampling and sample editing technology of the three companies.

It's clear that this is your opinion but I (in my opinion) disagree completely on that. Roland might have a cool sample editing feature which is not present on a Korg, but this all highly depends on the user and what kind of samples he uses.
Regarding Roland and especially the Fantom G (which I own), there are some big shortcomings based on what I do and need:
1. No loop crossfade function - without it it's almost impossible to create clean loops on a lot of sampled material
2. No multisampling file format - there is no way to add single multisamples to your existing project or exchange them with other users. Since the Fantom S you had to redo every single multisample you use in your work.
3. The Fantom G doesn't even import the loop, just the loop points from a wave. You have to turn it manually on on every single looped sample you import.
4. A cumbersome implementation requiring the user to apply Emphasis to every imported sample (to restore some high frequency contents) and Deemphasis if you export a wave file out from the Fantom G.

It's ok to disagree. We have had a different experience. What I am talking about is a culmination of Roland's technology and not singling out specific functions on a single model. My "opinion" is based upon the technology in the V-Synth and MV-8800 in addition to what the Fantom X/G.

1. Variphrase technology/real time time stretch(BPM sync) & pitchshifting
2. Graphic User Interface/Graphic waveform editor
3. Manual Adding Chop points
4. Effects processor

I will be more than happy to go into specifics. This functions are key aspects of Roland workstations that Korg and Roland can't match in my opinion. The problem I have with Roland is that they decided not to incorporate their best technology into a single workstation.

jahrome wrote:4. Effects processor

Korg has better effects than Roland, overall. And they are both trumped by Kurzweil. Yamaha is below them all.

EvilDragon wrote:

jahrome wrote:4. Effects processor

Korg has better effects than Roland, overall. And they are both trumped by Kurzweil. Yamaha is below them all.

Korg has better effects? Better then Roland's COSM effects? Please give some specifics. Because I have never heard anyone boast this before. I just can't recall a third party attempting to emulate Korg effects...which is the opposite of Roland whom also work with 3rd parties to give their products even better effects.

Overall. Better reverb, better chorus and phasers and flangers, better rotary speaker simulation. More interesting and weird sounding effects. Pretty good distortion - but Kurzweil has the best distortions that I've ever heard in a workstation. Puts Roland's COSM buzzword to shame.

It's obviously also a matter of taste as well - but that's just how it is with me. Don't care about 3rd parties.

Rob Sherratt wrote:
So why does the limited amount of sample RAM matter in that case? Why's there any constraint on sample sizes or number of samples? Why this thread? Why not just close the matter by saying the sample ram amount is now irrelevant?

Rob,

I believe the Kronos is only streaming samples from the "mega sample instruments" -- i.e. the 2 big pianos and big drums. All of the Kronos expansion sounds (the ones from the OASYS) are pre-loaded into RAM so they don't have to be streamed at all.

jahrome wrote: But it would be nice to have 1 GB of Ram to work with for sampling instead of 128 MBs at a time.

Yamaha would argue that you don't really need any more than 128MB for active sampling, since 128MB represents something like 20 minutes of sample time.

As posted by one of the Yamaha people over at their forum the idea is you use the 128MB for active sampling, then you store your finished sampling work on the Flash ROM.

EvilDragon wrote:Overall. Better reverb, better chorus and phasers and flangers, better rotary speaker simulation. More interesting and weird sounding effects. Pretty good distortion - but Kurzweil has the best distortions that I've ever heard in a workstation. Puts Roland's COSM buzzword to shame.

It's obviously also a matter of taste as well - but that's just how it is with me. Don't care about 3rd parties.

I see. You are saying that by your ears, Korg's effects sound better. Ok. I respect your opinion. I will for sure write about what I think about Kronos effects when I pick mine up this Spring. Since this is the first time I am hearing such a claim, I definitely want to hear about what others feel about this. *Just read some SOS reviews. Roland seems to always get good reviews about the sound quality of its effects..more than Yamaha and Korg.

MartinHines wrote:Yamaha would argue that you don't really need any more than 128MB for active sampling, since 128MB represents something like 20 minutes of sample time.

As posted by one of the Yamaha people over at their forum the idea is you use the 128MB for active sampling, then you store your finished sampling work on the Flash ROM.

And if I was Yamaha trying to sell workstations and expensive flash RAM, I would probably try to convince customers as well that 128 MB is enough. But when the previous version of Motif had 1 GB of sample RAM, it is still a hard sell (thus the reason why this thread was started). Ram is used by the sampler as well as for audio tracks. 20 minutes will be quickly eaten up if user decided to add multiple audio tracks to a song.