Author |
Message |
dorremifasol
Joined: Sep 28, 2006 Posts: 814 Location: Barcelona, Spain
Audio files: 7
G2 patch files: 49
|
Posted: Fri Dec 01, 2006 1:52 am Post subject:
G2 specs may lie miserably |
|
|
After having suffered a great deception with the quality of some sound modules, and having read about the 16 bit memory chips in the G2 I have something to say.
http://www.clavia.se/G2/G2_NM_comparison_chart.htm
This page told me that memory in the G2 was 24 bit (DSP Audio memory: 256kWords of 24 bits per DSP).
I thought I had bought a high quality instrument with 24 bit internal memory and now I see that I can't even use the 4 inputs for effect processing because the extremely low quality of the peak EQ's, delays, reverb and many other modules using the internal memory.
OK, the rest is great (except for the triangular waveform which incomprehensibly has much aliasing whereas the saw and square waveforms not), but the thing is that I feel cheated!!
If the memory is indeed 16 bit, this document where it is said that it is 24 bit must be changed. Well, this is the bug I'm refering to.
Or, perhaps it would be nice if the G2 used the memory as the document says, in 24 bit mode.
/rant mode off
Sorry but just at this time and day I'm really disappointed.
Now I feel better _________________ Cheers,
Albert |
|
Back to top
|
|
|
windchill
Joined: Jan 07, 2005 Posts: 90 Location: london uk
G2 patch files: 1
|
Posted: Fri Dec 01, 2006 10:12 am Post subject:
|
|
|
Clavia did indeed screw up badly here it seems - and you're not the only one who is annoyed. Some members of this forum have suggested that the "specifications may change without notice" clause exempts Clavia from this issue - but I think it is clear this is not the case..... if it was then you could find a banana in the G2 box when you got it home and the legal system would be unable to intervene! |
|
Back to top
|
|
|
mosc
Site Admin
Joined: Jan 31, 2003 Posts: 18197 Location: Durham, NC
Audio files: 212
G2 patch files: 60
|
Posted: Sat Dec 02, 2006 8:54 pm Post subject:
|
|
|
I'm not associated with Clavia and I can't speak for them, but the G2 uses Motorola DSP chips which are 24 bit fixed-point machines. The Kyma system uses the same chips, BTW. The internal memory of those chips is indeed 24 bits. Of course, programmers don't have to use all 24 bits for every application, and apparently Clavia doesn't store all data in 24 bit format. Still, I don't think their statement is dishonest. _________________ --Howard
my music and other stuff |
|
Back to top
|
|
|
blue hell
Site Admin
Joined: Apr 03, 2004 Posts: 24079 Location: The Netherlands, Enschede
Audio files: 278
G2 patch files: 320
|
Posted: Sat Dec 02, 2006 9:25 pm Post subject:
|
|
|
mosc wrote: | I The internal memory |
The issue is about the external memory, which is said to be 16 bits wide. And that's the memory used for delays and maybe some other delay based operations. _________________ Jan
also .. could someone please turn down the thermostat a bit.
|
|
Back to top
|
|
|
Fozzie
Joined: Jun 04, 2004 Posts: 875 Location: Near Wageningen, the Netherlands
Audio files: 8
G2 patch files: 49
|
Posted: Sun Dec 03, 2006 5:17 am Post subject:
Re: G2 specs may lie miserably |
|
|
dorremifasol wrote: | ......I thought I had bought a high quality instrument with 24 bit internal memory and now I see that I can't even use the 4 inputs for effect processing because the extremely low quality of the peak EQ's, delays, reverb and many other modules using the internal memory..... |
Let me first say I agree with you on the 16bit vs 24bit delay issue; this is not correct from Clavia and the fact that they still have that comparison chart on their website has a bad odour to it.
That being said, do you really think that the inputs are not suitable for sound processing? I disagree. I do it all the time - I have no golden engineer's ears, but I like what I hear. The delays may be 16 bit, which appears to be a bummer for some of the physical modeling stuff, but for normal audio I have never actually heard the '16-bitness'. The reverb, ok, it's a built-in reverb that is not very advanced, but it does it's job. For high-quality stuff: build your own or get an external reverb. Digital EQ's are notoriously sterile sounding, but they do work.
Of course, I'm not telling you to like everything in the G2, but are you sure you hear all the low-quality stuff you mention or whether you conclude it because of the 16-bit stories and stuff? I mean no disrespect with this question, it's just that it is sometimes easy to get carried away by discussions here while actually using and applying things in real life may be different. I have had these experiences myself, that's why I ask. _________________ Spinning at ~0.0000115740740741 Hz |
|
Back to top
|
|
|
dorremifasol
Joined: Sep 28, 2006 Posts: 814 Location: Barcelona, Spain
Audio files: 7
G2 patch files: 49
|
Posted: Sun Dec 03, 2006 5:56 am Post subject:
|
|
|
Don't get me wrong, I love this machine, but it incomprehensibly fails in some areas that cut my inspiration a lot.
For example, what's the deal with the peak EQs?? They sound like they were 8 bit when the volume is low, just move the gain knob a bit. The filters sound very good though so I don't understand why the EQs are so bad.
There is a very good patch that I love from Menno Meijer, MiniNord Bass. Well, the sounds have a sort of 8bit end to them, just listen. This nasty sound is produced by a simple peak EQ put just above the VA output. Disable the EQ and the sound is clean again (if you reduce the gain a bit). That means, from my point of view, that a single EQ can thrash the quality of an entire patch and make it sound like it wasn't a 24 bit / 96 khz machine at all.
Does the EQ use the 16 bit memory? I don't know.
The reverb and the delays do not end ever, is that simple. There's always a residual noise going on as long as the delay or the reverb has been feed once, and I can definitely hear it. Of course in a somewhat busy track you may not hear that but the thing is the base quality.
I use always headphones, K271 studio, and I can definitely hear all those low quality noises.
So the thing is, well, I have an incredible 24 bit, 96khz great sounding machine but if I use this or this other module it becomes a low quality machine.
Ok the inputs are good, but it depends on what modules will you use with them. My studio runs at 24 bit 96khz always, all my gear work with that quality and I bought the G2 (which is not exactly cheap) expecting the same quality as well. Still, the comparison chart is not true and I personally don't like being cheated.
However, when I use only good sounding modules I'm really happy with it and I really enjoy its sound. It is very user friendly as well! _________________ Cheers,
Albert |
|
Back to top
|
|
|
Fozzie
Joined: Jun 04, 2004 Posts: 875 Location: Near Wageningen, the Netherlands
Audio files: 8
G2 patch files: 49
|
Posted: Sun Dec 03, 2006 6:20 am Post subject:
|
|
|
That's fair enough, Albert. Sorry to hear you're disappointed. I use the EQ's seldomly, so I haven't noticed. Also, much of the stuff I do is quite experimental / lofi (or amateuristic if you like ) so it might not be strange that you notice these things while I don't. My external signals almost always involve single coil guitar or bass pickups, so those are always noisy.
Anyway, I hope the flexibility, workflow and NM community will keep you from selling the beast. _________________ Spinning at ~0.0000115740740741 Hz |
|
Back to top
|
|
|
dorremifasol
Joined: Sep 28, 2006 Posts: 814 Location: Barcelona, Spain
Audio files: 7
G2 patch files: 49
|
Posted: Sun Dec 03, 2006 7:21 am Post subject:
|
|
|
Oh in no way I'm selling this little red demon (for now)! I like it too much in spite of my complains. But, I think that the only way it could be improved in the areas where it's not good enough is complaining about those, loud and clear. Perhaps this way Clavia will do something. _________________ Cheers,
Albert |
|
Back to top
|
|
|
dorremifasol
Joined: Sep 28, 2006 Posts: 814 Location: Barcelona, Spain
Audio files: 7
G2 patch files: 49
|
Posted: Sun Dec 03, 2006 11:00 am Post subject:
|
|
|
Well I have been doing some experiments with the G2 and I have found some enlightening facts (for me at least, I'm sure most of you already know this).
The "internal" sound quality of a patch is greatly improved if the envelopes are put just at the end of the signal path, before the outputs, not before the EQs and other modules. I should have realized that before, it's a simple fixed point math problem.
I just tweaked 2 of my favorite patches by Menno Meijer (coneblaster dlx and mininord bass) to work this way and the noise is incredibly improved, they sound much cleaner now.
I'm much happier now
Sorry for the rants _________________ Cheers,
Albert |
|
Back to top
|
|
|
mosc
Site Admin
Joined: Jan 31, 2003 Posts: 18197 Location: Durham, NC
Audio files: 212
G2 patch files: 60
|
Posted: Mon Dec 04, 2006 9:37 am Post subject:
|
|
|
dorremifasol wrote: | Sorry for the rants |
Rant on, my friend. You have been respectful and courteous though it all, IMHO. I'm glad you discovered that patching technique. Sometimes, putting a EG/VCA at the end of a reverb effects chain can make things sound funny, but, one can usually figure out something. _________________ --Howard
my music and other stuff |
|
Back to top
|
|
|
cappy2112
Joined: Dec 24, 2004 Posts: 2465 Location: San Jose, California
Audio files: 2
G2 patch files: 1
|
Posted: Mon Dec 04, 2006 11:13 pm Post subject:
|
|
|
windchill wrote: | Clavia did indeed screw up badly here it seems - and you're not the only one who is annoyed. Some members of this forum have suggested that the "specifications may change without notice" clause exempts Clavia from this issue - but I think it is clear this is not the case..... if it was then you could find a banana in the G2 box when you got it home and the legal system would be unable to intervene! |
Any chance it was a cost-related issue as opposed to a design mistake? |
|
Back to top
|
|
|
Chet
Joined: Nov 19, 2004 Posts: 231 Location: Lititz,PA,USA
Audio files: 7
G2 patch files: 35
|
Posted: Tue Dec 05, 2006 10:45 am Post subject:
|
|
|
I think it was a cost issue. Making the delays 24 bits wide would require an additional RAM chip per DSP, and a bigger circuit board.
I like the workaround Tim proposed: add a 'high-quality' button to the delay modules, which instructs them to use two words of memory per cycle. The module would take more DSP cycles, and the total available delay time would be cut in half. But it would give us the option to use high-quality delays. |
|
Back to top
|
|
|
Kassen
Janitor
Joined: Jul 06, 2004 Posts: 7678 Location: The Hague, NL
G2 patch files: 3
|
Posted: Tue Dec 05, 2006 11:14 am Post subject:
|
|
|
dorremifasol wrote: |
The "internal" sound quality of a patch is greatly improved if the envelopes are put just at the end of the signal path, before the outputs, not before the EQs and other modules. I should have realized that before, it's a simple fixed point math problem.
|
That's not nesicarily related to fixed point math (though fixed point math does affect the rounding errors you noticed in the reverb never returning to silence and in fact becoming unstable). This particular envelope issue is caused by the lower bitrate of controll signals. Since the envelope curve (for volume) gets multiplied with the audio signal and the envelope is running at a 4th of the sample rate you get a effect of a sort of ring modulator (in adition to the amplitude modulation of the envelope's shape itself).
[edit] this effect can be increased if there are "resonator type" modules after the env in the chain, such as eq or reverb which I suspect is a big chunk of what you are hearing.[/edit]
The sollution (or at least a workaround) is to use a extra VCA controled by the envelope's curve fed through a LPF. Theoretically this LPF should be a brick wall filter at a 8th of the audio rate (no luck there) but in practice a steep LPF tuned as low as you can go without compromising punch will sort you out.
If you could make Clavia make audio-rate envelopes you'd be rid of that problem but clearly those would have a significantly higher cost in DSP (which is why people often do it like Clavia does, starting with Csound and it's companions).
It's completely possible to have a very solid go at anti-aliassing everything all the time and some designers do. The results can sound marvelous but be prepared for a rather large CPU hit.
Anyway, you have good ears. That's a blessing because you'll hear more beautifull details in music but you'll also be stuck with hearing many anoying things... _________________ Kassen |
|
Back to top
|
|
|
mosc
Site Admin
Joined: Jan 31, 2003 Posts: 18197 Location: Durham, NC
Audio files: 212
G2 patch files: 60
|
Posted: Tue Dec 05, 2006 11:29 am Post subject:
|
|
|
Putting the EG/VCAs at the end of the chain is something you'd want to do in any patch, digital or not. It's standard practice in patching an analog modular. _________________ --Howard
my music and other stuff |
|
Back to top
|
|
|
Kassen
Janitor
Joined: Jul 06, 2004 Posts: 7678 Location: The Hague, NL
G2 patch files: 3
|
Posted: Tue Dec 05, 2006 1:51 pm Post subject:
|
|
|
mosc wrote: | Putting the EG/VCAs at the end of the chain is something you'd want to do in any patch, digital or not. It's standard practice in patching an analog modular. |
Yes, but it's standard practice because of the hiss inherent in analogue filters. Proper digital ones don't have that kind of noise. I find that putting the envelope before the filter (or any sort of resonator) leads to a more natural decay which to me leads to a more relaxed listening experience. _________________ Kassen |
|
Back to top
|
|
|
mosc
Site Admin
Joined: Jan 31, 2003 Posts: 18197 Location: Durham, NC
Audio files: 212
G2 patch files: 60
|
Posted: Wed Dec 06, 2006 9:18 am Post subject:
Q |
|
|
Not to beat a dead horse, but analog and digital both are more noisy with low signal levels. In general, if you are going to both attenuate a signal and and filter it (EQ), for lowest noise it is best to filter it first, then attenuate it. _________________ --Howard
my music and other stuff |
|
Back to top
|
|
|
elektro80
Site Admin
Joined: Mar 25, 2003 Posts: 21959 Location: Norway
Audio files: 14
|
Posted: Wed Dec 06, 2006 10:20 am Post subject:
|
|
|
Sure, but Kassen is still quite correct. At times it makes perfect sense to filter an already enveloped signal or whatever. It all depends on what gear you have and what you want to do. _________________ A Charity Pantomime in aid of Paranoid Schizophrenics descended into chaos yesterday when someone shouted, "He's behind you!"
MySpace
SoundCloud
Flickr |
|
Back to top
|
|
|
elektro80
Site Admin
Joined: Mar 25, 2003 Posts: 21959 Location: Norway
Audio files: 14
|
Posted: Wed Dec 06, 2006 10:24 am Post subject:
|
|
|
elektro80 wrote: | Sure, but Kassen is still quite correct. At times it makes perfect sense to filter an already enveloped signal or whatever. It all depends on what gear you have and what you want to do. |
Anyways, the level thingie was always solvable way back then. It was just a matter of working with a fairly hot signal and use dynamics processing whenever that was called for. _________________ A Charity Pantomime in aid of Paranoid Schizophrenics descended into chaos yesterday when someone shouted, "He's behind you!"
MySpace
SoundCloud
Flickr |
|
Back to top
|
|
|
Electromagnetic Wave
Joined: Apr 28, 2013 Posts: 302 Location: Kebek
G2 patch files: 38
|
Posted: Fri Jan 26, 2018 10:39 pm Post subject:
|
|
|
I feel concerned by this topic because I spent 1 month editing all my patches this summer to add 2 EQ (L/R) before the final(physical) outputs to quickly mix left and right frequencies. As well as the volume for each side on all my patches. Very useful to mix any interslot because the 'patch level' does not act between slots.
About the envelope placed before or after the filter I can add : A gain more or less strong before the filter will treat the harmonics of the signal entering the filter differently. So if for example the velocity is assigned to the envelope (AM input), the signal entering the filter is not strong enough to bring out the high harmonics. But it is also useful to use a filter in series before a second filter to decrease or enhance the desired frequencies / harmonics before entering the second. Then this is not a big problem but placing the envelope before a filter can cause a TIC on the sounds with a fast attack (I'm not refering to problem of non-syncronized cycle of the source signal with the triggering of the sound here. Or with issue related to the release - N / R button to always restart the attack stage at zero level.). In other words, clic/tic problems on fast attacks can often be resolved by placing the envelope after the filter.
dorremifasol (or someone else?)> Regarding your remark and the phenomenon with the Peak EQ module (patch 'MiniNord Bass'), would it not be a form of clipping or overload of the headroom after conversion caused by the reverb module and delays (which are in 16 bits) placed before the EQ? Due to conversions 24 > 20 (VA to FX BUS) > 16 (delay and reverb modules) > 24 (OUT 1/2) bits? I noticed that there is also an EQ in the VA section. Do you disable this one too? I will replace the EQ of the patch that you speak by filters + volume modules to compare the sound of EQ and filters.
That said, I do not quite understand why this solution seems to work?
Quote: | The" internal "sound quality of a patch is also improved if the envelopes are at the end of the signal path, before the outputs, not before the EQs and other modules. |
Edit : for clic or other related issue I can understand. But here dorremifasol speak about a sound that suffers a loss of quality 'They sound like they were 8 bit'.
I will try the solution proposed by Kassen
Quote: | The sollution (or at least a workaround) is to use an extra VCA controlled by the envelope's curve fed through a LPF. Theoretically this LPF should be a brick of the LPF. |
So for me there is no quick fix. Just a custom/DIY chain like : gain > filter > filter > Env > gain > filter > env > etc according to the needs. |
|
Back to top
|
|
|
Electromagnetic Wave
Joined: Apr 28, 2013 Posts: 302 Location: Kebek
G2 patch files: 38
|
|
Back to top
|
|
|
Electromagnetic Wave
Joined: Apr 28, 2013 Posts: 302 Location: Kebek
G2 patch files: 38
|
|
Back to top
|
|
|
|