OK, since i feel this Compression RAM Issue is VERY important, i started a new thread on it.

On the next post you'll see a bunch of talk already on it (copied from the Leezone AUDYA review thread.)

lets get to the bottom of this :-)

leezone
Member posted 02-24-2009 02:18 PM

Memory Issues:
The one thing that was discussed was the miniscule 64MB of Ram. I think this is an issue which Ketron has to explain a bit better. I will attempt to explain this a bit…
From what AJ said and from what I understood, the AUDYA compresses your samples, or what’s loaded to keyboard. It compresses it NOT by degrading the sound as does an mp3, but rather it kind of .ZIPs it, but with a Ketron proprietary compressing program. I believe it compresses somewhere near 12x or maybe more. So lets say you want to load samples which total 62MB (which normally would fill up AUDYA’s entire RAM),,,, In reality AUDYA would compress the 62MB to let’s say about 5MB. So in reality what we think is 64MB of RAM is actually about 768MB. (assuming compression of 12x)
AJ or anyone, PLEASE do correct me if I’m wrong. But that’s how I understood it, and that’s how I can best explain it. I do not want to assume or put out false info on this so please do correct me if I’m wrong.
It just seems that IF this is how AUDYA handles the memory, Ketron should really think about HAMMERING this into our brains and making everyone aware of this. Promote this Compression feature… You can say something like it has 64MB but compression gives you the equivalent of roughly xxx MB
AJ, please do explain this a bit further as I am a bit uncertain how it works…
________________________________________

mrdave
Member posted 02-24-2009 03:51 PM

quote:

Originally posted by leezone:
...the AUDYA compresses your samples, or what’s loaded to keyboard....

What I can say is that when I tried the Audya and loaded samples, they occupied their actual size in RAM, i.e. loading the stereo grand occupied around 36MB of ram (the size of the file).
___________________________________________

leezone
Member posted 02-24-2009 04:58 PM

mrdave,
i am unsure as to exactly how this whole compression thing works, how it's used,
whether it's compressed to hard drive and then expanded to RAM, which would be useless, SO...
perhaps AJ can explain it in more detail...
___________________________________________

Ketron_AJ
Moderator posted 02-24-2009 07:19 PM

Hard Drive/USB ------> Compressed [12:1] ------> Audya RAM.
So files (when loaded) are "zipped" into RAM. When reading from RAM to Sound card ----> Unzipped for processing/use. Display shows actual file size though.
Hope this helps.
AJ
PS: Leezone - good explanation. I figured someone else would be better off explaining so more understand, than me!
___________________________________________

mrdave
Member posted 02-25-2009 04:51 AM

quote:

Originally posted by Ketron_AJ:
Hard Drive/USB ------> Compressed [12:1] ------> Audya RAM.


So yuo're telling me that the 36MB Stereo piano instrument when loaded into RAM will take only 3MB? PLEASE.. don't be fool... this is somewhat not possible for at least 2 reasons:
1) Lossless audio compression at 12:1 is pratically impossible, you can achieve 12:1 ratio with lossy compression algorithms like MP3 but you'll have a noticeable loss of quality in sound at these high ratios.
2) If this would be the case I could load many supersolo sounds on audya's RAM, instead when loading big files (like the stereo piano) I can barely load 2 or 3 of them before getting out of memory. I tried this on the Audya I had for experiments, so I know what I say...
Note that I tried os ver 1.0, but I don't think they added this "compression" to 1.1.
Do you want to prove it? Simple: make a video where you load the following 3 sounds together:
SUPERSOLO/STEREO_GRAND.INS
SUPERSOLO/POP_ALTO.INS
SUPERSOLO/POP_TENOR.INS
These together are around 96MB in size, if they would be compressed at 12:1 they would take 8MB and could be loaded together.
___________________________________________

Irishacts
Member posted 02-25-2009 05:34 AM

quote:

So yuo're telling me that the 36MB Stereo piano instrument when loaded into RAM will take only 3MB? PLEASE.. don't be fool... this is somewhat not possible for at least 2 reasons:

AJ said above that they are zipping the files. Which means you will likely fit about 70MB of data into the 64MB chip.
Regards.
James.
___________________________________________

trident
Member posted 02-25-2009 05:45 AM

quote:

Originally posted by Irishacts:
AJ said above that they are zipping the files. Which means you will likely fit about 70MB of data into the 64MB chip.
Regards.
James.

The word "zipped" was in quotes... there is a compression scheme called flac that looslesly (sic) compresses audio files (cd ripped audio etc) at about 1/2 their original size, but consumes a lot of processing power. You probably know it better than me.
12:1 ratio? Well, let's see it.
Onthe other hand, a "simple" waveform, such as a sax or piano would be more "compression friendly" than a full, complex song, right?
___________________________________________

LIONSTRACS
Member posted 02-25-2009 06

quote:

Originally posted by Irishacts:
AJ said above that they are zipping the files. Which means you will likely fit about 70MB of data into the 64MB chip.
Regards.
James.
[This message has been edited by Irishacts (edited 02-25-2009).]

yes James, this is about the correct calculation.
I'm working right now for programming the Dream 64Mb flash chip for the new MS too: http://www.lionstracs.com/demo/flashprogramm.JPG
so, I know well how the dream chip work from many years.
The Dream DSP allow to load only compiled soundsfiles.INS or complete binary bank files but NOT raw data like wav.
For playng the sounds in realtime must be compiled first and loaded in RAM/Flash, no compression is allowed.
For create a clean 64Mb file that can fit in one 64mb flash, I have used about 71Mb or raw wav/pcm data and during the compiler wil be optimized/compressed to get a 64Mb binary file.
Conclusion: the file that you have to load on DSP is already optimized and the RAM used wil be the same as the file size loaded.
___________________________________________


[This message has been edited by leezone (edited 02-25-2009).]

what i want to know is if the AUDYA can load 768MB into RAM...

it would seem so, if the AUDYA compresses 12x right? 768/12= 64MB

these 2 posts below seem to contradict...
maybe someone can explain it better???

posted by LIONSTRACS
"Conclusion: the file that you have to load on DSP is already optimized and the RAM used will be the same as the file size loaded."

posted by Ketron_AJ
"Hard Drive/USB ------> Compressed [12:1] ------> Audya RAM."

Hey, maybe they have made a miracle with a special firmware, I really dont know how they can compress audio data 12:1 and play by midi.
this is all what I know from this DSP manufacture and drivers.

Just search in google: ATSAM9707 and ATSAM9708
9707= max ram size: 64Mbytes
9708= max ram size :256Mbytes

MAYBE we are totally wrong again...AJ told that thet have used a PC Pentium 4or dualcore with 1Gb RAM or more, remember?
Ask in AJ, for sure he know better.

what i do know is that according to AJ, Ketron uses a proprietary compression algorithm, software or whatever its called, which compresses somewhere around 12x

how it works, how it benefits loading samples to RAM, how many MB can actually be loaded/used is still a mystery... at least to me

Hi leezone.



I know AJ works for KETRON and the above quote was extracted from his post, but I have serious doubts that this is possible or even what he meant. There seems to be a lanuage barrier here as something so simple is not being explained.

I've never heard of lossless compression even remotely close to 12:1

If FLAC is what was used as was mentioned in that other thread, then you would be lucky to even reach 2:1

FLAC will only give you at best 25MB of more memory.

Ketron need to explain exactly what is happening here and just how much data you can load into the Audya.

Regards.
James.

Yes, FLAC is not optimal in many ways...

James, does my thought that "a simple piano or sax waveform, would compress better than a complex song" have any logic? If yes, maybe that is the way they do it.

Of course all of the thread is based on the condition "we at Ketron managed to compress losslesly on a 12:1 ratio", being true.

(true as in Boolean logic, I am not implyying they lie to us)

Domenico is right, you can load .INS files only, and it will take on RAM the size of the .INS file. So, since these files are uncompressed audio they are loaded as is.
Here is the waveform of the stereo piano.ins loaded as raw pcm data, you can see it's totally uncompressed (playing it I think it's sampled at 22050Hz, since if I play it at higher sampling frequency notes sounds too high).
http://www.4shared.com/file/89421025/c363c0d/wave.html

maybe someone with an AUDYA can try loading let's say 300MB of WAV samples and see if AUDYA can accept it, compress, load & play all samples...

these 300MB should compress to less than 30MB i assume?