It is currently Wed Dec 11, 2019 7:04 am


Selecting Bit Resolution at Organ Loading

A discussion forum for anything even marginally Hauptwerk-related.
  • Author
  • Message
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Selecting Bit Resolution at Organ Loading

PostSat Mar 13, 2010 6:50 pm

Each of us, when loading a new organ sample set will select 24-bit
unless we are limited in our computer RAM. However, is that
necessary or suppose we used lower resolution, say 16 bit or
even 14-bit. What would the result be? We are told lower
signal to noise ratio and a noticeable decrease in sound quality.

Really? Try this listening test where the bit resolution is reduced.
See IF you can determine which track is 24-bit, etc . Resoutiom
is varied between 24-bits and as low as 9-bits.

Test here:

http://www.ethanwiner.com/BitsTest.html

I'll reveal which is which later. You might follow on here with your results
first. This demonstration first appeared on the Yahoo VTPO group
list.

A very fascinating forum on You Tube about all of this is here:

http://www.youtube.com/watch?v=BYTlN6wjcvQ

This Forum runs for nearly an hour, but you might find
the time spent worth while. Note, the first fellow you see and
here in this Forum is the guy who put together the above
BitsTest piece. For the very interesting bit tests, click ahead
in the Forum to the 45 min 10 second point where a similar demo
to the first above is run, but going on down to even 4-bit and
lower resolution.

Note that the first Allen digital generation, their MOS-1 instruments,
used only 8-bit. But also used only a single pipe sample per rank,
and the circuitry only played out 1/2 of the signal wave form; it
was then inverted 180 degrees and played again to form the
complete sample wave! This was caused by the extreme high cost of
digital storage back in 1971. But still, the sound was reasonable,
so Allen sold a lot of MOS-1 organs over the following 10 years.

We ARE very fortunate to now live while this technology has come
so far, and is still advancing!
Jim Reid
Offline
User avatar

Grant_Youngman

Member

  • Posts: 1035
  • Joined: Sat Jan 20, 2007 7:50 pm
  • Location: Savannah, Ga

Re: Selecting Bit Resolution at Organ Loading

PostSat Mar 13, 2010 9:00 pm

Jim Reid wrote:Each of us, when loading a new organ sample set will select 24-bit
unless we are limited in our computer RAM. However, is that
necessary or suppose we used lower resolution, say 16 bit or
even 14-bit. What would the result be? We are told lower
signal to noise ratio and a noticeable decrease in sound quality.

Really? Try this listening test where the bit resolution is reduced.
See IF you can determine which track is 24-bit, etc .


I haven't tried the test, but I can tell you that the difference (using high quality headphones) of the Caen Surround with the main samples loaded at 24 bit vs. 16 bit, which I did yesterday, is stunning. I can't say I thought the set loaded at 16-bit wasn't up to snuff -- not at all. But, that one time A/B experience was like going from colorized to COLOR :-)

I have the luxury of having enough RAM to load just about everything at 24 bits, save for the occasional limitation driven by the OS X VM management strategy. Thinking about putting Windows (7??) back up in Bootcamp, just for the Caen, and maybe the Hinsz and Zwolle.
Grant
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Re: Selecting Bit Resolution at Organ Loading

PostSat Mar 13, 2010 9:09 pm

Same bit reduction here with orchestra as that demo in the AES Forum:

http://www.ethanwiner.com/aes/bit_reduction.wav

I can just begin to hear the noise coming up at about
the 75% of the way through playing time. Or at a bit resolution
of below about 9 bits and on lower. How for you?
Jim Reid
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Re: Selecting Bit Resolution at Organ Loading

PostSat Mar 13, 2010 9:11 pm

Do try the test, Grant.
Jim Reid
Offline
User avatar

toplayer2

Member

  • Posts: 1055
  • Joined: Mon Jul 23, 2007 3:15 pm
  • Location: Michigan, USA

Re: Selecting Bit Resolution at Organ Loading

PostSat Mar 13, 2010 10:08 pm

After spending one hour watching the Ethan Winer video that Jim posted, the viewer will probably be much smarter and a lot more humble than before.

This discussion was kicked off by my intentionally provocative topic on the VTPO forum:

http://launch.groups.yahoo.com/group/VTPO/message/9340

Please forgive the following personal diatribe. For several years while working to pay tuition at one of the better US music schools at the University of North Texas, I managed a very high end audio store in North Dallas by the rather odd name "The Hungry Ear". We carried the best of the best lines of the day such as Mark Levinson, Audio Research, Magneplanar, Quad, B&W, etc. etc. I always marveled at the pomposity of the "golden eared" clientele who really believed they could hear a "huge" difference with "monstrous" speaker cables, or tube vs "solid state", or any number of "Audiophile" products. We were in business to make money, so we weren't about to argue.

If you want to really make a difference in the quality of sound, invest in good speakers (anyone notice the Mackie HR824s near the end of the video) and listening room treatment.

Joe Hardy
Offline
User avatar

mdyde

Moderator

  • Posts: 10845
  • Joined: Fri Mar 14, 2003 2:19 pm
  • Location: UK

Re: Selecting Bit Resolution at Organ Loading

PostSun Mar 14, 2010 8:08 am

Hello Jim & co,

I haven't had time to watch the video, I'm afraid (so please forgive me if this isn't completely relevant!).

Although there might not be much of an audible difference between 16 and 24-bit when playing back a recorded piece of music, there's a very significant advantage to higher bit resolutions for Hauptwerk's purposes:

The bit resolution determines the noise floor, effectively because the signal has to rounded to the nearest bit/integer, termed quantisation (there are techniques that help to reduce that effect, e.g. noise-shaped dithering, but none can completely eliminate it). When listening to a piece of recorded music you only have one recording playing, hence the noise floor for a 16-bit recording is still extremely low and not usually perceptible. However, with Hauptwerk you may be playing 8000 recordings (pipe samples) simultaneously, so that background noise is multiplied several thousands of times.

Also with wet sample sets the release samples inevitably have a very low signal level towards the ends. So if you play a large chord on a large registration on a wet sample set you will hear a *lot* more hiss/noise in the release tails (regardless of the quality of the noise reduction used by the sample set producer) if loaded in 16-bit than if you load in 24-bit. Try it and see!

It might be more noticeable with some sample sets than others because it depends on the signal level the sample set producer chose for the samples. Some producers favour a higher signal level, which gives less hiss in release tails when loaded in 16-bit but it doesn't compress very well in memory, so memory requirements for a sample set will be higher. Others favour a lower signal level, which gives lower memory requirements but at the expense of more hiss in release samples at 16-bit.

However, in either case loading in 24-bit (or 20-bit is often sufficient) should eliminate much of the hiss.

So if you're evaluating sample sets for quality it's essential that you load them in 24-bit for a fair comparison.
Best regards, Martin.
Hauptwerk software designer/developer, Milan Digital Audio.

[Please use email or the Contact page if you need to contact us privately, rather than private forum messages.]
Offline

GrahamH

Member

  • Posts: 546
  • Joined: Tue Sep 09, 2008 7:39 am
  • Location: Near Manchester, England

Re: Selecting Bit Resolution at Organ Loading

PostSun Mar 14, 2010 1:09 pm

Another factor which needs to be considered is that of hearing acuity. Several years have elapsed since I first realised I could no longer hear crickets chirping in the hedgerows, so the likelihood of my being able to hear the difference between 16-bit and 24-bit samples is very small - but that does not therefore mean there is no difference. I am blessed with friends who are considerably younger than me, and artefacts that I simply cannot hear, particularly in the attack and release transients of pipe organ samples, are so obvious to them that they can't believe I can't hear them - just as my sons couldn't believe that I couldn't hear the insects that both they and I could see quite plainly.

Colin Pykett published a good article on this topic not so long ago:
"Age-related hearing loss and organs" http://www.pykett.org.uk/arhlandob.htm

Then again, there is the matter of audio equipment. A couple of years ago, I couldn't understand why a friend was so scathing about some samples I had just finished editing and looping - until I removed my £7.50 Tesco headphones and auditioned them through his £130 Sennheisers. I promptly ordered a pair of Sennheisers.

So what I am implying is this - an experiment that involves a male who is past normal retirement age listening to something broadcast on YouTube is unlikely to "prove" anything!
I shall simply take Martin's word for it and load my samples at as high a bit rate as my 8 GB of RAM will allow - and IF the price of 4 GB RAM sticks ever comes down to a sensible level, I shall upgrade to 16 GB.

Graham
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Re: Selecting Bit Resolution at Organ Loading

PostSun Mar 14, 2010 2:10 pm

Excellent comments by both Martin and Graham.

Listening demos of signal to noise ratio, that is noise along with
the music are also illustrated in the 58 minute video.

To hear the noise level demonstrations, just play the YouTube
piece; start at minute 32, and listen only through minute 34,
if you chose not to hear all 58 minutes of the demonstrations
and explanations.

BTW, yesterday I loaded my new Pipeloops Schyven sample set at 24 bit.
I do have 12 GB of RAM installed and nearly exactly 50% of my
hard drive is loaded with HW and sample sets. Only sample set I had
to select lower options for loading so far has been the big PAB set.
Jim Reid
Offline
User avatar

johnh

Member

  • Posts: 696
  • Joined: Sat Mar 15, 2003 7:51 pm
  • Location: Monterey Bay Area of California

Re: Selecting Bit Resolution at Organ Loading

PostSun Mar 14, 2010 2:47 pm

GrahamH wrote:So what I am implying is this - an experiment that involves a male who is past normal retirement age listening to something broadcast on YouTube is unlikely to "prove" anything!


Just an aside, the clips in the YouTube video are offerered for download in their original form and this is mentioned in the video along with the problems of YouTube audio degradation.
Offline
User avatar

RichardW

Member

  • Posts: 827
  • Joined: Sun Oct 28, 2007 10:16 am
  • Location: UK

Re: Selecting Bit Resolution at Organ Loading

PostMon Mar 15, 2010 3:53 pm

... "huge" difference with "monstrous" speaker cables, ...


Joe, many years ago I had friends who were seriously into hifi. They spent hours listening to each others equipment and had exactly those golden ears you mentioned. I could sometimes tell the difference when they played things to me but not always. I thought they were making it all up as well - particularly with the speaker cables. I persisted in this belief until one day I changed the cables on my keyboard setup. I replaced the thin cables wit some thicker "speaker" cables (still cheap ones).

The difference to me was amazing when playing the keyboard. As far as I could tell, the sound of a CD remained the same through the same system.

Trying to rationalise this, I reasoned that when I played the keyboard the sound was part of a feedback loop. It controlled how hard I was playing certain keys - without the necessary harmonics the sound was flat. I can tell whether I am playing my piano with the lid open, shut or stacked with music, for instance.

When I listen to hifi, however, I suspect that the sound quality is only of limited consideration. I tend to listen to the music. Whether it is a historical recording of Rachmaninov or a new one of Lang Lang, the interesting part is how the music is played and not what it sounds like.

Regards,
Richard
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Re: Selecting Bit Resolution at Organ Loading

PostMon Mar 15, 2010 5:47 pm

Performed an experiment today following my necessary practice time.
This test to see if I could hear more noise in an organ loaded at 14 bits
rather than 24 bits. Had been using the OrganART Media Vollenhoven for
practice.

So checked the loaded RAM at 24 bit loading. Total, 5 GB.
Unloaded the organ; RAM remaining loaded with Win Vista 64
and Hauptwerk (no organ loaded) 2.20 GB. So at 24 bits,
the Vollenhoven needs to load 2.8 GB of pipe sample data into RAM.

Re-loaded with each of the 25 stops loaded at only 14 bit resolution.
RAM load at 14 bits, 3.7 GB, or for this organ, 1.5 GB of pipe sample
data is added to the RAM; or only about half the RAM needed at 24 bits.

With an individual stop drawn, or every stop drawn and the manuals and
pedal coupled, I could hear nor perceive no increase in playback noise
level, nor any deterioration of sound quality.

Results might be different for other sample sets, but this set by
Herr Prof. Maier has very low/no noise at any bit loading level,
at least none I could hear.
Jim Reid
Offline
User avatar

vidarf

Member

  • Posts: 429
  • Joined: Wed Oct 29, 2008 7:05 pm
  • Location: Narvik, Norway

Re: Selecting Bit Resolution at Organ Loading

PostMon Mar 15, 2010 7:43 pm

I was utterly dissapointed by the Metz sample set - it sounded awful. Until I loaded it in 24 bit, not 16 bit. :)

So perhaps some sample sets are more sensitive to 16 vs. 24 bit?
Offline
User avatar

toplayer2

Member

  • Posts: 1055
  • Joined: Mon Jul 23, 2007 3:15 pm
  • Location: Michigan, USA

Re: Selecting Bit Resolution at Organ Loading

PostMon Mar 15, 2010 11:27 pm

As the AES panelists in the YouTube video demonstrate most convincingly, our brains are wired to hear what we expect to hear, "expectation bias". Only blind tests are meaningful.

Joe H.
Offline
User avatar

Jim Reid

Member

  • Posts: 1764
  • Joined: Mon May 17, 2004 7:50 pm
  • Location: Kauai, Hawaii

Re: Selecting Bit Resolution at Organ Loading

PostTue Mar 16, 2010 12:28 am

And the correct answers for the BitsTest which began this thread are:

File 1 = dithered to 16 bits
File 2 = truncated to 16 bits
File 3 = truncated to 13 bits
File 4 = truncated to 11 bits
File 5 = truncated to 9 bits

So the HW option of loading using "only" 14 bits
might not be awful after all. I found that to be the
case with the Vollenhoven organ, as reported in an earlier
post here.

Just might do the experiment tomorrow using the Metz sample set
or the Bovenkerk Hinsz. Both have lots of acoustic reverb to
cause very high polyphony.
Jim Reid
Offline
User avatar

CHRIS 037

Member

  • Posts: 1006
  • Joined: Tue Dec 30, 2003 10:01 pm
  • Location: Spokane, WA, USA

Re: Selecting Bit Resolution at Organ Loading

PostTue Mar 16, 2010 12:56 am

Hi Martin,

What you have stated above suggests that the cleanest organ sound should come from a dry set that has some sort of reverb added to it. Is that fair to say? That is, the dry pipes would not have much, if any, background noise that comes through at the end of the sample. There wouldn't be the build up of noise from many pipes playing together. A good reverb wouldn't add very much noise, would it? Perhaps 16-bit would do in this case as well as 24-bit?

Leo Chris.
Next

Return to General discussion

Who is online

Users browsing this forum: anco111, bourdon, jpm, smfrank and 6 guests