Discussion:
[jOrgan-user] SF2 Limitation
John Beach
2017-06-21 22:07:02 UTC
Permalink
While I really believe that there must be an “objective reality” regarding the comparison of quality between recorded pipes as “real samples” and synthesized sounds, the following excerpt
was copied from the jOrgan Tutorial and does state that “This is a matter of personal opinion.”
That said, it is hard to believe that it has been six and a half years since the tutorials were written for jOrgan. I thought, however, that you would find this section particularly interesting
in view of our ongoing discussion on the topic. In the second question below, it is valuable to note that the same answers to the question which was considered 6 and a half years ago
remain the answers to the question today!

John Beach



Question: Why are jOrgan dispositions distributed with soundfonts rather than samples?

Answer: The world of jOrgan is bigger than just the dispositions that have been created for fluidsynth. Soundfonts are simply a way of bundling wave samples together for a sample player—in jOrgan’s case, typically the Fluidsynth soundfont player that is included as an extension for jOrgan. So, soundfonts can be synthesised sounds or real samples. Typically soundfont players (like fluidsynth) have a harder time with huge samples, so smaller sample sections are used, and wave files are also typically shared accross notes. [Graham G]
Question: Although soundfonts are much smaller, wouldn’t real samples give a higher-quality sound? Answer: This is a matter of personal opinion. In my opinion, real samples sound better than synthesized ones, although there are some wonderful exceptions. Longer samples sound better than shorter ones. A sample for each note sounds better than a sample shared across notes. Given the nature of computers and the requirements of fluidsynth and other sample players we have to make compromises for playability and share-ability,therefore stretching one wave file across 4 notes, for example, is a reasonable compromise. [Graham G] A soundfont file is simply a packaging structure for .wav files. The quality of the samples in the soundfont package is as good as the quality of the sample files used to create the soundfont. The issue of quality is bit density of the samples and degree of lossy compression, regardless of whether they are individual files or packaged into a soundfont structure. Individual samples files versus soundfonts is not a quality of sound dimension. [Lynn]
Question: Is jOrgan capable of playing samples? Answer: jOrgan is an organ MIDI relay that has three internal Sound Extensions—fluidysnth, creative, and linuxsampler —plus the Generic Sound element that sends MIDI data out to any MIDI receiving sound source. So jOrgan is capable of playing samples (and capable of playing real pipes too!), so long as the sample player reacts to MIDI. [Graham G] jOrgan does not play samples or soundfonts or, in fact, make any sound at all. jOrgan is a MIDI-based relay. Its input consists of MIDI data streams and mouse clicks on a console screen. Its output consists of one or more MIDI data streams...period. [Lynn]
Soundfonts and Sampled Sounds
Revised 11/6/2010 Revised 11/6/2010

Question: If jOrgan is capable of playing samples, is there any need for software such as GrandOrgue or Hauptwerk?

Answer: GrandOrgue and Hauptwerk are sample player software applications that specialize in low latency wave file reproduction. In my opinion, neither application has the versatility and ease of disposition creation that jOrgan has. Their wave reproduction is, however, generally better than the jOrgan extensions available at this time. My preference would be to have the low latency audio engine of GrandOrgue made available as a jOrgan extension. Until that time I use jOrgan with fluidsynth in Linux—which has much lower latencies than Windows—and jOrgan controlling GrandOrgue in Windows. [Graham G] Hauptwerk does indeed produce sound. If you want one software package that both operates as a MIDI-based relay and produces sound, Hauptwerk is probably the best—and the most expensive. As stated above, jOrgan does not produce sound, so in order to have its MIDI output rendered to audio you would need an auxiliary sound rendering software package—such as Fluidsynth, SFZ, GSO3, or any other MIDI-driven sample player—or a MIDI-driven hardware device like a Sound Blaster audio card or an external MIDI-driven sound module. Because Hauptwerk is more involved in the actual rendering of the sound, it is able to produce some unique articulations that are presently not possible with jOrgan and a separate MIDI-driven sound module, whether hardware or software. [Lynn]
Question: Which is better for soundfonts: sampled sounds or synthesized sounds? Answer: Soundfonts can be synthesized sounds or real sampled sounds. [GrahamG] SoundFonts offer a unique entrance to the world of sound creation, while the available wave sample sets are nothing more or less than the result of a carefully placed mike set and cassette recorder. Thus such wave sets provide an artificial imitation of the creative impulse of real organ builders with no provision for personal creativity. [Bernd] It all depends whether your main interest is to simulate a particular real organ as closely as possible, or simply to create a pleasant sound. [Roy] Sampled and Synthesized Sounds
Index.1
Index
Aaron Laws
2017-06-26 12:42:58 UTC
Permalink
Post by John Beach
While I really believe that there must be an “objective reality”
regarding the comparison of quality between recorded pipes as “real
samples” and synthesized sounds, the following excerpt
was copied from the jOrgan Tutorial and does state that “This is a matter
of personal opinion.”
That said, it is hard to believe that it has been six and a half years
since the tutorials were written for jOrgan. I thought, however, that
you would find this section particularly interesting
in view of our ongoing discussion on the topic. In the second question
below, it is valuable to note that the same answers to the question which
was considered 6 and a half years ago
remain the answers to the question today!
John Beach
Question: Why are jOrgan dispositions distributed with soundfonts rather than samples?
Answer: The world of jOrgan is bigger than just the dispositions that have
been created for fluidsynth. Soundfonts are simply a way of bundling wave
samples together for a sample player—in jOrgan’s case, typically the
Fluidsynth soundfont player that is included as an extension for jOrgan.
So, soundfonts can be synthesised sounds or real samples. Typically
soundfont players (like fluidsynth) have a harder time with huge samples,
so smaller sample sections are used, and wave files are also typically
shared accross notes. [Graham G]
Question: Although soundfonts are much smaller, wouldn’t real samples give
a higher-quality sound?
It seems there is some conflation between soundfonts and synthesized
sounds. The answer to this question (just above), for instance would be,
"Soundfonts commonly consist of real samples.", right? If the question
were, "Although synthesized soundfonts are much smaller...", it would be
legitimate.

In Christ,
Aaron Laws
John Beach
2017-06-26 21:54:00 UTC
Permalink
Aaron said, “It seems there is some conflation between soundfonts and synthesized sounds. The answer to this question (just above), for instance would be, "Soundfonts commonly consist of real samples.", right? If the question were, "Although synthesized soundfonts are much smaller...", it would be legitimate.”


Actually, the first soundfonts which were packaged with the Creative Labs soundcards were General Midi and were all recorded , real, orchestral instruments, sampled in several octaves. There were also special soundfonts which included collections of ethnic instruments from around the world. I have belonged to several organ forums as well as the former theatre organ soundfont user’s group, on yahoogroups.com, which is now defunct. That group used real, recorded, pipe organ soundfonts, but switched over to gigasampler format, which was so successful, that the soundfont user’s group went dead.
I have made and used synthesized sounds because they were what was available. I live in an area where pipe organs are a rarity.

Apparently, gigasampler is not compatible with jOrgan because no one has ever suggested using it.

Regarding the size of soundfonts, I don’t know what the size of a soundfont with real, recorded pipe sounds would have to be. On average, what the wave file size per note would have to be and how many
per octave, per stop, would determine the size of the soundfont.

With synthesized sounds, I have had soundfonts which were 1/10th second long and 6 soundfonts containing 400 stops were less than 17 Megabytes. I have since resynthesized all stops and
each wave file is 4 seconds long. The same soundfonts total 93 Megabytes. If I had to give an opinion on whether there is a noticeable difference in sound quality. Honestly, I can’t say.

The stops I find most difficult to replicate are diapasons. They are the ones with which I am always least satisfied. A principal is fundamental and it must be good. I find that synthesizing
principals using Hammond Drawbar registration recipes produces harshness—”cornet” timbers---which I don’t hear in real pipes. I think the use of frequency cutoff with those same synthesized
principals reduces or eliminates the harshness. A good pipe organ can be listened to for hours without getting obnoxious or monotonous. The best, synthesized principal leaves out the upper
harmonics above 2’ and sounds like a full flute rather than a string. This is a matter of personal taste, of course.

John Beach
Roy Radford
2017-06-26 22:07:08 UTC
Permalink
Hi, John,

I'm on shaky ground here but I seem to recall the samples
generated by Giga can be read by Qsampler in Linux. Presumably it could be
made the output of a Generic Sound in jOrgan???



Have fun,

Roy.
Post by John Beach
Aaron said, “It seems there is some conflation between soundfonts and
synthesized sounds. The answer to this question (just above), for instance
would be, "Soundfonts commonly consist of real samples.", right? If the
question were, "Although synthesized soundfonts are much smaller...", it
would be legitimate.”
Actually, the first soundfonts which were packaged with the Creative Labs
soundcards were General Midi and were all recorded , real, orchestral
instruments, sampled in several octaves. There were also special
soundfonts which included collections of ethnic instruments from around
the world. I have belonged to several organ forums as well as the former
theatre organ soundfont user’s group, on yahoogroups.com, which is now
defunct. That group used real, recorded, pipe organ soundfonts, but
switched over to gigasampler format, which was so successful, that the
soundfont user’s group went dead.
I have made and used synthesized sounds because they were what was
available. I live in an area where pipe organs are a rarity.
Apparently, gigasampler is not compatible with jOrgan because no one has
ever suggested using it.
Regarding the size of soundfonts, I don’t know what the size of a
soundfont with real, recorded pipe sounds would have to be. On average,
what the wave file size per note would have to be and how many
per octave, per stop, would determine the size of the soundfont.
With synthesized sounds, I have had soundfonts which were 1/10th second
long and 6 soundfonts containing 400 stops were less than 17 Megabytes. I
have since resynthesized all stops and
each wave file is 4 seconds long. The same soundfonts total 93
Megabytes. If I had to give an opinion on whether there is a noticeable
difference in sound quality. Honestly, I can’t say.
The stops I find most difficult to replicate are diapasons. They are the
ones with which I am always least satisfied. A principal is fundamental
and it must be good. I find that synthesizing
principals using Hammond Drawbar registration recipes produces
harshness—”cornet” timbers---which I don’t hear in real pipes. I think the
use of frequency cutoff with those same synthesized
principals reduces or eliminates the harshness. A good pipe organ can be
listened to for hours without getting obnoxious or monotonous. The best,
synthesized principal leaves out the upper
harmonics above 2’ and sounds like a full flute rather than a string.
This is a matter of personal taste, of course.
John Beach
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
jOrgan-user mailing list
https://lists.sourceforge.net/lists/listinfo/jorgan-user
Thomas Beck
2017-06-26 22:22:42 UTC
Permalink
Giga format can be played by linuxsampler, and there is already an existing
disposition for it. Please see Graham Goode's Barton organ.

I have mentioned this repeatedly.
Post by Roy Radford
Hi, John,
I'm on shaky ground here but I seem to recall the
samples generated by Giga can be read by Qsampler in Linux. Presumably it
could be made the output of a Generic Sound in jOrgan???
Have fun,
Roy.
Post by John Beach
Aaron said, “It seems there is some conflation between soundfonts and
synthesized sounds. The answer to this question (just above), for instance
would be, "Soundfonts commonly consist of real samples.", right? If the
question were, "Although synthesized soundfonts are much smaller...", it
would be legitimate.”
Actually, the first soundfonts which were packaged with the Creative Labs
soundcards were General Midi and were all recorded , real, orchestral
instruments, sampled in several octaves. There were also special
soundfonts which included collections of ethnic instruments from around
the world. I have belonged to several organ forums as well as the former
theatre organ soundfont user’s group, on yahoogroups.com, which is now
defunct. That group used real, recorded, pipe organ soundfonts, but
switched over to gigasampler format, which was so successful, that the
soundfont user’s group went dead.
I have made and used synthesized sounds because they were what was
available. I live in an area where pipe organs are a rarity.
Apparently, gigasampler is not compatible with jOrgan because no one has
ever suggested using it.
Regarding the size of soundfonts, I don’t know what the size of a
soundfont with real, recorded pipe sounds would have to be. On average,
what the wave file size per note would have to be and how many
per octave, per stop, would determine the size of the soundfont.
With synthesized sounds, I have had soundfonts which were 1/10th second
long and 6 soundfonts containing 400 stops were less than 17 Megabytes. I
have since resynthesized all stops and
each wave file is 4 seconds long. The same soundfonts total 93
Megabytes. If I had to give an opinion on whether there is a noticeable
difference in sound quality. Honestly, I can’t say.
The stops I find most difficult to replicate are diapasons. They are the
ones with which I am always least satisfied. A principal is fundamental
and it must be good. I find that synthesizing
principals using Hammond Drawbar registration recipes produces
harshness—”cornet” timbers---which I don’t hear in real pipes. I think the
use of frequency cutoff with those same synthesized
principals reduces or eliminates the harshness. A good pipe organ can be
listened to for hours without getting obnoxious or monotonous. The best,
synthesized principal leaves out the upper
harmonics above 2’ and sounds like a full flute rather than a string.
This is a matter of personal taste, of course.
John Beach
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
jOrgan-user mailing list
https://lists.sourceforge.net/lists/listinfo/jorgan-user
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
jOrgan-user mailing list
https://lists.sourceforge.net/lists/listinfo/jorgan-user
John Beach
2017-06-27 02:43:12 UTC
Permalink
Actually, I know nothing about gigasampler and, although I have an old computer with Linux on it, I don’t use it. I am still doing occasional reading about the file system used in Linux
as I find it a bit intimidating after the simplicity of Windows.

Anyway, about gigasampler, eight or ten years ago, the theatre-organ soundfont user’s group was able to get real pipe samples and a fellow named Joe Hardy worked tirelessly to get
them all into gigasampler. I forget how many ranks there were, but it took a long time. The were all posted on Allman’s music website for download. I don’t even know if that still exists, as I think the owner passed on several years ago. That was about the time that Miditzer was developed also with Bruce Miles’ soundfonts. Bruce’s were all synthesized, but they was excellent
quality. As a result of the switch to gigasampler, the soundfont user’s group died.

What are the requirements for “Generic Sound” in jOrgan? It must require some kind of formatted Wave Files.

For those people who do use soundfonts and there are a lot, I don’t know what they use besides Fluidsynth and some of the VST soundfont players that are available. People must be
fairly satisfied with what they are able to do since there seems to be no concerted effort to develop a soundfont sound engine, such as Fluidsynth with increased capabilities.
Since all user sound production must be wave files, there must be other sound engines which could be used with jOrgan.

John Beach
Graham Goode
2017-06-27 05:36:12 UTC
Permalink
What are the requirements for “Generic Sound” in jOrgan? It must require
some kind of formatted Wave Files.
Hi John,

No. The Generic Sound element in jOrgan is in fact the original sound
element that purely outputs MIDI messages to a MIDI port. Creative,
LinuxSampler, and Fluidsynth sound elements ADD specific program to
program communication from jOrgan to the external sample player.

So the Generic Sound element can literally connect jOrgan to ANY MIDI
based program, instrument, software, or hardware.

GG
Roy Radford
2017-06-27 07:54:24 UTC
Permalink
...As indeed I do with my Edirol units, simply set the Output property of a
generic sound to one of the Edirols.



Have fun,

Roy.
Post by Graham Goode
Post by John Beach
What are the requirements for “Generic Sound” in jOrgan? It must require
some kind of formatted Wave Files.
Hi John,
No. The Generic Sound element in jOrgan is in fact the original sound
element that purely outputs MIDI messages to a MIDI port. Creative,
LinuxSampler, and Fluidsynth sound elements ADD specific program to
program communication from jOrgan to the external sample player.
So the Generic Sound element can literally connect jOrgan to ANY MIDI
based program, instrument, software, or hardware.
GG
------------------------------------------------------------
------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
jOrgan-user mailing list
https://lists.sourceforge.net/lists/listinfo/jorgan-user
John Beach
2017-06-27 20:10:55 UTC
Permalink
Interestingly, Generic Sound in jOrgan works just like Creative Sound if one loads the soundfonts with the Soundfont Bank Manager and then selects the output device as Creative Synth A or B, under Generic Sound. I am still having trouble with the ha20x22k.sys file in both Creative Sound and Generic Sound. I have searched for a solution to the problem, but, apparently, there is
none.

John B.

Graham Goode
2017-06-27 05:41:27 UTC
Permalink
Post by John Beach
Actually, I know nothing about gigasampler and, although I have an old
computer with Linux on it, I don’t use it. I am still doing occasional
reading about the file system used in Linux
as I find it a bit intimidating after the simplicity of Windows.
Hi John,

LinuxSampler loads Gigasampler files, and runs on Windows, Linux, and
OS-X (just like jOrgan). The jOrgan LinuxSampler element loads a
script that loads all the files and sets up all the ports and sound
connections.

Gigastudio originally hosted the Hardy GIG samples, and yes - they
were controlled and played using jOrgan and the Generic Sound element.
I still have all the GIG files. The Allman site can still be viewed
(http://www.webring.org/l/rd?ring=theatreorgan;id=4;url=http%3A%2F%2Fallmanmusic%2Ehomestead%2Ecom%2Fvirtualorgan%2Ehtml)
but download links probably no longer work.

GG
Loading...