[linux-audio-user] Reply regarding our possible future tiMIDIty patchset.

New Message Reply About this list Date view Thread view Subject view Author view Other groups

Subject: [linux-audio-user] Reply regarding our possible future tiMIDIty patchset.
From: Roel de Wit (roel_AT_utopiasd.com)
Date: Fri Nov 29 2002 - 19:47:14 EET


Since I receive this list in a 'Digest' format I'll reply to all questions
(as far as they have not been replied by by others correctly) in one email.

- Regarding from 'scrap vs scratch typo': Sorry my native language is not
English and sometimes I make some mistakes sometimes.
- Regarding: how this patchset would sound when comparing to Fluid (assuming
it's played back under tiMIDIty): Fluid is a bank in SF2 format which was
made to sound a true (realistic) as possible with all it's cons and pros.
The main problem with 'realistic' patchsets is that it's hard to make all
instruments fit together correctly. Next to often existing tuning problems
they often just don't fit together. It's true that with some MIDIs it might
some great but with others it might sound horrible or just lacking depth.
All our past Utopia patchsets have been aiming at full sound and balance.
This means (that whenever we do the job right) that 99.x% of the well
sequenced MIDI files (in other words those who use the Roland Sound Canvas
as a refference) will playback near perfectly. As a result of this our
patchsets don't sound as realistic as a soundfont as Fluid might do but it
will sound (near) perfectly balanced.

Our previous products all have been in development since 1996 in some way
when we made our Utopia Soundbank for the GUS PnP soundcards (check dejanews
for some messages about that) from scratch (using .PAT files and a
home-written editor to allow stereo voices) and thus no sample in it is used
in any other soundbank/soundfont (legally). Ever since we've been
fine-tuning and tweaking this base set of work we've done up till now.. So
I'm pretty sure that the sound is unique and quite powerfull.

To get an idea how this new patchset would sound like have a look on our
website at http://www.utopiasd.com and listen to the Utopia Live!
demonstration MP3s for the general idea and to the Acoustic Piano add-on
v1.0 demonstration MP3s to get an idea how the piano that will be included
in this new patchset will sound. Note that the Utopia Live! recordings are
still from version 1.x and some instruments like the pianos are very

Note that the current patchsets are limited to 31.5 MB (and more recently 35
MB but no MP3 recordings have been made of the later versions) and since
size doesn't count too much with the tiMIDIty versions some instruments will
sound much more realistic (at least that is the plan).

Don't buy any of these products for use with tiMIDIty or another softsynth
or hardware other than the SB Live!/Audigy claiming to be compatible with
the soundfont format because it'll simply not work.

- The final version of this patchset will be released in the form of a
windows installer and a linux-perl installation script. So it should self
install up to a point. To make the downloadable size acceptable everything
will be compressed with the OggVorbis format. All samples will be 48 Khz 16
bit. My experience with Linux is still a bit limited (except for some
private server maintenance) so the initial beta versions will only come in
the form of a windows installer I fear.

- To get an idea how the current patchsets are layered download the free 2
MB demo version of Utopia Live! and open it in Vienna (under windows or
using an native SF2 editor under Linux). Note that the quality of these
samples (and looppoints) are very bad due to use of extreme compression (it
unpacks to about 36 MB). Still this should give you an idea how the layering
is done in our products.. E.g. our Acoustic Piano add-on (no demo available)
is in 48 Khz / 16 bit with 28 velocity layers. Layering use doesn't apply to
tiMIDIty since we are currently not happy with SF2 results.. Some examples:
Filters don't behave like they should, when playing a stereo voice you get a
phasing effect when timing is critical, early note-cuttoffs etc.. I'll have
a look at the tiMIDIty at sourceforge to see how things have improved and
will take that into consideration when determining whether to use .PAT or
SF2. If we use .PAT files make everything in Vienna in one use SF2 file and
sample it from an Audigy 2 to create a (huge) single layer set of
instruments. As a result, next to missing an stereo image the sound should
be equal or better (bigger instruments) than our current Utopia patchsets.
Regarding how they've been recorded or processed (e.g. our StereoID
percussion) I won't give any details. I know the SF2 format by head but I
don't know how tiMIDIty's implementation is. So far it seems terrible.

- Regarding .PAT implementation.. I guess that tremelo/vibrato is correctly
implemented ? From what I remember from back in 1997 it seemed to perform
nearly perfectly when comparing to the GUS Classic.

Roel / Utopia Sound Division

New Message Reply About this list Date view Thread view Subject view Author view Other groups

This archive was generated by hypermail 2b28 : Fri Nov 29 2002 - 19:46:53 EET