[linux-audio-user] Pianotest1 recording as requested by Mark and answers to questions by others.
22 Jan
2003
22 Jan
'03
6:27 p.m.
Hi,
I've uploaded the MP3 recording of this MIDI file to
http://project-portability.foad.nl/mp3/pianotest1.mp3
Note though that it was sampled from Grandest Piano beta v0.1 and thus will
not neccesarily represent the v1.0 release of the piano. Next week two new
sample layers will be added and the amount of velocity layers will be
increased to 127 (which will fully replace the current not so perfect ones).
Also the release values of each note (or layer.. not sure yet) will be
adapted.. I'll provide the website with a recording from v1.0 when it
arrives so I prefer to keep this recording low profile.
The recording was made without any reverb active or any post-processing so
that it shows the actual samples the best way.
Regarding 16 bit vs 24 bit.. I personally don't see the use of 24 bit
samples too much yet. When mixing them is concerned the more bits the
better.
Regading looping etc.. The samples are all unlooped with a maximum of 16
seconds (which is often used).
Yeah, but what is a tiMIDIty version, explicitly? If
you want to
please the Linux community, just send us 88 (more?) samples, one for
each key, and chuck the looping. Just let them all ring out till
silence. Better yet, 4-5 samples for each key at different striking
pressures would be great. Don't worry about how or what software will
sort this all out, we'll figure out something.
With all the sampled velocity layers and huge amount of filtered velocity
layers one soundfont won't simply work on any 'claimed' soundfont compatible
softsynth or device. E.g. LiveSynth Pro a very good SF2 softsynth can't
handle the piano correctly (yet). Our Utopia Live! soundfont (GM/GS) can't
be played back (or even loaded) on anything else than SB Live!/Audigy
hardware. It's the main drawback of 'complicated' soundfonts..
I have all the velocity layers sampled at 88 notes at my harddisk but the
Audigy is stuck at around 330 MB (maximum loadable soundfont size under
Windows 2000/XP). That's why I choose for around 25-28 samples per velocity
layer. The size of the full 88 key sampled piano with 5 sampled velocity
layers would be at least 800 MB assuming I'd stick to 48 Khz stereo
samples.. It might be done in the future, but not now, unless there is a
demand for it. I don't think the difference would be noticable too much..
I'm willing to add around 3 samples to each layer if the beta testers think
this is needed..
A good idea would be to double all of these samples,
one recorded
with the sustain pedal open, and one closed. This would allow the
creation of the interactive effect between the strings which is
so noticeably lost in most sample sets. This makes the set too
large you say? Not even. I wouldn't sweat if it was a couple of
That would mean that there would be 3 instruments (MIDI wise) which a
composer would have to switch between.. I'm not sure if this is usefull.
GIG even. Hey, it's only one DVD. Lets start
putting today's machines'
capabilities to good use. In fact, these samples should be AIFF, WAV,
or raw, 96k/24bit, mastered digitally with mics and equipment that can
register frequencies approaching the nyquist. That way, when we lower
notes an octave or two for effect, we will still have high end sizzle.
Try to explain to me what the use is of 96 Khz samples ? It's just marketing
nothing more.. 48 Khz makes it possible to represent up to 24 Khz of sampled
frequencies, far about what you can hear. More important is that the device
or softsynth used mixes at higher bitrates and that the DACs are of high
quality (unless you render directly to a .WAV/whatever file).
Yeah, make that, and I would buy it. $20.00 - $30.00
say. As long as
there were no restrictions through licensing, other than, of course, on
the resale of the package as a sample set.
No restrictions, other than you are not allowed to distribute any part of
what's in the package that will be sold.
---
Roel / Utopia Sound Division
http://www.utopiasd.com
23 Jan
23 Jan
12:25 a.m.
On Wed, Jan 22, 2003 at 11:23:55PM +0100, Roel de Wit wrote:
Note though that it was sampled from Grandest Piano
beta v0.1 and thus will
not neccesarily represent the v1.0 release of the piano. Next week two new
sample layers will be added and the amount of velocity layers will be
increased to 127 (which will fully replace the current not so perfect ones).
I'm not really a master of piano samples and synths, but let me say -
the test you've posted sounds really good. I like the way the samples
sound - they have a very unprocessed sound. It lets the piano come
through. If I were using the sample set, it would allow me to apply the
effects I needed without impeding that.
Nice work!
--
Ross Vandegrift
ross(a)willow.seitz.com
A Pope has a Water Cannon. It is a Water Cannon.
He fires Holy-Water from it. It is a Holy-Water Cannon.
He Blesses it. It is a Holy Holy-Water Cannon.
He Blesses the Hell out of it. It is a Wholly Holy Holy-Water Cannon.
He has it pierced. It is a Holey Wholly Holy Holy-Water Cannon.
He makes it official. It is a Canon Holey Wholly Holy Holy-Water Cannon.
Batman and Robin arrive. He shoots them.
4:20 a.m.
On Wed, 22 Jan 2003 23:23:55 +0100
"Roel de Wit" <roel(a)utopiasd.com> wrote:
Try to explain to me what the use is of 96 Khz samples
? It's just
marketing nothing more.. 48 Khz makes it possible to represent up to
24 Khz of sampled frequencies, far about what you can hear.
Ok, I'll have a go :-)
If you sample at 48Khz you need to filter out everything above 24Khz in
order to avoid alias noise. First of all this is not *really* possibkle,
filters don't block thay damp. Secondly a filter steep enough to let
ninimal amounts of energy through above 24Khz will be distorting the
upper part of the frequency range.Now if you sample at 96Khz the story
is the same, but not you have an extra octave to filter smoothly.
Another part of the story is that although a 48Khz sample can represent
frequencies upto 24Khz, the representation of the higher frequencies not
accurately describing the intensity.
--
peace, love & harmony
Atte
6:11 a.m.
If you sample at 48Khz you need to filter out
everything above 24Khz in
order to avoid alias noise.
Is there any advantadge to using 48KHz or higher sample rates if you
ultimately distribute your music on CD at 44.1KHz?
I can't help feeling that the 24 bit/96KHz thing is more about the upgrade
treadmill than it is about music. Suddenly you need a Pentium IV for simple
recordings...
Cheers
Daniel
10:51 a.m.
--- Daniel James <daniel(a)mondodesigno.com> wrote:
Based just on the theory for filtering above 24khz,
then yes for 96. Because you were able to filter above
before you resampled.
Remember though, that many hit songs were produced on
ADAT tape, 48000 and 16bit. :)
I've never done an a to b listening test for the
difference between 48000 and 96000 and don't intend to
do this anytime soon. Well, maybe I'll do it but just
for testing purposes, it's doubtful that there's a
good reason for me to change. Everything that
originates from me is produced in 44100 because I'd
rather not resample.
ron
If you sample
at 48Khz you need to filter out
everything above 24Khz in
order to avoid alias noise.
Is there any advantadge to using 48KHz or higher
sample rates if you
ultimately distribute your music on CD at 44.1KHz? I can't help feeling that the 24 bit/96KHz thing
is
more about the upgrade
treadmill than it is about music. Suddenly you need
a Pentium IV for simple
recordings...
Cheers
Daniel
__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com
4:15 p.m.
Atte André Jensen wrote:
On Wed, 22 Jan 2003 23:23:55 +0100
"Roel de Wit" <roel(a)utopiasd.com> wrote:
Another thing that makes 96k attractive to me, is the possibility
of capturing a sound that actually has frequencies up toward 40,000Hz,
(Have to be some mic!) and then dropping the speed of the sound down
and 8ve or two, retaining brilliance that is usually lost by slowing
down a sound.
Try to explain to me what the use is of 96 Khz
samples ? It's just
marketing nothing more.. 48 Khz makes it possible to represent up to
24 Khz of sampled frequencies, far about what you can hear.
Ok, I'll have a go :-)
If you sample at 48Khz you need to filter out everything above 24Khz in
order to avoid alias noise. First of all this is not *really* possibkle,
filters don't block thay damp. Secondly a filter steep enough to let
ninimal amounts of energy through above 24Khz will be distorting the
upper part of the frequency range.Now if you sample at 96Khz the story
is the same, but not you have an extra octave to filter smoothly.
Another part of the story is that although a 48Khz sample can represent
frequencies upto 24Khz, the representation of the higher frequencies not
accurately describing the intensity.
7977
days inactive
7978
days old
linux-audio-user@lists.linuxaudio.org
5 comments
6 participants
participants (6)
-
Atte André Jensen -
Daniel James -
R Parker -
Roel de Wit -
Ross Vandegrift -
Tobiah