BEST WAY TO RECORD SAMPLES FOR DS864

Page: < 1234 > Showing page 2 of 4
Author
Post
Essentials Only Full Version
techead
Max Output Level: -31.5 dBFS
  • Total Posts : 4353
  • Joined: 2004/01/24 08:40:20
  • Location: Macomb, IL, USA
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 19:46:29 (permalink)
ORIGINAL: b rock

I figured that the .wavs were coming in wrong anyway: out-of-tune due to the root note bug, and I felt that I had little control over the keymapping, so why not take advantage of that?


Root note bug? That didn't sound healthy on first read...explanation please?
#31
Digital Aura
Max Output Level: -24 dBFS
  • Total Posts : 5118
  • Joined: 2004/01/18 13:30:46
  • Location: Petrolia,Ontario ORIGINAL P5 FORUM JUNKIE
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 19:47:26 (permalink)
So am I out of luck then if I have only GoldWave?
Could I hook up my left & right outputs from my KORG into my 24/96 cards inputs directly and record into Goldwave as a 16-bit wav? Then after sampling the right ranges and velocities open DS864 and continue??


Official P5 FANBOY!!!
#32
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:01:42 (permalink)
ORIGINAL: Digital Aura

So am I out of luck then if I have only GoldWave?
Could I hook up my left & right outputs from my KORG into my 24/96 cards inputs directly and record into Goldwave as a 16-bit wav? Then after sampling the right ranges and velocities open DS864 and continue??


That should work. If you are able to open both P5 and GoldWave, you could try the following:

In P5, set up a MIDI pattern to play all of the notes you want to sample (e.g. C-3, E-3, G#3, C-4, E-4, G#4). You might also want to program multiple velocities for each note. Later on, you might want to try programming in sustain as well, since this will help when creating loop points.

Once your pattern is set up, use the Project5 MIDI Out plugin (from the customer downloads page) and route a MIDI cable to the input of your Trinity. If your keyboard is set up to receive incoming MIDI messages, then playing the pattern in P5 should trigger the sounds on your keyboard.

Why do this? It's a lot easier then manually playing notes while recording in GoldWave -- there is too much potential for variation in note velocity and duration (although I am not doubting your keyboard skills!).

If you can get this to work, then you can wire the audio outputs from the Trinity to your audio interface and record all of the notes in GoldWave. Just press record in GoldWave, then press play in P5.

It's been a while since I have used GoldWave, but it might have some functions for samples built in to it (e.g. root note, loop points). That would be something else to look into.

If you can't get P5 and GoldWave working at the same time, then it wouldn't hurt to record directly into GoldWave... let us know how it works for you.
#33
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:07:25 (permalink)
Every sample that I've brought in using the Add button after spanning a previous sample to a semitone under where I wanted the next one to land [see my first reply above; fifth paragraph] seemed to come in with the root note a semitone high. Perhaps is just my methods, and not a bug, after all.

I just loaded the primary sample in a blank program several times with Add, and it's come in variously at C4, C0, etc. This is contrary to my improvised method that I'd been using, which loads the primary sample from the second [.dp8] line in green using the folder icon, and the rest with Add. Those samples always seem to come in rooted a semitone high, but I'm beginning to think that the DS864 by default spans the range of notes one semitone higher and lower than the root note. I've seen widely different ranges on Add import, though, depending on the sample imported, so I'm trying to figure out a consistent method that covers all the bases. Suffice to say that you really need to pay attention to where the root note lands immediately upon import, or you'll end up with a mess on your hands. Not an easy task in the miniaturized world of the DS864.
< Message edited by b rock -- 1/2/2005 8:17:34 PM >
#34
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:10:29 (permalink)
ORIGINAL: techead

ORIGINAL: b rock

I figured that the .wavs were coming in wrong anyway: out-of-tune due to the root note bug, and I felt that I had little control over the keymapping, so why not take advantage of that?


Root note bug? That didn't sound healthy on first read...explanation please?


When you click the ADD button in the key map editor, you can then choose a wave file to add to the current program. After choosing a file, DS864 creates a split (that defines the range of notes and velocities that the sample should correspond to) and assigns it a root note value. In almost all cases, the root note it assigns by default will be wrong!

So, to change it, just click on the split, then click on the key that corresponds to the root note, or pitch of the original sample. This key will be colored red, so whenever you select a split, you will know where its root note is.

Also, when clicking around the DS864 key map editor interface, be careful not to reassign the root note by mistake! I did this several times -- the first piano library I created sounded totally bizarre until I went back and fixed all of the root notes.

Some sample programs let you define the root, upper and lower range notes, but I'm not sure if DS864 will read them in... I'll have to test that out.

As for the file format of the DP8 files that you looked into, you're right, the format of the audio data is pretty strange. I wonder if there would be a way to get the file format from Cakewalk (e.g. through a developer contract)? If so, designing a utility for creating DP8 files outside of DS864 would be a pretty fun project.
#35
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:14:48 (permalink)
ORIGINAL: b rock

Every sample that I've brought in using the Add button after spanning a previous sample to a semitone under where I wanted the next one to land [see my first reply above; fifth paragraph] seemed to come in with the root note a semitone high. Perhaps is just my methods, and not a bug, after all.

I just loaded the primary sample in a blank program several times with Add, and it's come in variously at C4, C0, etc. This is contrary to my improvised method that I'd been using, which loads the primary sample from the second [.dp8] line in green using the folder icon, and the rest with Add. Those samples always seem to come in rooted a semitone high, but I'm beginning to think that the DS864 by default spans the range of notes one semitone higher and lower than the root note. I've seen widely different ranges on Add import, though, depending on the sample imported, so I'm trying to figure out a consistent method that covers all the bases. Suffice to say that you really need to pay attention to where the root note lands immediately upon import, or you'll end up with a mess on your hands. Not an easy task in the miniaturized world of the DS864.


Yes, you are right about this. It might not be a bug, but it would be a lot easier on us if a) DS864 detected the pitch of the sample and defaulted the root note to that, or b) let us choose to default the root note to some other position (e.g. the lowest note value within the split instead of the middle value, which would help for working left-to-right).
#36
techead
Max Output Level: -31.5 dBFS
  • Total Posts : 4353
  • Joined: 2004/01/24 08:40:20
  • Location: Macomb, IL, USA
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:29:54 (permalink)
If using P5 as the MIDI sequencer to trigger the Korg Triton is a problem, I think DA could just program a MIDI sequence in the Triton's onboard sequencer to trigger itself, and use Goldwave to record the resulting audio.
#37
Digital Aura
Max Output Level: -24 dBFS
  • Total Posts : 5118
  • Joined: 2004/01/18 13:30:46
  • Location: Petrolia,Ontario ORIGINAL P5 FORUM JUNKIE
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 20:59:27 (permalink)
ORIGINAL: techead

If using P5 as the MIDI sequencer to trigger the Korg Triton is a problem, I think DA could just program a MIDI sequence in the Triton's onboard sequencer to trigger itself, and use Goldwave to record the resulting audio.


Ahhh .... good thought...although I don't have a Triton but a Trinity!!


Does it matter the note duration? Should I try and get the whole notes sound (including decay as it fades)? What are your thoughts on this?
Correct me if I'm wrong but P5 won't disengage the sound engine to allow for another "wav editor" to be open and usable!


Official P5 FANBOY!!!
#38
jardim do mar
Max Output Level: -66 dBFS
  • Total Posts : 1247
  • Joined: 2003/12/02 06:23:57
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 21:17:50 (permalink)
does the trinity use a flash card?,,,, you could use it to swap wav files between the trinity and your computer,,,,,,,,, gee, another one of my secrets is out ,, hope B rock is'nt listening,,,,,


Correct me if I'm wrong but P5 won't disengage the sound engine to allow for another "wav editor" to be open and usable!
if you're using "asio" in P5 you won't get this .....

can you set up your sound device's internal clock to digital? cause you could use soundforge's "remote record " window to record all data that is running through P5,, Oh wait ,,you don't have sound forge,,,,,

marcella
And Remember,,,,One thing at a Time.....
#39
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 21:31:36 (permalink)
I think it depends on your sound card. You might be able to use DirectSound or MME instead of your main driver in P5, and use ASIO in GoldWave. You can probably test that prior to messing around with any wires.

For the note durations, it's really going to depend upon what type of sound you are sampling. Plus, regardless of the case, getting those nice decay tails on certain instruments will result in a much larger DP8 file, so it's going to be a trade-off between quality and efficiency. After doing this a few times, I have really come to respect Squids and the SR guys, because this takes a lot of skill and know-how to get good quality sounds in as little memory as possible.

As a reference point, the Stage Piano samples were pressed for 2 measures long at a tempo of 100 BPM. The final file size for one velocity layer across four octaves was a ridiculous ~50 MB! It probably would have made more sense to sample fewer notes with shorter durations (plus sustain looping). For the higher notes, I could have truncated the samples such that they were shorter because the notes decay much faster, so that's another thing to think about.

It really does depend upon the instrument, though.

For the first time, you might want to sample a few notes from one octave using a few different sample lengths, then see what ends up working best for you with DS864. Otherwise, you might get stuck with a ton of samples that don't do what you want them to once in DS864.
#40
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 21:39:24 (permalink)
Oops! Blue beat me to it. Oh well; here it goes anyway:

I don't know if this'll work with your audio drivers (ASIO?), but you can try disabling the drivers in WinXP's Device Manager before opening up Project5. P5 will balk at this [Unable to open xxx driver ...], but it should still be able to output MIDI through the MIDIOut plugin. The driver could be re-enabled in the Device Manager for use with the wave editor. I haven't verify this procedure, because my audio & MIDI are provided by the same device [01X], but that might be worth a shot. You might also be able to switch the driver to DirectSound in P5's Options/Audio and follow the same procedure.

So you don't have an onboard sequencer in the Trinity? That might be the way to go. I miss having one with my MIDI-only controllers for situations just like this one, but I substitute some old hardware devices [drum machines, standalone sequencers; even a guitar synth with a mini-seq] instead. Unfortunately, the ol' Poly-800's seq is just too primitive.

If part of the characteristic sound is the decay portion, it'd be best to capture that as well. Some sounds can get away with using a looped portion in the sample while you hold your keys down, and impart a decay using the Amp EG in the DS864 for the release section.
gee, another one of my secrets is out ,, hope B rock is'nt listening
What was that? Did I hear something? I think that the garden is audibly sprouting ...Actually, this old dog already knew those tricks; Unfortunately, much of my hardware is on proprietary cards. I'm easing everything toward CompactFlash cards, and those USB flash drives, though. Even my portable remote audio recorder has one [CF slot]. Hmm ... now there's a thought ... has anyone used that new, cutting-edge technology to record samples? I think that they call them microphones ...
#41
techead
Max Output Level: -31.5 dBFS
  • Total Posts : 4353
  • Joined: 2004/01/24 08:40:20
  • Location: Macomb, IL, USA
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 21:53:14 (permalink)
Yeah, here it is in the help file (press F1 while having a sample loaded in the keymap editor then scroll to the bottom of the page:

"Loop points are imported when loading a WAV or AIF sample. The loop points are not editable within the DS864 and should be created in a sample editor when preparing the samples for a custom key map. The LOOP SAMPLE setting has three possible settings--OFF means the sample will only play once, FWD means the sample will repeat forwards between the loop points, and FWD/BKW means that the sample will loop forwards and backwards ("ping pong") between the loop points. If there were no loop points in the original sample, the loop points will be set at the beginning and end of the sample. "
#42
techead
Max Output Level: -31.5 dBFS
  • Total Posts : 4353
  • Joined: 2004/01/24 08:40:20
  • Location: Macomb, IL, USA
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 22:08:01 (permalink)
Every sample that I have recorded in SoundForge and imported into DS864's keymap editor has defaulted to 3-semitones width with the root note on the middle semi-tone.

I created a new sample in SoundForge and saved it with a different root note configuration. I then imported this into DS864 and it appeared to ignore the root note.

So, by default, my first wavefile imported with the "Add" button on the keymap editor always spans B3-C#4 with the root at C4, the second imported wave file spans D4-E4 with the root at D#4, the third imported wave file spans F4-G4 with the root at F#4, and so on.

If I modify the span of a sample then import another sample, the newest sample starts on the next semitone above the span of the last sample and it spans only three semitones with the root on the middle semi-tone.

Seems to me DS864 ignores the root note data stored in the MS .WAV file, defaults to a three-semitone span, and always uses the middle semitone as the root note. So far I have not been able to disprove this behavior--but I have yet to do an indepth sample setup in a single program. Mine have all been very simple splits and layers using usually only a couple of waveforms. I'll try some more complex things.
#43
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 23:36:02 (permalink)
my first wavefile imported with the "Add" button on the keymap editor always spans B3-C#4 with the root at C4, the second imported wave file spans D4-E4 with the root at D#4, the third imported wave file spans F4-G4 with the root at F#4, and so on.
Absolutely correct, Bob. But I couldn't get MIDI to trigger any sound, if the primary sample is introduced through the Add button. That's what led me to using the Open Folder icon to import the first .wav, and each one after that through Add. That, and I wanted to start correctly with a nameable and savable program for organization and frequent backups. Note that a sample imported there will have a root note at C5, and span from C0 to G10 by default, although I have seen some exceptions under random conditions. I would then have to correct my method to span that sample to remain two semitones under the projected root note of the second sample. That'll bring in the root into the correct position.

Understand that I'm using sample sets from CDs that contain the root note in the file-naming convention through much of this discussion. I thought that would be the route that most readers would travel, should I decide to write about the procedure in a tutorial. I'm pleasantly surprised at all the folks here trying the 'roll-your-own' approach.

At any rate, I was under the impression that the best (& easiest) solution for consecutive import would be to span upwards from the first root note to just under the next (by looking ahead too the next root position), and proceed in an orderly fashion from left to right. Cakewalk seems to assume that the best place for the root is centered, to then be spanned both upwards and downwards to cover the MIDI notes between root notes. That seems like a logistical nightmare, especially with this interface. Or at least they might thinking that a median value covers the most possibilities, or is the most logical position for the root.

But the focus should be on what approach sounds the best, and in that respect, both Cakewalk and I may be in error. At least one source (Scot Solida) recommends that the samples be spanned downwards from the root note, as by shifting the span upward, you are in fact increasing the sampling rate. That increases the chances of aliasing when running over the Nyquist frequency. This makes a lot of sense, and would necessitate a change to my whole approach in this little exercise that we have going here. Here's an interesting article that goes into some detail on this, with many other sampling tips as well.
< Message edited by b rock -- 1/2/2005 11:58:15 PM >
#44
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/02 23:57:53 (permalink)
Great article. I have a few of Scot's libraries, and they are all top-notch, so I am sure that he knows what he is talking about!

Spanning samples downward from the root note makes a lot of sense, so I'll try to put that advice to use in my next sampling attempt.
#45
rabeach
Max Output Level: -48 dBFS
  • Total Posts : 2703
  • Joined: 2004/01/26 14:56:13
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 00:44:15 (permalink)
initial testing in cool edit 2000 with a base 440 Hz sine wave shows many more artifacts with extreme up pitch shifting. i went up to 880 Hz and down to 220 Hz. the 880 Hz wave showed considerably more artifacts than the 220 Hz wave.
#46
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 04:27:44 (permalink)
For anybody interested, here is another "test" DS864 sample bank:

http://paintedblue.panicnow.net/Instruments/TranceKit.zip (~4 MB)

There are five programs in this bank and one Track Patch that loads up all five programs with effects. Since all five outputs aren't meant to be played together in the Track Patch, you'll have to mute/bypass the outputs you don't want to hear.

Anyway, they are just some detuned trance lead samples that I was using to mess around with while continuing my DS864 studies.
#47
techead
Max Output Level: -31.5 dBFS
  • Total Posts : 4353
  • Joined: 2004/01/24 08:40:20
  • Location: Macomb, IL, USA
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 08:09:37 (permalink)
ORIGINAL: b rock

But the focus should be on what approach sounds the best, and in that respect, both Cakewalk and I may be in error. At least one source (Scot Solida) recommends that the samples be spanned downwards from the root note, as by shifting the span upward, you are in fact increasing the sampling rate. That increases the chances of aliasing when running over the Nyquist frequency. This makes a lot of sense, and would necessitate a change to my whole approach in this little exercise that we have going here. Here's an interesting article that goes into some detail on this, with many other sampling tips as well.


That makes a lot of sense. I'll experiment more. Thanks Tom for sharing all of the hard fought info!
#48
Digital Aura
Max Output Level: -24 dBFS
  • Total Posts : 5118
  • Joined: 2004/01/18 13:30:46
  • Location: Petrolia,Ontario ORIGINAL P5 FORUM JUNKIE
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 15:58:43 (permalink)
ORIGINAL: PaintedBlue
Spanning samples downward from the root note makes a lot of sense, so I'll try to put that advice to use in my next sampling attempt.


I meant to ask this... maybe this is why the default for DS864 is root note and a spread on either side of it (as opposed to just spanning up).
What should the maximum spread be? Will the C's and G's be enough notes to use when attempting this?

P.S. B-Rock, yes the TRinity does have an onboard sequencer and was about the best one available for keyboard workstations because of the great touch screen LCD display!


Official P5 FANBOY!!!
#49
rabeach
Max Output Level: -48 dBFS
  • Total Posts : 2703
  • Joined: 2004/01/26 14:56:13
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 18:05:36 (permalink)
ORIGINAL: b rock

....... At least one source (Scot Solida) recommends that the samples be spanned downwards from the root note, as by shifting the span upward, you are in fact increasing the sampling rate. That increases the chances of aliasing when running over the Nyquist frequency. This makes a lot of sense, and would necessitate a change to my whole approach in this little exercise that we have going here. Here's an interesting article that goes into some detail on this, with many other sampling tips as well.


b rock i don't like scot's wording. this is why; when you shift pitch up you are downsampling (i.e removing samples; reducing the period of the waveform) this leads to a reduction of sampling rate therefore a narrowing of the bandwidth of the spectra that can be sampled without aliasing. bear in mind that sometimes my mind just freaks out and reverses all the know laws of physics in our universe; during these short sojourns i do some of my best work. so if this is one of those times just ignore what i said. :-) an increase in sampling rate reduces the chances of aliasing why 96k sounds better than 44.1k at high frequencies.
< Message edited by rabeach -- 1/3/2005 6:15:00 PM >
#50
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 18:26:00 (permalink)
What should the maximum spread be? Will the C's and G's be enough notes to use when attempting this?
Tough one, Grg. I'll echo Blue and answer the question like a politician: depends on the source. If I were sampling a "live" instrument, I'd err on the side of caution and sample every note, then discard until I was satisfied. True synth samples are the most forgiving: you might be able to get away with one per octave when sampling that. The tricky ones are the instruments that change timbre as the note value increases. I'd include most woodwinds, guitar, grand piano; that sort of instrument.

It also becomes problematic when an instrument changes timbre depending on how loudly it's played, or playing techniques. Think plucked strings, brass instruments, and, once again, piano. That begs for multiple samples for velocity layering. The toughest are those that combine the both problems; it's very difficult to create a convincing saxophone sound. And we haven't even considered all the novel quirks and performance techniques that most acoustic instruments possess, like guitar finger squeaks, pizzicato strings, or muted trumpets.

As a rule of thumb, I'd say that you can forget about C0 to under C2, and also the three-plus octaves above C7. Most instruments are not played in these ranges, and you can get away with stretching a few octaves there (except for the piano). You'll only use them for special effects, anyway. Many purists only sample the true range of the instrument being emulated, and don't sample (or stretch) any lower than the low E string on a guitar, or two octaves higher than the high E string. You can go that route, but you may never discover that unique "growl" sound that your vocal sample produces at C1, for example. Many free sample sets provide for a sample every perfect fourth, and the ones that cost heavily will have 12 velocities for every MIDI note.

Sounds like a cop-out, but you'll have to tailor your sampling technique to the instrument that you're emulating, and the accuracy that you'll tolerate. Sample drum sounds at several velocities; maybe from five to even twelve layers, but only one sample per specific sound (open, pedal, and closed hi-hats, as an example). Cluster your sax samples at every other note in its most playable range, and taper off at the extremes. Go with two samples per octave with an electric bass, especially if it'll be masked in the mix. I'm sure that you get the idea.

After all that theory, here's a practical tip for assisting with velocity layers: Don't worry so much about exact velocity input, then correct for it with a stroke of the mouse (and Shift) in P-Seq. Or even easier: Use the Cakewalk MFX plugin Velocity to shift all your notes to one velocity value, and you may be able to cheat, and use that one sample in several velocity layers by changing the plugin value. Which samples can you use that technique on? Well, that depends ...<g>
#51
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 18:52:50 (permalink)
an increase in sampling rate reduces the chances of aliasing why 96k sounds better than 44.1k at high frequencies.
I'm no expert, but the brick wall that you run into is the Nyquist frequency, above which you will get aliasing (a particularly nasty digital distortion). Now while you and I would consider that a valuable asset and a sought-after feature (<g>), most other musicians don't want that in their compositions. The thing to realize is that the Nyquist frequency is half of the sampling rate, so at 44.1, you'll get aliasing at 22.05 kHz, but at 96, your headroom improves to 48K.

As for the downsampling, I'm not so sure that you're actually removing samples by spanning the samples upward, as you would with a stretching algorithm. But who am I to argue?; you've actually done the diagnosis in a wave editor, and now even my head is now spinning from this thread [I had the strangest dreams last night ...]. I would think that spanning the samples upward would increase the "playback speed", but keep the samples intact. Now you're going to force me into an exhaustive .wav editor session to back up my speculation! And I was so looking forward to a good night's sleep ...

And while you're reversing the laws of physics, see if you can perform some of that quantum magic and actually place the samples into the DS864 by osmosis, upload/download those huge .dp8 files at the speed of light, or at least convert that small mass of a Key Map Editor into visible energy that we can really see...
#52
Digital Aura
Max Output Level: -24 dBFS
  • Total Posts : 5118
  • Joined: 2004/01/18 13:30:46
  • Location: Petrolia,Ontario ORIGINAL P5 FORUM JUNKIE
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 19:44:28 (permalink)
ok...thanks for shedding some light on that B-Rock.

Now.... I've noticed there are DS8 files (banks) and DP8 files which are program files or patches right?

So... why do DP8's have 8 slots on their drop down and Banks (DS8) have 128 ?

I am going to get ambitious and record all 127 programs in bank A and B of my Trinity into DS864 (maybe one a day or so...)
What's the best way to do this!? I'd like to have this series of patches called Trinity Bank A and a separate one called Trinity Bank B. Each would contain the 127 program patches.
< Message edited by Digital Aura -- 1/3/2005 7:51:55 PM >


Official P5 FANBOY!!!
#53
rabeach
Max Output Level: -48 dBFS
  • Total Posts : 2703
  • Joined: 2004/01/26 14:56:13
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 20:00:24 (permalink)
ORIGINAL: b rock

an increase in sampling rate reduces the chances of aliasing why 96k sounds better than 44.1k at high frequencies.
I'm no expert, but the brick wall that you run into is the Nyquist frequency, above which you will get aliasing (a particularly nasty digital distortion). Now while you and I would consider that a valuable asset and a sought-after feature (<g>), most other musicians don't want that in their compositions. The thing to realize is that the Nyquist frequency is half of the sampling rate, so at 44.1, you'll get aliasing at 22.05 kHz, but at 96, your headroom improves to 48K.

As for the downsampling, I'm not so sure that you're actually removing samples by spanning the samples upward, as you would with a stretching algorithm. But who am I to argue?; you've actually done the diagnosis in a wave editor, and now even my head is now spinning from this thread [I had the strangest dreams last night ...]. I would think that spanning the samples upward would increase the "playback speed", but keep the samples intact. Now you're going to force me into an exhaustive .wav editor session to back up my speculation! And I was so looking forward to a good night's sleep ...

And while you're reversing the laws of physics, see if you can perform some of that quantum magic and actually place the samples into the DS864 by osmosis, upload/download those huge .dp8 files at the speed of light, or at least convert that small mass of a Key Map Editor into visible energy that we can really see...

my point exactly the nyquist changes based on sampling rate. should have been a little clearer with my wording. the low pass filters of the 96k sampling device don't have to be as tight as the ones in 44.1k hardware to get rid of aliasing frequencies because of the added headroom. you can get better resolution of higher frequencies with 96k you can hear frequencies above 22k; well if you are able to hear them. at 44.1k you have 2 samples of a 22k Hz signal at 96k you have about 4.

as an example about using downsampling to increase frequency, if you sample 9s/s so for a 1 second wave form you have 9 samples and you throw out every 3rd sample (decimation) you now have a wave form sampled at 6s/s and have increased the frequency of the original wave form. in order to keep the new wave form the same duration or length as the old wave form you would just add samples 1, 2, and 3 (the new samples) to the end of the new wave form. old wave form samples 1, 2, 3, 4, 5, 6, 7, 8, 9. new wave form samples 1, 2, 4, 5, 7, 8, 1, 2, 4. now you have 9 samples again but the frequency and period have changed. look at it this way you have gotten to the peaks of the wave form in less samples therefore the increase in frequency and reduction in period of the new wave form. in other words if you got to the first peak at sample 4 in the original wave form after throwing out every third sample you have now arrived at the peak at the 3rd sample in the new wave form which leads to a reduction in period and increase in frequency.
i should have been a little clearer in my statement, i'm referring to up shifting pitch while still maintaining the length or duration of the original wave form (which i assume most samplers do; of course i have never looked into it)


edit...as an after thought the new wave form would have a nyquist based on 6s/s even though you have 9 samples since technically you sampled at 6s/s. this applies whether you keep the new wave form the same length or duration as the original or not. the opposite is true for pitch reduction or pitch shifting down you increase the number of samples (through interpolation) to increase the period and reduce the frequency.
< Message edited by rabeach -- 1/3/2005 10:22:24 PM >
#54
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 20:48:03 (permalink)
i'm referring to up shifting pitch while still maintaining the length or duration of the original wave form (which i assume most samplers do; of course i have never looked into it)
Never did get into the testing yet; some unexpected stuff came up, and I returned to see your post, Robert. I have no exhaustive evidence to back this up, but I'll offer a quick example with a fixed sample (no looping) that we all have for the sake of discussion.

Load the 15G-KIT01A.wav file (it's the first one in the KIT 01-06 sub-folder of the Pro Samples Dance Drums on the Project5 CD). Use the second line Open Folder icon to map the sample across the entire MIDI key range, with a root note at the default of C5. Now play various notes from the extremes of your keyboard, or use the P-Seq green & black keys to do so. Make sure that you hold each note through its final decay. As the key range spans the entire keyboard, the sample does shorten with an increase in the note#.

You're making a persuasive argument, and (as I've said) I haven't done enough research to disprove it, but if you take away the static length assumption, cracks appear in the logic. You know that I'm not arguing the point for argument's sake, or targeting your logical conclusions. I'm just after the correct answers, and if I'm wrong, then I'll be the first to admit it. After all, we anarchists have to stick together ...<g>
#55
rabeach
Max Output Level: -48 dBFS
  • Total Posts : 2703
  • Joined: 2004/01/26 14:56:13
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 20:54:48 (permalink)
not a problem i added some more to the post i don't think it matters whether the sample stays the same length or not. there may be another way to raise pitch other than decimation i'm just not aware of it. to get to the truth is what is important to me nothing else matters well except life.
#56
rabeach
Max Output Level: -48 dBFS
  • Total Posts : 2703
  • Joined: 2004/01/26 14:56:13
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 21:00:54 (permalink)
ORIGINAL: b rock
..... if I'm wrong, then I'll be the first to admit it. After all, we anarchists have to stick together ...<g>

me too. i believe you might have brought carnivore to the board with that link.
<g>
#57
b rock
Max Output Level: 0 dBFS
  • Total Posts : 8717
  • Joined: 2003/12/07 20:31:48
  • Location: Anytown (South of Miami), U.S.A.
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 21:23:09 (permalink)
Agreed. Now I see the edit to your reply, and that's definitely food for thought. This thread just keeps getting deeper and deeper ... (do you see what you've started, Aaron? <g>). I'll do what I can on my end here to help to mold the final conclusions with everyone that's contributed to this thread. All this theory really does have the most basic of practical applications: Which way do we approach the the sample editing in the DS864 to achieve the best-sounding results? An uninformed direction at this point will cause everyone needless work in possibly having to redo their banks and program files, and it's difficult enough to accomplish the first time through.

I find it very interesting that there was only the occasional post about the DS864 over the past year, then suddenly this topic explodes with an intensity that's rarely sustained here. It's as if everyone had the same questions and problems with the sampler simmering in the background, and were just subconsciously waiting for something trigger their ideas. I think that this is some great stuff posted here, and I'd like to thank everyone at this point for the collaborative efforts. Of course, there's still much to work out here, so it's back to the research ...
#58
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 21:58:20 (permalink)
Here's another question to consider: what are the best practices for managing input levels when recording an instrument at different velocities? In particular, I am wondering how to deal with recording quieter samples for softer-played notes.

For example, I can try to adjust my input level for different velocities to keep all recorded samples at roughly the same level, and I would then adjust the gain for each velocity layer within DS864.

Or, I can record my samples at their natural levels, but I think that the quieter samples would hold less audio data (i.e. and be of lesser quality than if they were recorded at a higher input level?). From there, I could then choose to normalize my samples and manage the gain within the sampler, or I could load the samples into DS864 as recorded.

Any ideas on this?
#59
PaintedBlue
Max Output Level: -77 dBFS
  • Total Posts : 651
  • Joined: 2003/11/08 00:44:43
  • Location: Project5 v2.5
  • Status: offline
RE: BEST WAY TO RECORD SAMPLES FOR DS864 2005/01/03 22:01:29 (permalink)
ORIGINAL: b rock

I find it very interesting that there was only the occasional post about the DS864 over the past year, then suddenly this topic explodes with an intensity that's rarely sustained here. It's as if everyone had the same questions and problems with the sampler simmering in the background, and were just subconsciously waiting for something trigger their ideas. I think that this is some great stuff posted here, and I'd like to thank everyone at this point for the collaborative efforts. Of course, there's still much to work out here, so it's back to the research ...


What's next, Cyclone? Maybe nPULSE? Spectral Transformer? Going back to PSYN would be fun as well.
#60
Page: < 1234 > Showing page 2 of 4
Jump to:
© 2026 APG vNext Commercial Version 5.1