MakeMusic
SmartMusic Finale Garritan MusicXML

55ChAhsG3_0002102C

For solving technical issues involving installation, operation, bug reports and troubleshooting

55ChAhsG3_0002102C

Postby enaef » Sat Mar 21, 2015 10:43 am

Instruments for Finale; Choir Ahs
Used in Finale 2012

There is a "bump" in the choir sound, when the file 55ChAhsG3_0002102C.audio is in use. I guess it is the audio-file itself which is corrupted?

Can anyone help?

I tried to attach the audio-file but the extension audio is not allowed.

Thanks
  • 0

enaef
 
Posts: 7
Joined: March, 2015
Reputation: 0

Re: 55ChAhsG3_0002102C

Postby SysExJohn » Sun Mar 22, 2015 6:07 am

Hi enaef,

Usually if there's a bump it's because the loop point in the sfz file has been set incorrectly.
I found this to be the case with the lowest notes of the oboe d'amore in the GPO4 library, which I was able to fix.

I'll try to take a look at the file over the weekend and see if I can spot the problem, then whether I can fix it.

Just as a bit of information, you can open an audio file in Audacity as a 'raw file' and at the foot of the screen enter the sample points from the sfz file to find out where exactly the loop 'turns over'.

John.
  • 0

Author of http://midi-tutor.proboards.com/ MIDI tutorials.
SysExJohn
 
Posts: 245
Joined: December, 2013
Location: Betelgeuse, shoulder of Orion.
Reputation: 21
Software Owned:
Sonar 7 PE, Finale 2012c, GPO4, COMB2, JABB3, IO, World, CPO, Harps, Steinway Basic + GPO5.

Re: 55ChAhsG3_0002102C

Postby enaef » Sun Mar 22, 2015 8:40 am

Hi John

Thanks for your reply.
Your hint regarding opening the file in Audacity "in raw" was helpful. I tried to open the file before but was not aware of the possibility to import as raw data, and failed therefore.

Actually, the problem lies, as suspected, in the audio file itself. My term "bump" may be a bit misleading (I'm from Switzerland and English is a second language for me).

The file is nearly 7.7 seconds long. About at the position of 1 second there is a slight deviation from proper intonation. And again about at point 6.8 seconds; this time a bit less deviated.

I wonder, if I can correct that by myself?

Ernst
  • 0

enaef
 
Posts: 7
Joined: March, 2015
Reputation: 0

Re: 55ChAhsG3_0002102C

Postby enaef » Sun Mar 22, 2015 10:08 am

Ok, I wasn't able to remove the part with the incorrect intonation in Audacity without producing an even worse deviation - this time a real bump.

I guess it is possible to do the correction without bump, however, I'm not capable of it.

Should I ask makemusic for an improved 55ChAhsG3_0002102C?

...

A bit later I had another idea:
I took 56ChAhsG#3_0002102C into Audacity, changed pitch a half-step down, exported it in raw, renamed it into 55ChAhsG3_0002102C.audio and put it in the folder (keeping the original after renaming it into 55ChAhsG3_0002102C.orig).
This is a workaround for the moment only, of course ...

Ernst
  • 0

enaef
 
Posts: 7
Joined: March, 2015
Reputation: 0

Re: 55ChAhsG3_0002102C

Postby SysExJohn » Mon Mar 23, 2015 4:01 am

Hi Ernst,

And an excellent workaround too ... for short notes.
But, the loop points defined within the sfz files for the two samples are quite different.
I would expect therefore, that you would experience some clicking on notes that are held for more than the sample length.

In my experience you'll get no new audio file from MakeMusic.
But, I could be wrong, so it is worth filling in a bug report on the problem.

There is another workaround which involves getting the sfz file to play the G#3 for both G3 and G#3, which is quite easy to program.

The sfz files can be opened in a simple text editor like NotePad (for Windows).
They are usually held within a folder called 'Instruments' in most Garritan libraries, but in the case of Instruments for Finale they are within 'Programs\009. Choir\Choir Ahs.sfz'

If you scroll down the file until you come across the first <region> definition, you'll see it looks like this:

<region> lovel=1 hivel=127 lokey=36 hikey=40 pitch_keycenter=40 amplitude=100 tune=0 loop_mode=loop_continuous loop_start=85041 loop_end=409124 offset_oncc68=120511 sample=40ChAhsE2_0002102C.audio

This definition uses the sample for E2 (MIDI note 40) for C2 through to E2, MIDI notes 36 to 40.
If you look at all the other definitions, except the last, they are one sample per note. The last <region> again uses one sample (G5) for G5 to E6.

lovel and hivel define that this sample be played for all note velocities.
lokey and hikey define which MIDI notes 'play' this particular sample, with the pitch_keycenter telling ARIA which note the sample actually is taken at. ARIA will adjust the frequency of the sample to correspond to the MIDI note when instructed to use the sample for adjacent notes.

The workaround is to remove the key=55 <region> and modify the key=56 to play the G#3 sample for both 55 and 56.

I can do it for you, but since you seem to be keen to experiment, why don't you give it a try.
Do remember to save an unaltered copy as a backup before you start, won't you?

I'll post my altered version in a day or so.

Regards,
John.
  • 0

Author of http://midi-tutor.proboards.com/ MIDI tutorials.
SysExJohn
 
Posts: 245
Joined: December, 2013
Location: Betelgeuse, shoulder of Orion.
Reputation: 21
Software Owned:
Sonar 7 PE, Finale 2012c, GPO4, COMB2, JABB3, IO, World, CPO, Harps, Steinway Basic + GPO5.

Re: 55ChAhsG3_0002102C

Postby enaef » Mon Mar 23, 2015 3:50 pm

Hi John

Thanks for your input!

You are right, there is a bump in my workaround after about 7.7 seconds.
Before I tried your suggestion, I tried something else:
Starting from my workaround with the pitch-down-changed 56 audio, I copied the loop-point definition from the 56 audio to the 55 audio (attachment Choir Ahs_0.txt). This, because my new 55 audio based on the 56 audio and since you mentioned the loop-point definition responsible for the bump I thought this might also be a remedy. However, there was no change regarding the bump.

Then I tried to do what you explained, resulting in the attachment Choir Ahs_1.txt. Somehow I probably did something wrong since the notes 55 and 56 are played too deep.
You might tell me, what I did wrong.

For the moment I changed back to my first workaround.

As I also work with SmartScore X2 Pro, which also includes the ARIA Player, I looked into 55ChAhsG3_0002102C there, but as suspected, it comes with the same flaw.
By the way: do you also hear the problem in intonation in this file?

Best wishes
Ernst
  • 0

Attachments
Choir Ahs_1.txt
Choir Ahs_1.txt (sfz-file)
(18.94 KB) Downloaded 274 times
Choir Ahs_0.txt
Choir Ahs_0.txt (sfz-file)
(19.1 KB) Downloaded 275 times
enaef
 
Posts: 7
Joined: March, 2015
Reputation: 0

Re: 55ChAhsG3_0002102C

Postby SysExJohn » Tue Mar 24, 2015 4:16 am

Hi Ernst,

Yes, I hear two slight 'warbles' in the G3 sample, one at around a second in, the second flaw at about 6.6 seconds. The G#3 sample is a good clean one.

In your file you forgot to define what the key centre is, which is why it doesn't play correctly, so pitch_keycenter=56 is necessary after the lokey and hikey definitions. Take another look at the definitions for the lowest and highest choir notes again, you'll see they both have the MIDI key of the sample defined as the key centre.

These two examples are slightly unusual in that the other notes that the samples are used for, are either all above or all below the key note. In most libraries it's more common to use a sample in the middle, e.g. lokey=40 hikey=42 pitch_keycenter=41 and the sample would be at MIDI note 41 or F2. But that's the joy of the sfz format, it's very flexible in defining playback.

Try that edit and see how you get on.

Regards,
John.
  • 0

Author of http://midi-tutor.proboards.com/ MIDI tutorials.
SysExJohn
 
Posts: 245
Joined: December, 2013
Location: Betelgeuse, shoulder of Orion.
Reputation: 21
Software Owned:
Sonar 7 PE, Finale 2012c, GPO4, COMB2, JABB3, IO, World, CPO, Harps, Steinway Basic + GPO5.

Re: 55ChAhsG3_0002102C

Postby enaef » Tue Mar 24, 2015 12:19 pm

Hi John
Thanks again, that did the trick!
Had I spent a bit more time with the sfz-file, I should have found the solution myself. I was too lazy however ;) . After my choir rehearsal yesterday I went to bed early in order to be fit for work today ...

Now, may I ask some more questions?

I'm interested in how loop-start and loop-end work. I guessed, that the numbers somehow point to the "position" of the audio-file where to begin and where to end with the loop.
I tried to find some information and found this webpage: http://drealm.info/sfz/plj-sfz.xhtml

The text seems to confirm what I suspected: "The player can be told that a certain region of the file can be looped: either using loop_start and loop_end (both range 0 to 4Gb) to specify the sample numbers to start and end looping".
However, I do not understand the term "sample numbers" nor the range 0 to 4Gb.
Can you give me a hint?

With this information I might also understand, why there occur "bumps" when loop-informations are not set correctly.

Although the workaround with the adapted sfz-file works fine at the moment, I also wonder if it were possible to define the loop_start after the first flaw and the loop_end before the second flaw in the original 55 audio file ...

And I also wonder, if it were possible - just for fun of course - to audio myself, safe it in raw and use it in the aria player ... ;)
I would nobody else than me let listen to that of course ... :oops: :oops:
  • 0

enaef
 
Posts: 7
Joined: March, 2015
Reputation: 0

Re: 55ChAhsG3_0002102C

Postby Credo » Tue Mar 24, 2015 10:00 pm

enaef wrote:And I also wonder, if it were possible - just for fun of course - to audio myself, safe it in raw and use it in the aria player ... ;)
I would nobody else than me let listen to that of course ... :oops: :oops:


You can indeed make your own sounds/patches for Aria. You don't need to save anything in a raw format. You can save your samples as WAV, AIF, OGG, FLAC, etc. The sfz format is an open standard, so making your own sfz patches for Aria, sforzando, Dimension, Rapture, etc....is actually encouraged. Drop by http://www.plogue.com/ to get an idea of how to get started.

The best documentation on the opcodes for the sfz standard that I know of can be found here:
http://www.amazon.com/Cakewalk-Synthesi ... ension+pro

The raw.audio format is actually an encrypted OEM thing for commercial ARIA library developers. If you look really close at all of the Garritan *.audio samples, you can see 'issues' with some of them in something like Audacity. They're not all tagged with the correct base pitch or octave (or need alterations during play-back if they are). Some appear to have altered sample rates or some sort of time-shifting stuff required to get them playing back nicely. Some are out of phase or otherwise distorted. Some of these samples even sound 'BAD' in Audacity....but quite nice when ARIA corrects all that during play-back (probably using information from our registration keys somehow).

Credo
  • 0

Credo
 
Posts: 112
Joined: September, 2014
Reputation: 24
Software Owned:
Cubase 8.5; Sibelius 7.5; Finale 2014.5; Bidule;
Garritan Ultimate Collection, Halion 5, Halion Symphonic Orchestra, SONiVOX Film and Orchestra Companion

Re: 55ChAhsG3_0002102C

Postby SysExJohn » Wed Mar 25, 2015 3:57 am

enaef wrote:Hi John
Thanks again, that did the trick!
Had I spent a bit more time with the sfz-file, I should have found the solution myself. I was too lazy however ;) . After my choir rehearsal yesterday I went to bed early in order to be fit for work today ...

Great stuff, Ernst!

Yeah! When messing with programming there's no half measures or "that'll do for now". :)
Like building an aircraft, it's gotta be right first time, every time.
(I served an electronics apprenticeship in the mid-sixties in the British aircraft industry, before moving into computers in 1966.)

Now, may I ask some more questions?

I'm interested in how loop-start and loop-end work. I guessed, that the numbers somehow point to the "position" of the audio-file where to begin and where to end with the loop.
I tried to find some information and found this webpage: http://drealm.info/sfz/plj-sfz.xhtml

The text seems to confirm what I suspected: "The player can be told that a certain region of the file can be looped: either using loop_start and loop_end (both range 0 to 4Gb) to specify the sample numbers to start and end looping".
However, I do not understand the term "sample numbers" nor the range 0 to 4Gb.
Can you give me a hint?

There are two meanings (at least) to the term 'sample'.
The general use is when we refer to the playing of a note (or short phrase in pop music) for the purposes of playing (typically) a virtual instrument of some kind. There is often one sample per note.
The second use is in audio engineering, when we're describing, usually a 16 bit value (but it could be 8 (old Akai samplers), or 24 bit for studio quality mixing (and DVD?)), that defines the volume of an audio signal at a specific moment in time. Many of these tiny samples go to make a sample in the general use of the word.

A sample, in audio engineering terms, depends upon the sample rate.
On a CD and, for the rest of what I write here, lets assume 44,100 samples per second.
It has to be (according to Nyquist who came up with the idea) at least twice as fast as the highest frequency we want to record. We usually reckon that hi-fi audio is 20Hz (cycles per second) to 20 kHz (20,000). (I'm deliberately spelling everything out in detail. I'm sure you know a lot of it.

So, in practical terms how do we cope with this?
Our good friend Audacity comes to our rescue ... again.

If you import the sample choir ahs G3 again and Click in the audio wave at roughly the point where the first wobble occurs, the focus will now always be on that point of the file.
Now, using the magnifying glass with the plus in it (it should be up on a tool bar somewhere) you can magnify the image. If you keep on clicking, the wave will eventually resolve itself into individual samples with points. (For me, about ten clicks shows a series of dots on the wave.)
Each sample is simply a positive or negative two byte value defining the volume the signal is at at this moment in time. 44.1 kHz, 16 bit samples.

Press the "Z" key and the vertical line will 'snap' to the nearest 'zero' crossover point. This is where, if we are trying to create a loop, we would want to start or end a loop point, so there is no audible click.

How do we examine the loop points that have been set in the Garritan libraries, or any other for that matter?
Audacity again, look at the foot of the window.
You should see "Project Rate" with the pane showing 44100.
Next to that are three boxes.
In the first one can type in the loop start position, in the second the loop end.
Look to the top of the screen and you'll see a grey bar has appeared (you'll have to zoom out again to see it).
Using these various tools you can work on a complete instrument sample (the whole thing) sample by sample (1/44100).

However, the choirs are stereo samples. Many solo instrument samples are mono.
Try "OboeC#5_0001102C.audio". The sfz file introduces another concept offset=.

With stereo samples it's more difficult to find a point where both the left and right audio channel are at zero. In the case of the choir sample one channel is zero but the other channel is positive, but the same value, so there will be no sample amplitude mismatch causing an audio glitch. It's necessary to be aware too of the overall direction in which the wave is travelling, +ve or -ve. And so on.

With this information I might also understand, why there occur "bumps" when loop-informations are not set correctly.

Although the workaround with the adapted sfz-file works fine at the moment, I also wonder if it were possible to define the loop_start after the first flaw and the loop_end before the second flaw in the original 55 audio file ...

And I also wonder, if it were possible - just for fun of course - to audio myself, safe it in raw and use it in the aria player ... ;)
I would nobody else than me let listen to that of course ... :oops: :oops:

Have a look on the internet for the player Sforzando by Plogue. It's a single MIDI channel version of ARIA. I sometimes use it for messing about with samples. It's free. BTW, Plogue made ARIA for Garritan. They have a forum too where the writer of the drealm site, pljones, can be found to contribute.

Er, does any of that make sense?
Regards,
John.

Oh! And the range?
Loop points can be gigantic (4GHz = 4,000,000,000).
An instrument sample four thousand million samples long at 44,100 samples per second is ... er ... 90,702 seconds or ... um ... 1,511 minutes or ... more than 25 hours!
How daft is that?
  • 1

Author of http://midi-tutor.proboards.com/ MIDI tutorials.
SysExJohn
 
Posts: 245
Joined: December, 2013
Location: Betelgeuse, shoulder of Orion.
Reputation: 21
Software Owned:
Sonar 7 PE, Finale 2012c, GPO4, COMB2, JABB3, IO, World, CPO, Harps, Steinway Basic + GPO5.

Who is online

Users browsing this forum: No registered users and 0 guests

cron