Topic: A new Italian temperament: CHas

This is a recording done on a Steinway S to demonstrate a recently developed Italian temperament, called CHas:

http://www.box.net/shared/od0d7506cv

It's a variation of Equal Temperament that divides two octaves equally instead of a single octave. The creator argues, in part, that expanding the octave into two octaves provides a better sound that partly takes care of IH almost automatically on a modern piano. 

When it was first introduced, without any recording, and only defended as a mathematical formula, it was attacked, but I'm liking the recordings I'm hearing. (EDIT: I should add that CHas is not exactly just a temperament. It is a theory based on dividing the two octaves equally. The tuning\temperament emerge from the theory, which in turn arose from the developer's extensive experience in tuning.

Seems as though there are several recent variations of ET out there. There is a long, long, often contentious, thread about this temperament over at http://www.pianoworld.com/forum/ubbthre...874/1.html .  The developer, from Sicily, had a few problems with English, with the form of his presentation, and used formulas and variables lost some readers. He was hit fairly hard there. It took him some time to understand that people needed to hear what he was talking about. Now, there are several recordings, and people have become more interested. There's a tuning sequence, but not one with cent deviations or herz specifications, at http://www.pianoworld.com/forum/ubbthre...050/1.html . Not for the faint of heart.

Not suggesting that this replace conventional variations on ET, of course. Just thought it was an interesting sound.

Last edited by Jake Johnson (06-03-2010 18:13)

Re: A new Italian temperament: CHas

lovely sound Jake!

Re: A new Italian temperament: CHas

Very interesting...I like the slow beating it seems to introduce, as well as the great music and playing.

If you ever find a way to implement it in pianoteq (scala or otherwise) I would be interested.

This kind of beating is what I try to do with my own favorite presets by working with unison and stretching.

Re: A new Italian temperament: CHas

A very sweet, singing sound, yes. I like the equal beating, too. Let's hope that we hear more of it.

Re: A new Italian temperament: CHas

I witness that this tuning set is very particular.

Being so much near from what our usual intention is that analysis would certainly not show where are the differences, but the way the resonance is enhanced is immediately noticed by pianists, whith comments about the beauty of the tuning.
The ones who tunes knows that this is not so common.

What would be necessary for a scala file ?

I wish to test that set with harmonic tones. The tuning is done by ear and with a very particular auto-settling method which is installing the tension in a coherent and stable way in one pass (may be possible up to a few HZ pitch difference)

Harpsichords can be tuned that way, so it is not only piano related (while it provide a new way to manage the piano iH).

I only have tuned 8 pianos that way,  I am each time still amazed by the way things fall in place naturally.

May be pianoteq would show significant results in term of harmony.

Please let us know what kind of data is needed.

Thank you in advance.

Isaac OLEG

P.S the "temperament" is ET indeed, with coherence based on the 12th double octave relation. For a result all beats are progressive, the 12th double keep the same relation all along, the 5ths goes from slightly tempered in the mediums to pure in the high treble and bass.

The octaves are indeed enlarged, but not much more than in an usual ET tuning, the enlargment is pretty much hidden in the spectra). The larger span gives more precision and a slower beat rate acceleration for 3ds10ths 17ths, etc.

Just modern contemporary ET. Ill take some time to understand the scala files use next week.

Last edited by OLEK (22-02-2010 14:34)

Re: A new Italian temperament: CHas

Olek,

I've heard the recordings of your tunings. Those pianos sound great.

Scala may be good for trying this tuning. Would just a manual tuning, using the Detune pane, be just as good? More specific questions and concerns-- number 4 really worries me the most:

1. For this tuning, can you just set the pitches of the 24 notes, and then have Scala repeat the same intervals between the pitches for all of the other notes? Or do you have to tune the notes (using the Detune NE pane in PianoTeq) by ear and coincident partials to get the equal beating?

2. Obviously, there's also the question of the way that PianoTeq automatically adjusts for IH (unless the Temperament is set to Flat, which is simple to do). I think it does it by matching up coincident partials, but we can't control which partials it automatically matches on. So it may be better to use the Flat setting and then set the upper 12's by 3:1 by watching them in the Spectrum profile NE pane while using the Detune NE pane. (I don't think that Scala can do any partial matching--it just sets absolute intervals and offsets.) So would a combination of a Scala file and manual repitching in the Detune NE pane be best? Seems like it, but I worry about tracking changes in a tuning if there are two sources for the pitch changes.

3. I worry that one problem with Scala is just that it's a separate program: each time you want to make a change, you have to open the Scala file, make the change, save it, open the file in PianoTeq to listen for the effect, make an adjustment, etc. Lots of going back and forth. Might be simpler, in the long run, to just tune it in the Detune pane?

4. The larger concern: PianoTeq lets us control the inharmonicity of each string with the String length parameter. Thus there will be no single proper tuning in CHAS. We might want a different IH setting to get a different sound instead of the default string length setting. In other words, there's the issue of designing the piano and doing the tuning at the same time. The final tone will be something to look at FIRST, since there is no physically imposed IH that is beyond control.

In an case, I'm looking forward to hearing what you do with CHAS and PianoTeq.

Last edited by Jake Johnson (22-02-2010 15:38)

Re: A new Italian temperament: CHas

Hi all,

I made the Scala file for s=1.

-----%<----------Cut here ---------%< ------
! CHaS.scl
!
CHaS s=1
12
!
100.038318440222
200.076636880444
300.114955320666
400.153273760888
500.19159220111
600.229910641332
700.268229081554
800.306547521776
900.344865961998
1000.38318440222
1100.42150284244
1200.45982128266

-----%<----------Cut here ---------%< ------

I'm impressed by the increase of harmonic resonance you get with this temperament.

Don't forget to set octave stretching to 1.00 otherwise it won't work !

Nikos

Last edited by Nikos (22-02-2010 23:52)

Re: A new Italian temperament: CHas

Nikos

Thank you for the scala file - it's really good!

Re: A new Italian temperament: CHas

Hello, thanks so much, all for your interest, and Niko for the instructions.

OK I think I get it (was mistaken on the file extension)

I really did not go very deep in the Pianoteq software yet, only played a little with the different parameters and find them amazing and for some, realist.

I will have more time for that during the week end.

I let Alfredo Capurso tell what he think about the numbers .

How are those expanded afterthat ? (what mean s=1 ?)

The octaves are not zero stretched in Chas. Is the stetching done automatically because of the scala file ? in that cas eyou may be able to provide a "pure 12th"  progression as well, may be interesting for comparaison purpose.


I hope for more useful information soon


I will try to listen with a more advanced setup in a few days.

Thanks. Isaac

Last edited by OLEK (23-02-2010 00:59)

Re: A new Italian temperament: CHas

I only listened to some midi files, and yes the harmony is recognizable. It is absolutely amazing how the musicality is raised, as resonance, and you even can change the tuning during the listening (not really in favor of standards ET even with some stretch)

That Pianoteq software always surprizes me !

I like to know if some  harmonic tones may be find (even at the harpsichord the tone is more gentle, but I wish to listen to some organ type tone, any idea ?)

I'll setup my mid keyboard in a few days then I will listen at that as a tuner... (have no midi output yet on that puter , BTW is a good USB midi port availeable ?)

Thanks again, Nikos, I guess it is as simple as what you provided : a new ratio.

Last edited by OLEK (23-02-2010 01:42)

Re: A new Italian temperament: CHas

@Olek:
According to Capurso's formula, the s ratio changes the way harmonics are matching.

With s=0, you get the standard equal temperament (Scala: 100 200 300 ... 1200) where 2nd partial of a note matches exactly its octave (in Pianoteq at least because inharmonicity is compensated).

With s=1, you get the chas temperament where 3rd and 4th partials matches the octave of the fifth and the double octave +- a small delta.

With other values, you can vary the way the partials are matching but I think s=1 gives the best result.


The octave stretching is given by the LAST Scala number (1200.4598...) so you have to set octave stretching to 1.00 in Pianoteq because Scala file is already stretched.
For your info: there is no stretching if it is exactly 1200 (or 2/1 ratio in some Scala files).

Nickos

Last edited by Nikos (23-02-2010 08:26)

Re: A new Italian temperament: CHas

Nikos wrote:

@Olek:
With s=1, you get the chas temperament where 3rd and 4th partials matches the octave of the fifth and the double octave +- a small delta.

With other values, you can vary the way the partials are matching but I think s=1 gives the best result.
Nickos

I worry that the difference between CHAS theory and tuning may not be taken into account. Scala is fine for absolute pitches and offsets, but Alfredo's tuning is largely aural. (I don't worry that there may BE a difference--no one tunes a piano to straight ET, either)

One big difference: Alfredo emphasizes that he uses a 12th check, matching on 3:1 from A4 (our A3) up.

The tuning sequence is here:

http://www.pianoworld.com/forum/ubbthre...ost1326050

Other bits and pieces, responses to questions, etc, appear in several threads in  the Technicians forum at PianoWorld. A search for his name may be valuable.

Last edited by Jake Johnson (23-02-2010 13:49)

Re: A new Italian temperament: CHas

It's good to know that a 12 note scala file is sufficient to get the CHAS tuning effect. I had previously done it with 24 notes and also 88 notes to be sure of the effect in pianoteq, and they all sound very slightly different, probably due to pianoteq's treatment of scala files (maybe 64 bits calculations inside) but I think the 12 notes file sounds the best in fact.

I still think it's not completely exact because the standard A=440Hz is supposed to be the starting point for the tuning which should go in both directions to flatten the bottom octaves and sharpen the higher ones slightly. With a scala file, I am not sure where the fixed point is. Maybe the lowest A.

EDIT: Comparing CHAS with the flat temperament, only middle C (C3) stays the same to my ear, so that's probably the fixed point for scala files in pianoteq. Can't discriminate with my spectrum analysis program, only integer Hertz values are displayed...

Also, I hear no beating of consecutive octaves with the 12 notes version, so I think maybe the 24 notes one (similar, just extended) is maybe closer to the real CHAS tuning.

Last edited by Gilles (23-02-2010 21:03)

Re: A new Italian temperament: CHas

Hello all !
Alfredo Capurso will chime in soon, but he told me that the formula is well implemented with the scl file.

In real Chas on a piano, the octave beat is pretty much hidden in the specrta, to me it is due to iH and the slowliness of the beat that will tenhd to couple durinf sustain.
unless you can hear a beat coming in 3seconds, or have it in evidence with tests, it is coupled with the partials and then disappear to the ear.

Is it possible to use a 24 notes scl file ? (or any size ?)

I had fun listening to different midi files and changing the tuning during listening, comparing the musicality...

if pianoteq bases the tuning on a pure 2:1 relation by default, this seem too compact and straight for a realistic tone. That is the Yamaha approach, but the octaves are opened then, for practical purpose, for what I know... ( can be opened without beating audibly)

Re: A new Italian temperament: CHas

I worry that the difference between CHAS theory and tuning may not be taken into account. Scala is fine for absolute pitches and offsets, but Alfredo's tuning is largely aural. (I don't worry that there may BE a difference--no one tunes a piano to straight ET, either)

One big difference: Alfredo emphasizes that he uses a 12th check, matching on 3:1 from A4 (our A3) up.

The tuning sequence is here:

http://www.pianoworld.com/forum/ubbthre...ost1326050

Other bits and pieces, responses to questions, etc, appear in several threads in  the Technicians forum at PianoWorld. A search for his name may be valuable.


The complete tuning was also recorded, :
http://www.box.net/shared/1te90to9yd  (108 Mo file, I'll make a smaller one to simplify upload)

then you can see that the way the tuning is done tend to stay near the theory. No partial match tests are used, octaves 12ths and doubles are evaluated for their lowest beat (and 12th is not tuned as such, it only may be the proof of the remaining being coherent).

Talking of the usual checks we use I have there a mistery. on a vertical I tuned : the 6:3 octave is still "short" from 1/3 bps +-) the 4:2 octave is large by 0.3 bps, but the 2:1 relation is still slow at  1/4 bps.

The theory state that the 2:1 may be larger than the 4:2 , than the 6:3.
It seem that practically it may be not the case sometime.

Last edited by OLEK (24-02-2010 09:33)

Re: A new Italian temperament: CHas

Hello.

I would like to thank you, Olek, for your attitude, what makes you a demanding colleague and, today, a generous friend.

Thank you all for your interest. I'm so glad if you are able to transfer Chas onto Pianoteq. If this were to make you enjoy this resort, tuning and playing more, a substancial part of my goals would be achieved.

Nikos, you got the "s" variable too (section 3.3). When s = 1 we may get Chas in its ideal/optimum semitonal beating form, and what I wonder about this s discretional variable is if, in practice too, it can open to infinite forms of temperament, microtonal too, so overtaking many circumstance distances. In my intention, this s variable translates the model's meta-cultural approach and its intimate "elastic" properties.

I'm not at all familiar with sound processing, softwares or similars. Nevertheless I hope to be able to contribute in this direction too by replying to your questions, both about Chas theory or practical aural tuning.

Jake, a special thank to you for reading through my writings.

Best regards, Alfredo.

.

Last edited by alfredo capurso (24-02-2010 14:07)

Re: A new Italian temperament: CHas

Could someone upload the Scala file to the Files area?

I thought I had this file working in Scala, but lost it somehow. Now, if I just copy the above text into Scala, and click on OK or Apply, I get an error message saying:

"Illegal pitch value(s)"

What am I doing wrong?

Some confusing things about the scale as posted, too. Does Scala use s=1 to denote stretch? According to the Scala Help file, to enter stretch you type Stretch 1.  But I'm relatively new to Scala.

EDIT: Or did you enter the numbers manually using the Input command in Scala? When I do this, and set the stretch, the resulting file is a mess, with the upper octaves grayed out in PTeq and odd pitch changes.

Last edited by Jake Johnson (24-02-2010 17:23)

Re: A new Italian temperament: CHas

I just made a text file named chas.scl containing the lines above.

ascii plain text, it worked.

Re: A new Italian temperament: CHas

I was able to get it to work in Scala by cutting out all of the text and just using the numbers.

Two other worries, now:

Nikos' Scala file puts the stretch into the 12th note. (No text such as s=1 is needed in the file.) But:

1. Alfredo specifies a temperament based on two octaves. Does the Scala file start the stretch an octave too soon? Is the stretch supposed to appear at the start of each octave or at the start of the unit of 24 notes?

2. If the stretch is added in the 12th interval, does that work out correctly as Scala "copies" the temperament to the left of the keyboard? I can see that it might work, but I just don't know Scala well. Do you see the possible problem? : Does Scala know to NOT add that stretch for the last pitch of the octave below the bearing\temperament octave? What could logically happen is that, since the last note of the temperament is also the first note of the next octave, the first note for the temperament octave (middle C) could have its pitch raised by the stretch.  (The raised pitch would be small, too small to see in the hz readout in PianoTeq, but would have affect the tuning\beating of partials and octaves.

(If the stretch had been added in the conventional way as part of the Scala file, I wouldn't worry.)

Re: A new Italian temperament: CHas

I was wondering if the various Scala implementations in pianoteq that seems centered on a fixed C instead of A did justice to the real CHAS (s=1) tuning, so I generated with a javascript program the resulting frequencies to plot the differences: (Right click on it to see larger)

Well, visually there is not much...as you can see, since all is in the decimal places, as can be seen in the numerical values left.

http://lh5.ggpht.com/_TjduP84EQE0/S4V9OOQPTrI/AAAAAAAACTU/iHmkCouHK7Y/s720/CHAS_graph1.jpg

Another view, with the same series ordered from inside out in the ring where individual differences show more, especially in the low frequencies.

http://lh3.ggpht.com/_TjduP84EQE0/S4V9NW-cEtI/AAAAAAAACTQ/PsG_3I4wsMI/s720/CHAS_graph2.jpg

The curious thing is these decimal places that separate the pianoteq "flat" temperament (equal temperament with no stretching) from CHAS are quite audible...even if it is hard for anybody (for me at least) to discriminate between pure tones one hertz apart. It seems all is in the partial interactions with the piano. These really produce an audible difference.

Last edited by Gilles (24-02-2010 21:33)

Re: A new Italian temperament: CHas

When going into fine details like those mentioned here, you may find it interesting to know how inharmonicity is taken into account in Pianoteq w.r.t. scala files, as it has an important influence on the resulting sound:

  • if the last note in the scala file has exact value 1200.0 (or 2/1), then the natural inharmonicity of the piano is taken into account. That means that the effective frequencies are those obtained by an equal temperament with inharmonicity (as would be tuned by ear by a piano tuner) modified accordingly to the values in cents specified in the scala file,

  • if the last note in the scala file has another value, then inharmonicity is "bypassed" and the effective frequencies are exactly those specified by the scala file.

Re: A new Italian temperament: CHas

Olek,

When you've tuned pianos using CHas, do you get the same sound on the 10ths and 17ths in the bass and tenor as when you use the Scala file in PianoTeq? There's some beating there, but it may only be what it is intended. (Listen to F1-A2, for example. Remember that Middle C is C3 in PianoTeq.)

Last edited by Jake Johnson (24-02-2010 22:36)

Re: A new Italian temperament: CHas

Jake, for the moment I cant use real checks, or I have to produce a midi file with all test intervals and play it.

I will do tests with a real midi keyboard this week end (in the meantime I will try to record that midi file with a music writing software)

On a real piano you get a progressiveness of beats acceleration for all rapid beating intervals up to 3 octaves difference (24ths), and all along the keyboard. very surprising that it is obtained relatively easily (without having to modify yet tuned notes or make lot of compromising, only the low bass ask for some smoothing with for instance control of the 14ths , mostly because too much different beats are in the way so it is for practical purpose)

To me as soon as the logic is respected the output may be similar. (whatever central note is used, the incremental ratio only differs, the remaining is supposed to be symetric and coherent- indeed the real piano modify things, but no reason that pianoteq does it differntly)


What concerns me more is the absence of any iH correction (it even surprised me in that case that the Chas resonance is so audible).

What makes me think of Chas is also the better clarity of the right hand , as the warmer basses, that stay focused and are only smoother (too low basses are making a strong rrr tone but they sound easily out of range) . Now may be Chas have a little more "move" on a real piano.

In fact if no correction for iH is used (the correction is not tha high , in the a3-a4 octave it is ofen around 1cts just to have the 2:1 partial matching.

I have no idea on how iH is corrected, in Pianoteq, if it refer to the global tone impression due to the hammer hardness and level of inharmonicity, or if it relates to the octave (most tuning theory and practical way rely on octaves, it is may be the biggest problem for tuners !)

So I cant say for the moment how is the differenece.
Simply the chas formula seem to apply "as it is," on a real piano, with no need to add any additional st-retch in the treble and low basses (may be the reason why it seem to work in that case)

I have also no idea on the tuning model used in Pianoteq, most French tuners use an enlarged octave (little stretch added a with  1/3 bps at the M3-10Th level) but other ways exists (as Japanese tuning model that rely on a flat octave at the 4:2 level , when seeing that thru the anglo saxon analysis).

One have to understand that "pure intervals" cant exist at the piano, so any real partial relation only correct partly for iH.

The pitch impression is always modified by the level of the iH , and also by the ear's own filtering. Tuners put the partilas more or less in "sight" when tuning unisons, that allow to modify the pitch impression (not purposely probably) .

Then a straight octave will sound more "open" with open unisons than with tight ones.

I really need to setup that keyboard. !

Last edited by OLEK (25-02-2010 00:25)

Re: A new Italian temperament: CHas

"...I have also no idea on the tuning model used in Pianoteq, most French tuners use an enlarged octave (little stretch added a with  1/3 bps at the M3-10Th level) but other ways exists (as Japanese tuning model that rely on a flat octave at the 4:2 level , when seeing that thru the anglo saxon analysis)."

I had no idea that there was so much difference between countries.

Re: A new Italian temperament: CHas

The way to compare temperaments may be to use the "10 meter" string lenght, no sympathic resonance, no quadratic effect, no stretch, unisons to 1.0 .

The iH is then reduced but still remains( on theh C3 model)

I loaded the 3700 scl files from the scala web site.

Cant the scl files be applied before the iH correction ? would it be very difficult to implement ? those scl files are a great tool.

Is the building of an instrument "from scratch" allowed" with another version of pianoteq ? Where can I know the available parameters , out of the interface ?

I have read on the forum that the piano description files are allowing only the modifications within a margin (the tone stay somewhat policed even with worn hammers).

Thanks for your insights.

Isaac

Re: A new Italian temperament: CHas

1. "The way to compare temperaments may be to use the "10 meter" string lenght, no sympathic resonance, no quadratic effect, no stretch, unisons to 1.0 . The iH is then reduced but still remains( on theh C3 model)"

Just to be sure that you understand--setting the string length to 10 meters will mean that the scale is set so that the longest string is 10 meters long. The other strings will then take their relative length.
You must use the pro version to change the length of each string individually. Then, the string length readout will mean that you are setting the length of any given string as though the longest string was X. So: if you change  C3 to 2 feet, you are setting it to be the length it would if the longest string was 2 feet long. About the unisons: with a setting of 1.0, they are still detuned. Setting them to 0.0 puts them exactly on pitch with the center string.

2. "Can't the scl files be applied before the iH correction ? would it be very difficult to implement ?"

I'm not sure. To take off the iH correction for any file, one sets the Temperament to Flat. However, if one then loads a scl file, the setting changes to Default (for the keymapping). I'm not sure if PianoTeq then applies iH correction or not.

3. "Is the building of an instrument "from scratch" allowed" with another version of pianoteq? Where can I know the available parameters , out of the interface ?"

The Pro version lets you control the parameter settings for each note, including the pitch, length of each string, and almost everything on the interface. Are you searching for other, specific parameters? (I hate to say it, but have you reviewed the manual?) But please do ask any questions you have. Your tunings are very good. We want you to learn PianoTeq well--you're insights will be valuable.

Remember that setting the Temperament to Flat kills the iH correction.

4. "I have read on the forum that the piano description files are allowing only the modifications within a margin (the tone stay somewhat policed even with worn hammers)."

The hammers can be worn in several ways: you can make them too hard at any of the three velocities, and you can set their angle of attack with Direct sound duration, which off-sets the hammer so that it doesn't strike one unison string at precisely the same time as it strikes the other trichord strings. You can also increase the sound of the hammer strike, and with Strike position, change the point at which the hammer hits the string. But is there another quality of the hammers that you want to control?

I hope these answers help a bit. PianoTeq has a learning curve. We've all gotten lost in it.

Last edited by Jake Johnson (25-02-2010 17:20)

Re: A new Italian temperament: CHas

Thanks, Jake for your explanations, and the 10ths test file
I'll answer them after finish my testings.

Something I just realized, is that the 12th step in scala files is the octave, not the b (as by defaust the 0 step is C).
Opening the chas.scl file in scala.exe shows that each interval  is enlarged by 0.038 cts (not much). more information when the scl file is in the editor. A very interesting software I may say.

So the results I had while keeping 1200.0 for step 12 changed all the octaves, reimposing the iH correction hence the heigh difference I felt for the b's was due to the scale jump.

It will be better this afternoon with a real keyboard.

BTW in pianoteq, the 10ths beat rates are at the same time at the top note level and at the partial match level - probably as in a rela piano but more noticeably).

Last edited by OLEK (27-02-2010 13:39)

Re: A new Italian temperament: CHas

Since it seems obvious that the CHAS tuning by itself doesn't stretch the upper and lower octaves and is more or less the same as the flat tuning in these regions, and also that any playing around with the 1200.0 or 2/1 value in the scala file to get the pianoteq stretching destroys the CHAS tuning (or so it seems to my ears) I think the best way is with a 24 notes scala file along with stretching on the note edit detune pane using values from a Steinway piano that were once posted.

The CHAS tuning works best for the M3 along the current presets, so I post an fxp in the files section to demonstrate this.

It is named "M3 Recording (CHAS24 Stretched)"

Here is the 24 values scala file:

----
! CHAS 24 notes
!
CHAS 24 notes
24
!
100.038318440222
200.076636880444
300.114955320666
400.153273760888
500.19159220111
600.229910641332
700.268229081554
800.306547521776
900.344865961998
1000.38318440222
1100.421502842442
1200.459821282664
1300.498139722886
1400.536458163108
1500.57477660333
1600.613095043552
1700.651413483774
1800.689731923996
1900.728050364218
2000.76636880444
2100.804687244662
2200.843005684884
2300.881324125106
2400.919642565328
----

And here is the stretching in the detune pane:

Detune = [-17, -16, -14, -13, -12, -9, -9, -7, -6, -4, -3, -3, -3, -3, -2, -2, -2, -2, -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, +1, +1, +1, +1, +2, +2, 0, 0, 0, 0, 0, 0, 0, +1, +1, +1, +2, +2, +3, +3, +3, 0, 0, +1, +1, +2, +2, +2, +3, +3, +3, +4, +4, +4, +5, +6, +6, +7, +8, +8, +9, +10, +11, +12, +13, +14, +15, +16, +18, +19, +22, +24, +26, +28, +31, +33, +36, +40]

EDIT: Uploaded demo : "Debussy Reflets dans l'eau" in the files section.

Last edited by Gilles (27-02-2010 19:20)

Re: A new Italian temperament: CHas

That is great, Gilles,; thank you

I am just testing the beat values and octave sizes in the default tuning on the C3- better use a standard piano setup.

You know, the high treble is not "stretched with Chas, in fact it siumply follow the same progression than under it (same for bass) .

Phil, can you tell me if you try to use a "standard (or whatever) beat set for the tuning, or, if the method is an automatic partial match provided by computation A very interesting result in that case I may say, but then it should be very interesting also for us to know what "weight" is given to the partials , depending of the interval, if it is a complex model based on octaves and 5ths or if other intervals play a role within ?.
Can you express that in a few lines without giving any secret recipe you may have worked on ?

Thanks

Isaac




)

Re: A new Italian temperament: CHas

Please, I need to know if the temperament is from C3 to C4

ANd if the stretch = 1.00 is yet adding stretch and from which notes.

Re: A new Italian temperament: CHas

Gilles,

I could only listen to "Debussy Reflets dans l'eau" and I'd say it is not bad at all. What is your opinion? And thanks very much.

I can not compare that tuning, and comparesons do help a lot. But you may have...

It would be great if we could have a sequence of chromatic 12ths and 15ths, played slowly, say 3 secs each. That could give an idea of what temperament we are listening to.

Regards, a.c.

Re: A new Italian temperament: CHas

I am reading this fascinating thread with half a mind to 12 string guitar tuning - 4 lower courses are octave pairs. There is a big difference in several practical ways, but string inharmonicity is common to both piano and guitar.

The octave stretching in CHAS, is less than 1 cent. For a guitarist, it would be difficult (and, I think, make for a boring performance) to play in such a way that plucking notes does not introduce pitch distortion (deviation) that is greater than this value.

What, I am wondering, is the variation in pitch in an actual piano between playing a note say pp and ff?

Re: A new Italian temperament: CHas

Gilles, thank youfor your MP3 and fxp files.

II have since yesteday the possibility to check intervals, but infortunately my keyboard  is only 2 octaves .

Styll that allow for more good try than before, with ufor the moment too small audio quality, I will try to correct that this afternoons (BTW are midi ports on USB availeable and good ?)

At first I've find that the Chas stretched is a little too much stretched. Not that it does not sound nice, but it seem to me. Can you post  a longer file.

Did you check how the 12ths are in the treble ?do you know how to compare beat rates ? (even the slow 12th beats are audible if the audio is good, but some beats appears where they are not supposed to be usually, as in 10ths with a beat at the top note level)

I have also tried differnt fxp files I find in the files section, and I may say I find some of them convicncing, I guess I begin to get the point about what you say on the more musical behaviour of the pianoteq, ves a sample based digital piano or software.

What is still missing to me is the change in the attack of tone that occur when a lower note is open and allow a partial to match. On a real piano, my guess is tha in that case the stabilsation of the tone differs, hence the attack is different, pobably slower.

Here the sympathic resonance is realist , but it is only half of he effect.

May be I only miss the tactile vibrations under the finger, that kind of sensation, but I dont think so. I am sure recordings and analysys have been done that show the change i the enveloppe of the high notes when wower ones help.

Sorry, out of suject there... But I have a strong feeling that different gestion of the unison "detuning" would pair with the same effect. Unison opening is playing within iH of the note, to me, as in octaves.
The enveloppe vary with a faster or slower stabilisation of the attack energy. WHen we tune , we couple the frequencies, but we are at the same time working with the way the energy is adbsorbed or delayed.

A MP3 file wih a selectioof intervals woud show if the tunin is CHAS or no. BTW when you use the detune function, can you check other notes than the oe worked  at the same time aurally, or is it  back and forth proccess ?

Best regards .

Last edited by OLEK (28-02-2010 15:54)

Re: A new Italian temperament: CHas

hyper.real wrote:

I am reading this fascinating thread with half a mind to 12 string guitar tuning - 4 lower courses are octave pairs. There is a big difference in several practical ways, but string inharmonicity is common to both piano and guitar.

The octave stretching in CHAS, is less than 1 cent. For a guitarist, it would be difficult (and, I think, make for a boring performance) to play in such a way that plucking notes does not introduce pitch distortion (deviation) that is greater than this value.

What, I am wondering, is the variation in pitch in an actual piano between playing a note say pp and ff?


As I have listened to guitars tuned in pure 12th and that did not sound strange, the CHas (which is less large at the octavce level) may not be.
(are you confusing 1cts and 1Hz ?)

Because of the high tension on piano wire (75 Kg for the palin wire is usual) , the deformation and pitch change is limited to the first miliseconds of tone, the tone iùmmediately settle in its final pitch (while this one is not as stable as one believe, due to the coupling or other things it  tend to lower in time, more or less depending of the instrument, and the unisons tuning) For what I have seen with a precise spectrum analysis tool a few 10th of cts is not impossible.

At that time I did not try to analyze if the tone could be raising or descending depending of the "shape" of the 3 strings unison.

I was seen that as a different pich level for the strong focused tone and for the sustained harmonic resonance that finish the sustain.

If the strings are only plucked, this probably dont happen at all.

Re: A new Italian temperament: CHas

OLEK wrote:

Phil, can you tell me if you try to use a "standard (or whatever) beat set for the tuning, or, if the method is an automatic partial match provided by computation A very interesting result in that case I may say, but then it should be very interesting also for us to know what "weight" is given to the partials , depending of the interval, if it is a complex model based on octaves and 5ths or if other intervals play a role within ?.
Can you express that in a few lines without giving any secret recipe you may have worked on ? Thanks

Isaac)

Yes, it would be valuable, so that we could better predict the results of working with any temperament.

Re: A new Italian temperament: CHas

"BTW when you use the detune function, can you check other notes than the one worked  at the same time aurally, or is it  back and forth proccess ?"

Olek:

You can just play the notes to listen to them as you repitch another note. In the pro version, you can also easily see the pitch of each partial of other notes as you repitch one or several notes. To do this: Open the Detune NE pane and detach it. Open the Spectrum pane and leave it where it first appears.

You can then change the pitch of any note and move the mouse pointer over to select the corresponding note or any other note in the Spectrum NE pane to see the pitch of each partial. After selecting the note, just hover the mouse over the bar for each freq.

Actually, it would be nice if one didn't have to hover the mouse over a partial to see its pitch. It would be good if just selecting a note in the Spectrum NE pane showed all of the partial freqs, and in fact the changes in these were shown as we changed a note's pitch or changed the string length. By default, of course, the numbers might be too small to see in the narrow bars that represent them. But if we detached the pane and dragged it bigger, we could see them clearly.

A big advantage, here: We could see where partials go from inharmonic to harmonic as we made changes in the Detune or String length NE panes, or when we imposed a Scala temperament.

But the answer to your question is that, yes, the pro version lets one see the pitch and partial feqs of any note while retuning any other note.

Last edited by Jake Johnson (28-02-2010 18:17)

Re: A new Italian temperament: CHas

To all:

I notice that in Gilles' 24 freq Scala file, posted above in this thread, the 12 freq = 12 * freq 1. In other words, 100.038318440222 * 12 = 1200.459821282664, which is the 12th freq.

Should that happen with CHas?
Not trying to criticize your file, Gilles. Just trying to determine if we're getting this correct.

Re: A new Italian temperament: CHas

OLEK wrote:

Phil, can you tell me if you try to use a "standard (or whatever) beat set for the tuning, or, if the method is an automatic partial match provided by computation A very interesting result in that case I may say, but then it should be very interesting also for us to know what "weight" is given to the partials , depending of the interval, if it is a complex model based on octaves and 5ths or if other intervals play a role within ?.
Can you express that in a few lines without giving any secret recipe you may have worked on ?

Thanks

Isaac

Nothing really original there in the world of tuning: inharmonicity is taken into account just in the same way a piano tuner does (listening to the partials beating), making the standard octave compromises and aiming at a regular interval beating progression. The result usually lies somewhere between 4:2 and 6:3 octaves. As we all know, tuning is an old problem which has no "exact" solution...

Re: A new Italian temperament: CHas

Olek:

Here's an image showing how one works with partials in the pro version. See the notes below:

partialviewingcorrected.png

Note that:

1. This image is flawed, for it does not show that I am holding the mouse over the bar for the 2nd partial. (My screen capture didn't include it, and I can't seem to make it include the mouse pointer.)
2. I'm playing and listening to the E, so the piano key is held down.
3. I could also have the Detune pane open, and dragged to the side of the main interface and be detuning a note, such as the E. (Hard to show that in a screenshot--the images get too small.)
3. I've selected another note, making the box above it yellow--the lower C. (To select a note, you just double-click on that little box, which then turns yellow.)
4. I'm hovering the mouse pointer over the second gray bar from the left, so the freq for the 2nd partial of the selected (yellow boxed) note appears to the right of the words Spectrum profile in the NE pane.
5. You can also, of course, just do the simpler thing of selecting the same note that you are detuning, to see its partials change in freq as you repitch it.

Sorry if this is all self-evident. I just thought that seeing the picture would let you better grasp what can be done.

I understand that you do not just want to match partials. However, listening to beats while being able to see the partial freqs can't hurt.

Last edited by Jake Johnson (28-02-2010 19:28)

Re: A new Italian temperament: CHas

Just for thoroughness, here's an image that shows how to work with both panes open at once:

partialviewingcorrected.png

The E is being detuned while the user looks at partials on the lower C.
Again, the mouse pointer is not shown. Here, it is over the yellowed bar for the detuned note in the Detune pane, so that the cent amount of the pitch change appears at the top of the Detune NE pane.

(These screen shots do not show that one must move the mouse pointer from pane to pane to see the freqs of partials.)

If you have a Windows system, you can hold the CTRL key and roll the mouse scroll wheel towards you to enlarge the screenshot. (May be impossible on a notebook with a touchpad. however.)

Last edited by Jake Johnson (28-02-2010 19:44)

Re: A new Italian temperament: CHas

Philippe Guillaume wrote:
OLEK wrote:

Phil, can you tell me if you try to use a "standard (or whatever) beat set for the tuning, or, if the method is an automatic partial match provided by computation A very interesting result in that case I may say, but then it should be very interesting also for us to know what "weight" is given to the partials , depending of the interval, if it is a complex model based on octaves and 5ths or if other intervals play a role within ?.
Can you express that in a few lines without giving any secret recipe you may have worked on ?

Thanks

Isaac

Nothing really original there in the world of tuning: inharmonicity is taken into account just in the same way a piano tuner does (listening to the partials beating), making the standard octave compromises and aiming at a regular interval beating progression. The result usually lies somewhere between 4:2 and 6:3 octaves. As we all know, tuning is an old problem which has no "exact" solution...


Thanks , Philippe, but you lost me here a litle.

Does  it mean that each fxp  filez contains the iH correction (I would call that a "tuning" , out of the temperament zone) for the modelized instrument, that you (or the creator) tune by ear note by note ?
Oris it that you have tuner auraly a standardized model to define the way te iH may be corrected for all coming pianos ?

I understand that each creator may be able to fine tune his model (while this iscertainly not esy if he i nota piano tuner)

But a set of rules may apply to the iH correction, that last being related to the "stretch" cursor I suppose.

Then, a model may exist that is adapted to some kind/size of piano.

For instance if we are in the anglo saxon approach, the treble is stretched by a 2:1 ratio, tempered with other larger partial relation at the double and triple octave level, while in the basse different ways may apply depending of the tolerance to widening, octaves may be as tight as 6:3 or as large a 12:6 wth all partil match in between.1
. A good feature would be a "tuning model" with beat relation that can be weighted, as for instance in the VT 100. "Verituner EDT"

I understand that you may have worked at lenght to provide a "one size fit all" model (which is good, for what I can hear I admit), but may be the cause some user made models are not sounding so much in tune is there. (I listened to a Fazioli that is beating in a strange way)

In the real word, we dont tune a Fazioli like a Steinway, it is even impossible if we apply a standard beat set and stretch, and a Bechtsein will also be different.

May be I just dont get the point, and the stretch parameters are within some file.

Last edited by OLEK (28-02-2010 20:18)

Re: A new Italian temperament: CHas

Isaac, the default iH information is stored in the instrument (inside Pianoteq), each instrument having its own iH. The fxp file contains the modifications done by the user, so if the user changes the string length, then the iH is recalculated from the new lengths, and the tuning is also recalculated by taking into account the new iH values. The new tuning is calculated by the algorithm that I explained that takes into account the proper iH of the piano (no "one size fit all").

Re: A new Italian temperament: CHas

Philippe Guillaume wrote:

Isaac, the default iH information is stored in the instrument (inside Pianoteq), each instrument having its own iH. The fxp file contains the modifications done by the user, so if the user changes the string length, then the iH is recalculated from the new lengths, and the tuning is also recalculated by taking into account the new iH values. The new tuning is calculated by the algorithm that I explained that takes into account the proper iH of the piano (no "one size fit all").


Thank you Philippe, I begin tyo get that (did not undestood the "instrument" concept).

SO there are Instruments to work from, each having its own ih curve and the tuning formula apply to that within pianoteq.


ANnyway as the user can modify that with the detuning function, he can make an aural tuning when building his model.

I have find highly valuable he formula from Alfredo because it limit a lot approximations, and to me it look like if a good compromizing set is find each time because the model, wich sem to fit well in the natural resonance of the piano (what most tuners are looking  for in the end)

What you say here means also to me that an algorithm is there to reduce beats between partials in the consonant intervals, then the "octave size" (or "stretch")  is find depending of the iH (mdified or not) and the volume/duration of the partial in the mix for one note.

If not it is a model with interval size predetermined.
When one use the "stretch" cursor, does it move the note 88 and 1 , o does it enlarge the octaves ?

Last edited by OLEK (28-02-2010 21:26)

Re: A new Italian temperament: CHas

Jake Johnson wrote:

To all:

I notice that in Gilles' 24 freq Scala file, posted above in this thread, the 12 freq = 12 * freq 1. In other words, 100.038318440222 * 12 = 1200.459821282664, which is the 12th freq.

Should that happen with CHas?
Not trying to criticize your file, Gilles. Just trying to determine if we're getting this correct.


I was waiting for Alfredo's answer, but he have very little time this week ..
Yes the 12th step is the ocatve and it is the correct expanding (he told me)

there is a Zeo step which is not apparent in the scl file so 12 is not b it is C (octave) this is logical with the octave stretch parameter.

I was unable to make a 24 steps scl file for some reason, but I will try again.

The result may be the same with the 2 scl file sin my opinion.

My brother, Raphaël, violinist, and his wife, harpist, acknowledge the beauty and the justness of the tone with the scl 12 steps file (and no stretch)

They are amazed by Pianoteq, BTW, find it very good

The CHAS tuning, on pianos, is sounding as if no additional stretch is used, because the same size remain for 12ths alla long  - most of our usual (non Chas) tunings are 12th driven in all the medium range and the beginning of the bass because of the listening for the 6:3 partial match when tuning octaves (does not mean we leave the partial match but it is taken in account aurally to judge for stretch) Also, many tuners are attentive not to have 12 ths than may be larger than the 3:1 ratio (I dont say "pure "purposely  , as really pure intervals are inexistent at the piano, they can be at some level, but the iH will always be unmatched to some point then).

Thank you for the scree shots.
If the detune function allows increments of 1 cts, it is not really precise enough for a fine tuning, where 10ths of cts are used. Hence the interest of a formula to compute the scale, I suppose that in that case the precision is raised.
The precision is huge in the basic tuning provided within Pianoteq, and beats are well progressive - may be a little break due to the stretch parameter or to the Ih correction computation, but more intervals are lining than with an aural tuning, which use generally some "more weighted" interval for time constrain and  because of the methodology.
We should use all intervals available, and it is the case with the Chas "preparatory tuning" (on one string). But most good tuners check the intervals in the octave span "on the fly" ; when expanding their temperament, with some occasional checks at the double octave or/and 12th level, very rarely at each step, hence some intervals are more in favor than others.

That said, because of experience, and if the model is good, the result may be coherent. (and the specialized hearing, or in some case "perfect pitch" or "musical hearing" give the remaining precision and musicality).

All that in my point never relate to a real consensus model. by following a progression, the end result is progressive, but it is not always optimal in term of harmony or resonance, in fact the brain catch on the construction of the intervals and analyses that as being "in tune" because of consistency (I believe).


BTW the "Ghosting technique"  : hearing beats with an excitation of the partial responsive for beats while having the interval notes only open (dampers up, but note not played) works perfectly fine in Pianoteq.
I believe that the sympathic resonances are very well reproduced

If something misses it may be at the "phase" level, which may be certainly very difficult to modeling , even to analyze, and may be is even partially lost in the sampling rate in any case (be it on samples or on models, as to me it involve instant changes in the direction the sound radiates (?) ). I heard that Yamaha installed vibration goodies to provide tactile sensations within the keys, on their last model, may be they give a more realistic reproduction of the phase effects (?).

Re: A new Italian temperament: CHas

Olek,

1. Just to be sure that I understand: You are saying that in Chas, the octave for the first note should be exactly freq 1 x 12? I'm worried: since this is a 24 pitch temperament, that 12th note will come BEFORE any stretch factor. Won't this arrangement mean that there is an absolutely pure octave?

2. You CAN fine tune (tune by using 10ths of a cent).  But you can't do it just by moving the slider. [EDIT: What I say is not entirely accurate: We can change the pitch in increments of a cent, but that's very hard to do with much precision, and the readout doesn't show the small changes.] To do it with the slider, the scaling for the slider, the ratio of slider movement to change in pitch, would require either an enormous interface or not permit a large change in pitch. (I'm not certain, but I believe that tuning in finer increments, while simple, does require the pro version.) In any case, you are NOT limited to a formula.

There was a debate about this subject. Essentially, the argument against having the slider move in smaller increments was that human hearing can't detect less than a cent in pitch change. Which is true. (But I've begun to think, after this debate, that those increments add up across the keyboard--a .3 cent change on an interval in each octave, for example, would add up to several cents by the time you reach the top?)

In any case, I hope time opens up for you soon. I must say that I'm a little amazed by this collaboration. How many languages are involved? Italian, French, English. And Alfredo studied with Japanese tuners who had worked with German tuners? And you are trying to learn both PianoTeq and Scala at the same time! Zut alors!

(Scala, to me, is a program that seems straightforward, but can be dangerous, because of its mathematical precision--one number in the wrong place wrecks the result all across the keyboard. But that is math. If I really knew the program well, it would frighten me less.)

Last edited by Jake Johnson (01-03-2010 18:33)

Re: A new Italian temperament: CHas

Euh, no , Jake, sorry if my post is confusing , I said that the 1200 value in scl is the octave (which was pretty evident) , and that the 1200.49xx value in cts is correct Chas ratio, so it can remain the same above and under.


I get the point for the sliders, we need smaller increments than 1cts to fine tune with beat rates (which are well perceived in Pianoteq). may be a double level slider would do the trick. as I did not try to make a tuning on a software I have no idea  on how practically it is easy or not.  1cts increments may well suffice to put you in the ballpark, then you may refine - I recall analysing a Yapaha P something (8 years ago). and the tuning was done by full cst increments, stretching by blocks of 1-2-4 cts...

the ear cant hear small differnces but can hear beat speed differnces at a much precise level.

I am gone ...

Re: A new Italian temperament: CHas

I posted previously an fxp file of CHAS tuning with a Steinway stretching which I agree is too much with the M3 preset. So I implemented the average Railsback curve instead, in another fxp.

This sounds much better with the Debussy piece that I used as demo. I upload also the complete midi file so that the large arpeggios that come later on can be used to hear the complete effect. I had truncated it at first also because there are a couple of wrong notes in the arpeggios that might be misleading. This file comes from a Disklavier performance from one of the piano-e-competition sites.

I am not a piano technician. I only like the effect this tuning has on some kind of music with some of the pianoteq presets. I think it sounds better on the M3 than the C3 for example which is already very resonant by itself. For Ravel, Debussy, some jazz and surely other pieces with complex chords that use the piano as a singing, resonating instrument, rather than a percussion instrument as is often the case with Prokofiev or Stravinski, I think the CHAS tuning is great.

For more percussive music, I feel that the standard equal temperament separates the notes better and is more appropriate. But then, I am not a professional musician, this is only the opinion of somebody that listened to a lot of music...and played some.

I think listening to the complete midi file while switching between M3 equal temperament and the uploaded fxp is quite convincing. To me, it is a different instrument.

Further adjustments of the Railsback curve by listening to beats can only be done by qualified piano technicians (Olek or Alfredo) using the detune panel, even if it is limited to integer values (no tenths of cents).

I would sure appreciate them posting the result.

Re: A new Italian temperament: CHas

OLEK wrote:

I get the point for the sliders, we need smaller increments than 1cts to fine tune with beat rates (which are well perceived in Pianoteq). may be a double level slider would do the trick...the ear cant hear small differences but can hear beat speed differnces at a much precise level. I am gone ...

OLEK:

No, no. I said that you CAN fine tune in PianoTeq pro. But not with the sliders. (Actually, that's an oversimplication: You can do it with the sliders, but it's very hard to move them in such tiny increments, and the readout doesn't show the decimal places, since it might create an ugly, busy\complicated- looking readout and confuse users.)

Last edited by Jake Johnson (01-03-2010 18:29)

Re: A new Italian temperament: CHas

Gilles wrote:

Further adjustments of the Railsback curve by listening to beats can only be done...even if it is limited to integer values (no tenths of cents).

Gilles,

I think you already know this, but just in case:

That workaround of right-click\Copy\paste to text file\edit the text\copy the text from the text file\right click in the PTEq tuning ne pane\ Paste works for tuning and for string length\inharmonicity, too. In the text file, you can just add the decimal point and values to the pitches you want to change before you recopy the data and paste it in.

(Of course, PianoTeq already lets us use smaller increments with the sliders, but making such small changes is hard, and we can't see the increments in the readout. Making the changes in the text gives you precise control, etc...)

Last edited by Jake Johnson (01-03-2010 18:37)

Re: A new Italian temperament: CHas

I didn't remember that one...but still it's dangerous because the copy from the note edit pane gives you back only integer values regardless of the decimals pasted. Are these decimals really used for something other than just moving the displayed bars?

EDIT: I don't think so...otherwise it would be incoherent when reloading the fxp. Remember the WYSIWYG principle...

Last edited by Gilles (01-03-2010 19:25)