Ce forum permet à des personnes du monde entier de communiquer, c′est pourquoi les messages échangés sont en anglais.
-
Hey all,
I've read the manual, guides, tutorials, and can't figure this out. I've made a soundfont from scratch but no matter what I do to attenuate the sound (global, instrument, preset) with positive values (even tried negative values), the sounds clip (play way too loud) if played in a DAW (Cubase, etc) beyond the approximate velocity 66. Thing is, I don't want people to have to play (or 'draw') notes that are half quiet in order not to distort.Attached is my SF2Uploaded my SF2 to box (https://app.box.com/s/3duck4h8yhpdlmvgduyt90uqfxgzvm9y) created via Polyphone (42.2 MB). Any insight would be greatly appreciated -
you forgot to assign each sample to a specific key (sample middleC to key 60, etc...) this is easy in Polyphone; Tools> Instrument> Automatic Distribution.
for all soundfonts you are going to make in the future;
1. give your samples a proper name, so you can find them back in the Sample Folder of bigger soundfonts. i myself add the Root Key of the sample as the number of the key it will be assigned to.
2. always assign a few milliseconds Attack to your Instruments ro avoid possible popping at the beginning of the sample
3. always assign 1 second Release to Volume to your Instruments (to start with, you can change that later if it is too much or too little)
https://drive.google.com/file/d/1D9IG_p4…3eS/view?usp=sharing
regards bottrop -
bottropyou forgot to assign each sample to a specific key (sample middleC to key 60, etc...) this is easy in Polyphone; Tools> Instrument> Automatic Distribution.
Wow, this worked! Thanks soooo much! And thanks for having taken the time, I hugely appreciate this.
bottrop
2. always assign a few milliseconds Attack to your Instruments ro avoid possible popping at the beginning of the sample
3. always assign 1 second Release to Volume to your Instruments (to start with, you can change that later if it is too much or too little)
Good advice, but I had already taken care of that when cutting up the samples as I always do, cut from the nearest zero-crossing point (where a signal is neither positive or negative) before the transient, and added a quick fade out at the end which is the same as release. It's manual ADSR but I like to have granular control over the sounds before I export them. It's actually the 1st time I embed them into a "sound font" so I was new to that aspect, so many thanks for that!
Have an excellent day!
p.s.: If ever you see this sound font uploaded, I've restricted the key range back to 36-96 instead of 0-127 because I want this to be like the original source without any time-stretching. It can't be worse than a lot of sample "patches" I have for Kontakt which are very limited in their range, like say a Baritone Sax, having been sampled from (in that case) real instruments which don't have such an extended range. But it's good to know you can do the 0-36 and 96-127 trick for scenarios where it doesn't really matter that much -
the fade of a sample is not the same as Release in a soundfont. you will only hear a fade when the whole sample is played, with shorter notes you wil hear an annoying cut off.
play a few short notes at key 55 of both Presets in the attached sf2 and you will hear the difference.FadeRelease55.sf2 -
bottrop user has already written what is needed.
Some minor remarks:
A value of around 0.300 ~ 0.600 for the vol-env-release is sufficient for jazz guitar. (it may vary depending on whether it is pure electric or acoustic body)
Values around 0.600 ~ 0.900 can be used for cased instruments: Violin family and Nylon Guitar.
Values of 1.00 and above can be used for grand pianos and ensembles: Strings, Choir, Church organ, etc.
Don't be afraid of these values seeming too high. The fade-out time is fast, and when half of the entered fade-out time is passed, it usually reaches the inaudible part.
Technically you should use a minimum value of 0.001 (~0.003) for Vol-env-attack. And if you don't want a dominating guitar-pick sound or piano-hammer sound or something similar at the beginning, this value can be 0.007 or higher. -
bottropthe fade of a sample is not the same as Release in a soundfont. you will only hear a fade when the whole sample is played, with shorter notes you wil hear an annoying cut off.
Man, I don't know why my mind was acting up and not identifying it as such, of course if I cut a note halfway or anywhere, especially with lower frequencies having less cycles to complete per second, it's going to create clicks/pops, thanks for clarifying that there!bottropplay a few short notes at key 55 of both Presets in the attached sf2 and you will hear the difference.
Do you mean note 55 as in G3 ? I tested both your GtrJazzRel & JtrJazzFade and none seemed to click (yes, very good studio headphones + over a decade of almost daily mixing). I'm wondering, maybe my soundfont loader (TX16Wx) has a built-in release, because I hear clicks all the time usually, especially with things that control dynamics such as compressors & limiters.
Wow -> No wonder I couldn't test the release properly or hear your examples, they put a 50ms default release in FX settings. I turned it off and now I hear them very clearly. I wouldn't want to publically release a soundfont not knowing what the end-user might have like settings if settings at all. Good to know, thanks.
[img size=350]https://images2.imgbox.com/63/23/3lLnGXRO_o.png[/img]
ziyametedemircanA value of around 0.300 ~ 0.600 for the vol-env-release is sufficient for jazz guitar. (it may vary depending on whether it is pure electric or acoustic body)
Values around 0.600 ~ 0.900 can be used for cased instruments: Violin family and Nylon Guitar.
Values of 1.00 and above can be used for grand pianos and ensembles: Strings, Choir, Church organ, etc.
Yah, that's what I thought about when thinking the way a compressor would act. Thanks for the info. I (almost) always stay on the safe side and put 50ms minimum especially when there are those lower frequencies.ziyametedemircanTechnically you should use a minimum value of 0.001 (~0.003) for Vol-env-attack. And if you don't want a dominating guitar-pick sound or piano-hammer sound or something similar at the beginning, this value can be 0.007 or higher.
I would understand for the release but I'm not sure about the attack here. I mean, in my case I prepare them before importing them into Polyphone and the attack is already taken care of. Or is there something I'm ignoring about sound fonts the same way I hadn't caught on early about the release being not the end of the note but ceasing to play it before the end -- as for the attack, it always starts there anyway. My question sounds sarcastic but it's not, I'm really curious. -
This is a safe value and is equivalent to "no attack".
Type 0 in the vol-env-attack cell and press enter on your keyboard. What do you see?
Soundfont players should have set 0.001 as the default value if nothing is written in that cell, but it is useful to write this value here to be safe. So the value 0.001 should already be the default value. (Equivalent to an empty cell in an error-free soundfont player and doesn't change anything.)
The internal values here are -32768 (most negative=1msec), 0 (=1sec), and 32767(most positive=100sec). Instead of these values in the user interface, we see them converted to seconds. 0.001, 1, 100
Try typing 200 in that cell in Polyphone. What did you see: 101,593 right? This is an error, the number here cannot be greater than 100. (I wonder if the number 32768 is used for the most positive value, or is this a rounding error?)
In order to avoid some mistakes like this, I suggest you not leave this cell blank and enter a value.
For example, with some older soundfont players you will sometimes hear a strange pop or crackle at the beginning of the sample. Because there is a calculation error; the synthesizer wouldn't know what to do with this extreme unexpected value. And it tries to reverse-attack. Convexly from maximum decibel to minimum. I remember solving this problem by typing 0.003 in that cell.
In reality I don't think a value between 0.001 and 0.003 will be a problem.
A note: In versions 2-01 and older of the Soundfont specification, filter-frequency was not applied for samples with attack times less than 0.007, in version 2.04 this requirement was removed.
And if a soundfont player accepts 0 as a value (Houston, we have a problem) that value would be equivalent to 1 second (for technical reasons in the spec). Because this value cannot be multiplied or divided.
But since you will not publish this soundfont and use it for your own work, what I wrote is nothing more than satisfying your curiosity. -
ziyametedemircanThis is a safe value and is equivalent to "no attack".
Type 0 in the vol-env-attack cell and press enter on your keyboard. What do you see?
I know, I found that really annoying the other day (I think first time I used it was 2 days ago), but glad it does that now that I know what you outlined!ziyametedemircan(I wonder if the number 32768 is used for the most positive value, or is this a rounding error?)
I'm guessing (programmer here) that since -32768 -> 32767 is a signed int (interger that allows negative values), those are the max values that can be held. I'm reminded of that often when I see 0-127 (that one signed) for 128 (including zero) is just because 16 bytes (8 bits * 16 = 128) was the limitations at the time with synths.ziyametedemircanFor example, with some older soundfont players you will sometimes hear a strange pop or crackle at the beginning of the sample. Because there is a calculation error; the synthesizer wouldn't know what to do with this extreme unexpected value. And it tries to reverse-attack. Convexly from maximum decibel to minimum. I remember solving this problem by typing 0.003 in that cell.
In reality I don't think a value between 0.001 and 0.003 will be a problem.
Really good to know, thanks! I'm taking screenshots of all these posts, including @bottrop's, and including it in my Polyphone installer folder for future reference.ziyametedemircanBut since you will not publish this soundfont and use it for your own work, what I wrote is nothing more than satisfying your curiosity.
I actually do plan to put it publically available. They're not the best samples out there (I even own Nexus) but I like classical stuff and I was doing it more to immortalize someone's "old" Casio keyboard and his two favorite presets, but I'm sure a couple people around the world might feel the nostalgia.
Connectez-vous ou inscrivez-vous pour participer à la discussion.
Polyphone a besoin de vous !
Polyphone est gratuit mais il y a des coûts associés à son site web et à son développement. Un petit coup de pouce aidera beaucoup.
Faire un don
Apprenez les bases
Voir le tutoriel
Haut de
page
page