I've got a new version of Xhip which is almost ready to enter RC testing. The only reason it isn't officially RC yet is that I'm still just adding a couple of the last features on the to-do and I haven't personally tested it (which I need to do before I hand over testing to anyone else.)

http://xhip.net/alpha/

Have at it, and let me know if you find any bugs. Please read the info on the alpha page and follow it, with one exception: You can probably create presets safely with this version of the alpha if you want to participate regarding the following request.

I'm looking for presets to build a few additional banks for the next release. Contributors of course will be attributed / given credit.

I'm looking for the same with respect to a few demos. Requirements here are 100% use of Xhip and Xhip Effects. Samples are fine for breaks, vocals, loops, whatever else, just keep the focus on Xhip of course

Those will be uploaded to the Xhip SoundCloud account with optional description and link to your account/website/whatever.

Here is an example (whether or not it's any good...):
http://soundcloud.com/xhip/aciddose-whattheheckis

What's new list is too long to include here, see the alpha page for that.

Thanks from the number one fan of Xhip

no problem

this is undoubtedly the very best feature of all:
http://xhip.net/temp/name_randomization.png

OK i found one little thing i am not sure if it is intentional or not.
Mixer/Noise - the knob has a big jump in level from 0 to -35.95dB, so it is impossible to set it at lets say -50dB or -80dB.
OFC It can be worked around by setting oscillators volume much higher and global volume lower.

Also would it be possible that in some future version all the controls in the Control section could be working with MIDI learn ?
Would be very nice to be able to turn a physical knob to set things like global tune and volume, unison parameters and push a physical button to turn "use prg change" and "mono retrig" on and off.

flags that probably won't happen right away, but the control stuff already works. unison, tuning, polyphony, routing everything works. just use mapcc. that's the last thing on the list actually so maybe i'll fix that (the flags) as well.

last thing on the list is to add options to load a file with the cc map in it. also need an option to enable only when the gui is open / focused and of course you'll also be able to save the settings into a file.

the mix levels are a thing i've been a little bothered by for quite a while. i changed them to -36 range because in almost all cases that's lower than needed. why do you need the noise level so low?

it used to be -48 for oscs and -60 for the noise but i realized i was never using a level below -30 while it was on, and otherwise i'd just have it at -60 all the time which was pretty much the same as having it off.

i'd really rather leave the ranges nice and wide, i just had a lot of trouble justifying it to myself.

I like how it comes across here. Has a lot of presence. I'm going to try my hand at presets for this thing....

Great stuff.

Xhip seems to be tuned one octave higher than normal. Is this intentional or am I doing something wrong?

Moonbeam preset, change sustain to 100%.

Basically all pure triangle wave sounds without xmod etc will have this little problem.
Don't get me wrong, it is just a tiny annoyance.

Yes, I adjusted the root frequency calculation and hadn't measured the result. Turns out I did it in my head backwards.

I'll fix that along with a couple other bugs I've found today and anything else anyone posts before I upload a new alpha. These are the sorts of reasons I say "don't use alpha for long term projects "

Right!

Actually the moon-beam preset in the last version is an example of why I changed the control from -60 to -36. It was supposed to be off in that preset, but it was accidentally left at -59 or so instead. With the level at -35 you'll be able to actually hear that it's on.

Do you think anyone will ever make a preset where they'd want the noise at a lower level intentionally though? It could be something you'd tune in just high enough to get a light fuzz over the sound before it really becomes something you can identify as white noise.

I went over this with this exact preset when I made the change. I figured most likely it would be better to avoid having levels so low you couldn't hear them outside extreme cases. In this case, I found you can get a very similar effect if you use some quantizer (bit reduction) with dither when you specifically want it.

Not something that can be done directly in Xhip without effects, but again it's a sort of trade-off between where the level is low enough vs. becomes too low.

But you're pointing out something I forgot to do. When loading old version presets I need to check if the noise level is < 35 and if so set it to off.

Yep
A noise that is not loud enough to be noticeable as a noise but is an integral and important part of the sound.
Last edited by Mutant on Fri Sep 14, 2012 4:17 am; edited 1 time in total

well it's difficult to accommodate that because even with -36 range, the control tends to always be either off or -6, 0 and so on.

a lot of synthesizers work around this problem by throwing out accuracy. they use something like a simple power of two (4*volume^2) rather than using db. problem with that is of course the display then never says perfect db values, it will always read something like 3.159649264 rather than it's current neat 3.0 steps with the mouse wheel.

of course the range is -inf to +12.0412~db and the contour is "ok", but it's a bit difficult to select anything below -48db because that is below 3%. so from a practical stand point it tends to be identical to a proper db scale -48 to +12. if the 0db point is placed at 80%, the curve is continuous which is what xhip used for quite a long time. (actually, xhip mistakenly used 75%, but it was close enough and never tested.)

to get lower values with reasonable accuracy the db scale _must_ be used because the contour of more than ^2 becomes unattractive in most cases.

other systems use other solutions, for example some mixers use a split at -20db and -80db or other positions. the slider then would generally go +20db to 0db to -20db in the 100-60 range, -20db to -60db through 60-20 and then 20-0 is from -60db to -inf.

that looks ok when you're using ticks on a slider or knob because they simply get twice as wide above 60% and usually the word "-inf" is placed in 0-20%.

so in that case you'd have 3db steps for a while, then 6db steps, then one large step to zero at the bottom.

you have to also consider the number of steps in total. in xhip i'm using 16 in most places.

48 / 16 = 3, and because of the -36 ... 0 ... +12 range, it totals to 48 and gives this nice round 3db step size.

(which reminds me, this is another reason i switched to 36.)

if you use some other range you either have to change the number of steps or change the step size, or both.

so the -36 ... 0 ... +12 range is due to several of my preferences mixed together: 3db step size, 16 steps, smooth contour, accurate db representation, minimal 'useless' range.

so it's actually fairly complicated, there are various consequences to be accounted for any particular decision.

the idea i like most is to use multiple step sizes. 3db until -24 and 6db until -42, then the last step -42 to -inf. that gives 12 steps 3db, 3 steps 6db, 1 step to inf.

4/12 = 1/3, so only a third of the dial exists below 24db which i guess is ok.

only problem is you don't get any steps below -42 and if you want to use accuracy greater than 1.5db (assuming you hold shift and use the wheel, that's a 4x magnification, 6db/4 = 1.5db) below -24db you can't use nice round values.

anyway like i said it's all terribly complicated for such a simple thing when you really focus on it

i'll think about whether the multiple step size is a good solution and if yes, i'll try various configurations of it.

thank you.

ok, i tested the stuff i described in the volume rant post.

one issue i didn't think of before was with the -inf feature, you can't really do that if you're printing the values because they'll just go nuts in the -inf range.

so, instead of using -inf i just made the last step to -90, which i hope is enough.

new version on the alpha page also fixes the tuning issue.