Ultrabeat sample init changes the sample? [SOLVED]

I want to control the pitch of my audio file kicks using Ultrabeat. I right click on one of the instruments for ultrabeat and select 'Sample Init' and drop my kick into the sample window and the kick sounds totally different to the same audio file in an audio channel.

The kick sounds really filtered but the filter isn't turned on, the signal flow seems to be right (unfiltered/modulated) but maybe i'm missing something...

anybody have any ideas?

maybe using exs24?

Is the attack of envelope 4 as short as possible? From memory the sample init setting might have a few ms of fade-in on the envelope.

The other unfortunate thing is that sometimes UB does f**k with your samples. See this thread on Apple Support. This really really bugs me, as I love the workflow of making beats in Ultrabeat, but it often ruins the sample's sound.

So the possible solutions are:

1) Use EXS24 instead - treats your samples nice.

2) Manually add 30ms or so of silence to the start of the audio sample you want to use. Pain in the butt!

Thanks for the replys,

The attack on envelope 4 I manually set to 0 and still had the problem, it's a serious shame that Ultrabeat can't do this, the onboard kits i find mostly terrible and it's workflow would make things alot easier if it would work properly.

I think the EXS24 will have to take over in this department, one thing i want to add is using Logic 8 you can't use Bazzism (doesn't register) and I have never heard a kick generated using any of Logics onboards that sound anywhere near it,

surely there is a way to get a nice clear cracking Kick with a bassy tail.....

A DANCE KICK?

The attack on envelope 4 I manually set to 0 and still had the problem

None of the envelope attacks in Ultrabeat go to 0. The minimum setting is "< 0.001 ms", which is not 0, so it still softens the attack of your samples. That's just the way Ultrabeat was designed. If you don't like the way it sounds, you need to use another sampler, such as the EXS24 for example.

Please excuse me David,

So are you saying that because the minimum attack setting is less than zero, it's causing a soft/slow attack? which is affecting the sample?

isn't that what attack is supposed to do when increased?

did you read the thread on this at apple support?

I don't see why they would have incorporated this feature if Ultrabeat creates an unchangable envelope.

So are you saying that because the minimum attack setting is less than zero

It's actually MORE than zero. I've never heard of a time being less than zero (that would be great news if there was such a thing)!!

ah ok,

thanks for all the help guys it's now time for me to get that EXS24 rollin

Smaller than 0.001 ms...

that is les than 1/1000 of a millisecond, aka a microsecond, of which there are a million in a second.

Pretty short. Still, light travels 300 metres in that time... sound only 0.3 millimeter...

I don't think there's an audible difference between 0 µs and 1 µs attack. I also don't think that in practice, in any synth, there IS an attack of absolutely 0 µs.

One sample at 44.1 kHz already lasts much longer than that (almost 23 µs)

Smaller than 0.001 ms...

that is les than 1/1000 of a millisecond, aka a microsecond, of which there are a million in a second.

(...)

One sample at 44.1 kHz already lasts much longer than that (almost 23 µs)

True. The number doesn't mean any softening by itself. Well, maybe I will examine this one by using an oscilloscope later, but meanwhile here's a workaround: Use a sample editor to insert some silence before the start of the sample (maybe 1ms); this will add latency (which you can compensate if you think you need to do so), but the sample's sound will start when the envelope is full up.

Smaller than 0.001 ms...

that is les than 1/1000 of a millisecond, aka a microsecond, of which there are a million in a second.

(...)

One sample at 44.1 kHz already lasts much longer than that (almost 23 µs)

 

True. The number doesn't mean any softening by itself. Well, maybe I will examine this one by using an oscilloscope later, but meanwhile here's a workaround: Use a sample editor to insert some silence before the start of the sample (maybe 1ms); this will add latency (which you can compensate if you think you need to do so), but the sample's sound will start when the envelope is full up.

It makes sense and no sense... add a useless 1000 µs to get rid of something smaller than 1 µs... and still it is a valid option... but adding just 1 (one) sample of silence would also suffice....

Still, I like academic questions. You need to rewire logic and reason with math to solve them...

... but adding just 1 (one) sample of silence would also suffice....

Yes and no... Obviously the real attack time must be much longer than 1µs. I think a good sample should alway start at a zero crossing, so the second sample would already meet the envelope's maximum. Even if the second sample were involved you would hardly hear it.

And yes, it makes sense. Delay can be compensated while a sloppy attack can't.

Smaller than 0.001 ms...

that is les than 1/1000 of a millisecond, aka a microsecond, of which there are a million in a second.

Pretty short. Still, light travels 300 metres in that time... sound only 0.3 millimeter...

I don't think there's an audible difference between 0 µs and 1 µs attack. I also don't think that in practice, in any synth, there IS an attack of absolutely 0 µs.

All I can tell you is that Ultrabeat changes the attack of your sample, EXS24 does not.

Jope wrote:

I think a good sample should alway start at a zero crossing

I have no idea why you would think that. I definitely don't agree with you on that. Have you ever listened to a sample that does not start on a zero crossing? Do you just not like that sound, or do you arbitrarily decide that it shouldn't be used?

I think a good sample should alway start at a zero crossing

I have no idea why you would think that. I definitely don't agree with you on that. Have you ever listened to a sample that does not start on a zero crossing? Do you just not like that sound, or do you arbitrarily decide that it shouldn't be used?

If it didn't start at a zero crossing I would not be sure if something is missing. Before the sound is silence, it has to start at zero, hasn't it?

All I can tell you is that Ultrabeat changes the attack of your sample, EXS24 does not.

I actually did find that out a long, long time ago (the topic linked by Widdershins earlier this thread). I had forgotten all about it. And it is a bit more than 1 µs, too...

I actually did find that out a long, long time ago (the topic linked by Widdershins earlier this thread). I had forgotten all about it.

Haha, that sounds like something I would do

There must be something else wrong but the attack of Ultrabeat's envelope...

Here are two waveforms showing what Ultrabeat outputs using a square wave sample. Using "Amadeus" I synthesized an exact square wave of 882Hz at 44.1kHz sample rate, so one cycle is exactly 50 samples long, 25 of which are maximum or minimum each. The amplitude was 80% of the full 16 bit range.

Well, I took this wave (.wav file) and dragged it into the sample area of oscillator 2. I bypassed the filter, switched the EQ off and set attack time of envelope 4 to minimum ("<0.001ms"). No pitching was applied to oscillator 2 (=C3).

I made a "song" with one single note using this UB voice and bounced the thing at 16 bits, 44.1kHz (which also was the song's setting) without dithering. What you see is the output waveform re-imported to Amadeus, one pixel on the X-axis is one sample. The very first cycle of the square starts with 25 maximum sample values followed by 25 minimum sample values. No sample is missing.

For the second picture I used a modified square sample, I had set the first sample value to zero instead of maximum. The bounced song now shows 24 maximum sample values at its start.

Everthing as expected. UB's envelope definitely does not creep in if you set it to minimum, it is as fast as is possible, no sample value is changed.

If everybody still is convinced exs24 does it better, there must be a different reason for that. Maybe it adds something (sort of overdrive) that UB doesn't?

No Zero at start

Zero at start

There must be something else wrong but the attack of Ultrabeat's envelope...

(...)

If everybody still is convinced exs24 does it better, there must be a different reason for that. Maybe it adds something (sort of overdrive) that UB doesn't?

No, it's the attack of the envelope. Note that the EXS24 envelope attack goes to "0.000ms", while the Ultrabeat envelope attack only goes to "<0.001ms". I agree with Erik that that last value in Ultrabeat doesn't make much sense, it seems like in reality the minimum attack is actually longer than that.

Jope, on your pictures I cannot see where the sample starts - is all that silence in front of the first cycle part of the sample? That would explain your results.

Here are my tests. The EXS24 sounds exactly like the original, Ultrabeat sounds (and looks) completely different.

Pic 1 = Original sample

Pic 2 = Sample played back through EXS24, A = 0.000ms

Pic 3 = Sample played back through Ultrabeat, A < 0.001ms

Original sample

Sample played back through EXS24, A = 0.000ms

Sample played back through Ultrabeat, A < 0.001ms

Note that the EXS24 envelope attack goes to "0.000ms", while the Ultrabeat envelope attack only goes to "<0.001ms". I agree with Erik that that last value in Ultrabeat doesn't make much sense, it seems like in reality the minimum attack is actually longer than that.

:idea: :idea: :idea: :idea: :idea: :idea:

I have come to the conclusion that it must be a typo in Ultrabeat, they mean either (less than) 1 millisecond, or (less than) 0.001 seconds - that would make sense since it seems to affect the first (less than) 44.1 samples...

(...) Jope, on your pictures I cannot see where the sample starts - is all that silence in front of the first cycle part of the sample? That would explain your results.

The silence is because I started bouncing some ticks (or how the smallest unit in the counter is called) before the note starts. Keep in mind for the first picture the original squarewave was used (starting at maximum), while for the second picture I used the squarewave with the first sample value had been set to zero. If you really count the pixels (I did), you will find in the second picture the first cycle starts with 24 maximum sample values instead of 25. Nothing is missing.

I will do some further tests later. Maybe it's a matter of sampling rates (project's or sample's) or If the sample starts with negative values or how the shape of the attack is set or...

Later.

Later is now.

This time I took an identical square wave, but starting at its minimum value/negative phase. To make sure we're watching the very beginning of the wave, I put gaps (some zero samples) into both the first minimum and the first maximum. Again, here are two pictures.

The first one shows the bounced waveform with UB's attack slider hard left, tag says "<0.001ms". The second one shows the bounced waveform with the slider... well, not quite left, but... If you switch the autozoom for the attack phase on, there are many positions for the attack slider for wich the tag says "<0.001ms", and this is the first position you can reach at maximum zoom that is not hard left. And yes, the real attack is obviously much longer than 0.001ms, it's about 28 samples = 635µs =0.635ms long.

My conclusion is, UB as included in LP 9.1.3 is able to play samples accurately if the attack is really set hard left. The number "<0.001ms" doesn't mean anything. BTW the next change of the attack phase can be seen with an attack time of about 0.71 ms (tag's reading).

Maybe something has been fixed in this version or maybe sometimes the attack slider behaves different... I don't know. For me it works.

This bug was fixed in Logic 9.1.4:

Logic 9.1.4 fix ultrabeat sample attack cut-off? [sOLVED]

For the sake of completeness in this thread, too:

9.1.3 does the trick as well, for those who are afraid of the ES2 bug of 9.1.4.