AudioSuite Signal Generator not accurate?

[rantanplan]

Hi there

Strange thing here: I did generate a -18dbFS / 1kHz reference tone of 1 min. length and put it on the timline of a 48kHz/25fps session.

Now, both ends of the segment should match with a waveform zero cross point. But only the start does.

Why is this? Something seems not accurate here...
Any idea?

Cheers

[albee1952]

Why SHOULD both ends be at zero points? If the time of the wavelength does not allow it to cross zero at the 1 minute mark, its simple math(isn't it?).

[joachim]

It should end at a zero crossing; if 1kHz cycles to zero 1000 times a second, then also after 60 seconds.
Same "thing" in PT 11.2; about 4 samples difference at 00:59 minute.
Maybe a graphic issue ?

[Shan]

It's based off the selection and the math will very unlikely add up to be a zero crossing on the end, as there's also going to be sample rounding. Apply a 5 sample fade or so on the end.

Shane

[albee1952]

Yes, I see your point now. Maybe graphics, maybe the 1KHz has a small plus or minus factor on exact frequency....and Shan's comment carries as much validity(or more) than my 2 cents

[rantanplan]

Originally Posted by Shan

It's based off the selection and the math will very unlikely add up to be a zero crossing on the end, as there's also going to be sample rounding. Apply a 5 sample fade or so on the end.

Not really...
With 48kHz sampling there should be exactly 48 samples in 1/1000 sec = 1kHz, i.e. the waveform should be symmetrical. Why should there be any rounding?

My selection is exactly 1 min. (or the timecode grid is wrong).

How I did it: I entered 00:58:30:00 in the counter window and added a marker. Then I entered 00:59:30:00 and added another marker.

Then I selected the timeline between the two markers. If this wouldn't be sample accurate, I would have had a lot of issues in my postproduction work...

[Darryl Ramm]

1 minute at 48kHz samples of a 1KHz synthesized signal and you see the phase out by *4* samples out the end? How accurate do you think these signal generators are designed to be? I can't possibly imagine why you are even looking at this.

[rantanplan]

Originally Posted by Darryl Ramm

1 minute at 48kHz samples of a 1KHz synthesized signal and you see the phase out by *4* samples out the end? How accurate do you think these signal generators are designed to be? I can't possibly imagine why you are even looking at this.

Because I can I didn't say anything about 4 samples BTW

I see no reason why this plugin shouldn't be designed to be accurate enough for the resolution we have...

[Eric Lambert]

Originally Posted by rantanplan

Because I can I didn't say anything about 4 samples BTW

I see no reason why this plugin shouldn't be designed to be accurate enough for the resolution we have...

It should be accurate.

And what does the sample rate have to do with it if it's a 1KHz tone? It cycles 1,000/second regardless of the frequency at which you're sampling the signal. Zero crossings will occur twice every 1/1,000th of a second.

[joachim]

Originally Posted by Eric Lambert

It should be accurate.
Zero crossings will occur twice every 1/1,000th of a second.

Rendered with Audiosuite 20Hz a 1 second selection.
There are 20 waveform cycles, but the 20th cycle does not end with a zero crossing.
Even the 1st waveform cycle is "off" by 8 samples.
Set Grid to 100msec, deviation is clearly visible. One Grid Line should represent 2 waveform cycles. But the zero crossing is later , about 16 samples , and deteriorates naturally from there on.
Does not matter to me a lot practically; but Signal Generator seems not be exact.