Subject:And, to start the 9.0c bug list...

Posted by: jeffjams

Date:8/17/2007 1:39:53 PM

Here's one I've never seen before:

I installed your latest release, 9.0c, this morning. All of my compressors now do not properly limit the output when I attempt to compress only a selected area. All compressors seem to work properly when compressing an entire file.

I have observed that my compressors (when processing a selected area that's not the entire file) allow an overage during the first 10 milliseconds of the selected area. Both in preview and after processing.

I first noticed this with my Wave's L3, but it is consistently problematic with the Izotope Limiter, Wave Hammer and the Wave's L1. ASIO vs. classic wave driver...the problem is the same with both.

Sorry to start your weekend off with this one.

J

Subject:RE: And, to start the 9.0c bug list...

Reply by: Kennymusicman

Date:8/17/2007 5:28:37 PM

Been having a little play with regards to this and wavehammer..

With no hardware meters on display, I get an audiable click on preview playback (testing a simple 440hz sine wave @ -2.8dB - length 61 sec, sample selection 2 sec)(other samples lengths also produce clicks). Start and end on zero crossing, with phase cycle set to be complete (ie starting in positive, endng after negative). Peak meter shows -1.1dB, with running average at -9.9dB.
Now if I add the hardware meters. The peak meter now shows correct dB values, and the hardware meters show the -1.1dB...

Turn off realtime, and both meters agree on -1.1dB! Occasionally the VU/PPM meters drop off their reading (literally for a flash - not long duration - and happens with realtime on).

RMS setting creates more random peak values - much less consistency to speak of.


The CPU % indicates only 1% being used, whereas taskmanager tells me it's more around 10-12%.

Bypass both compressor and vol.maximiser creates a clip/peak 0.7dB....

Increase the attack time from 0 - and the peak readings increase. 6ms bumps up to -0.2dB, 8ms is 0.0dB, 8.5ms is 0.0dB(no clip) but 8.6ms(and anything above) is 0.0dB(or more) with clip....

Turning off smooth saturation gives about another 1.1dB of headroom to play with on the peak value.

Changing the compressor setting from input to output, the compressor display shows -10dB vs peak meters' -9.9dB vs hardware meters -1.1dB = 3 different meters and 3 different readings!!!!

Subject:RE: And, to start the 9.0c bug list...

Reply by: Kennymusicman

Date:8/19/2007 7:02:56 AM

Ok, - played even more with this, on another machine (the decent one :) ).

Using my hardware meters (not SF - the ones that go with my hardware for real...) - they agree with the output of Sony HW Meters and the Peak meter agrees with the HWMeters. - Although mine refresh constantly, so I get a more accurate view of what is happening with the peaks. Getting similar results as to my previous post - although in smaller differences - less "peakage" by the compressor. The realtime/non realtime is not making any change - meters agree on this machine regardless of realtime state.

Interestingly, if I select the [sys] bypass- then the preview plays, then the compressor "kicks in" (and gives a click noise to prove it, and bump up the dB momentarily 3dB, then normal until the end of the preview where the compressor "kicks out" with a clip/over (0.3dB).

VU/PPM meters work fine on this machine - no dropouts/flashes.

Task manager still showing SF taking up more than the 1% value - but this time only running at aroud 6%.

The Izotope compressor works very well however - giving a slight click on triggering, but not throwing any random or incorrect levels at entry or exit.

Granted a sine wave file isn't a music test - and I don't complain went it comes to my music - SF works very well for me. Just thought I would look into the original post and post my results FWIW.

TTFN

Ken

Subject:RE: And, to start the 9.0c bug list...

Reply by: ForumAdmin

Date:8/20/2007 7:05:57 AM

Make sure you've removed the edge fades or change the default curve to the (visually) linear one. The default fade type is not equal power, so if the processed material is in phase with the source, you could easily get bumps (easily reproduced with, e.g. Volume).

You can save the fade types and lengths as the defaults by expanding the "More..." button, right-clicking in the blank area below the button, and selecting "Save Fade Values".

When the entire file is selected, the edge fades are set to zero width (regardless of the defaults), so that's why you aren't seeing it there.

Oh, as for the the CPU indicator, it is just an estimate of the effect's CPU load, not the current system load. The reason it is often 1% is because Sound Forge 9.0 processes effects on a separate thread ahead of playback. For CPU-intensive effects, you'll see it spike at the beginning of preview, then drop down to 1%. For very long selections, you'll see it kick up again when it is processing additional buffers.

J.