At 9:05 AM -0500 10/4/09, Rob Danielson wrote:
>At 5:17 PM +1100 10/4/09, Paul Jacobson wrote:
> > On 01/10/2009, at 5:18 PM, Rob Danielson wrote:
>
><snip>
>
>I took your original snippet and played with EQ
> >> and made a comparison movie which may make what I'm trying to
> >> describe somewhat audible:
>>>
>>> QuickTime Movie 12mb uncompressed AIFF soundtrack
>>>
>>><<https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_Lrg.mov>https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_Lrg.mov><https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_Lrg.mov>https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_Lrg.mov
>>>
>>> QuickTime Movie 3mb compressed AAC soundtrack
>>>
>
> >><<https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_AAC.mov>https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_AAC.mov><https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_AAC.mov>https://pantherfile.uwm.edu/type/public/temporary/IzoTopeNoiseReductionVsEq_AAC.mov
> >>
>
><snip>
> >Hi Rob,
>>
>>Thanks for making the comparison. I always find your approach to very
>>educational, and I try to follow the less is more method as far
>>possible.
>>
>>I'd noticed when listening to the filtered recording that there seemed
>>to be a slight phasing effect in the wing beats. Looking at a
>>spectrograph display the is a slight but visible loss of detail in the
>>calls and wingbeats. I was using the plugin version in Wave Editor
>>so didn't have access to the spectrogram display for selecting the
>>training sample, so I'd expect to get better results using a licensed
>>copy of the standalone app.
>>
>>I've put up a 15 second edit of the raw file that corresponds to the
>>section you used in the comparison. There is no post-processing of any
>>kind, so it might be a better basis for comparing?
>>
>><<http://www.urbanbirder.com.au/audio/download/241/RoundHill_edit_for_RD.wav>http://www.urbanbirder.com.au/audio/download/241/RoundHill_edit_for_RD.wav><http://www.urbanbirder.com.au/audio/download/241/RoundHill_edit_for_RD.wav>http://www.urbanbirder.com.au/audio/download/241/RoundHill_edit_for_RD.wav
>
>The high frequency complexities of the wing whirs was the first thing
>I noticed when I heard the unprocessed mp3 material. I'll take a
>listen to the original and see what else we might be able to learn.
>Did you reach an opinion about whether there is some "sizzle" in the
>recording from the vegetation? My memory of the situation can be
>useful in guiding tone shaping in post. Rob D.
>
>>cheers
> >Paul
Hi Paul--
Here's another test comparison using the original this time:
https://pantherfile.uwm.edu/type/public/temporary/Jacobson-EQ%27dOrig_vs_EQandIso_Sor3Lrg.mov
Its gain is boosted considerably to help with assessment; you might
want to play it back at what seems like a "natural" level to you.
Given the amount of steady sustain and the "grainy" character of the
"hissy" noise, I do suspect its not just from environmental
vegetation/wind interaction. You can see from the EQ curve, much of
the noise is between 4K Hz and 6K Hz with a large concentration
around 4300 Hz. The amount of noise is different from channel to
channel and pretty fixed in ratio which also suggests its not
environmental. I cannot account for what others may hear above 10Hz
with my ears. I bet others can hear plenty up there. This noise is
not typical for AT-3032 recordings I've EQ'd before.
There seems to be stridency in the loud calls in the original too,
but less than in the mp3. I didn't address this in interest of time.
The wing flutters have a lot of natural phasing but I'm not hearing
it as exaggerated by the EQ or the noise reduction. There are peak
resonances in the fluttering that could be doctored more if its
offensive to you; I cut two of them a tiny bit.
The main perceptual difference between the approaches is the 1K to 5K
Hz lift I gave the recording with EQ. The lift is a bit too much
during the first wing flutter event but I left it at this setting
because I feel the lift improves the sense of spatial depth in the
setting. The center of the lift is around 2600 Hz. These are very
subjective adjustments and they may or may not have a similar impact
on others' monitoring set-ups. I made the adjustments referencing
Mackie MKII 824 speakers. It would be quite different on phones I'm
sure. The lift is an adjustment I make fairly often on 3203
environmental recordings.
As for dealing with the hiss-noise, I personally felt I could get
better spatiality and detail using only EQ than with using Izotope by
itself. Izotope quickly started to take the air and life out of the
3KHz to 6KHz region to my ears. I included a quick test that I tried
using mild EQ before a mild Izo setting where the background hiss was
fairly similar to that of the EQ section. Here are those settings:
https://pantherfile.uwm.edu/type/public/temporary/Jacobson-MildEQ_and%20_MildIzotope_Settings.jpg
I'm not really surprised that addressing the noise manually could be
effective because, as I described before, the noise seems to be
mostly concentrated in narrow bands. Rob D.
--
|
