At 10:26 AM +1000 8/19/10, Paul Jacobson wrote:
> On 18/08/2010, at 4:08 PM, Rob Danielson wrote:
>
>>
>> Hi Paul--
>>
>> I trust you are hearing something that's important. I want to hear it
>> too. Can you describe the differences in pre performance in terms of
>> specific, audio qualities? As you well know, comparative
>> "preciseness" and "clarity" can stem from a number of traits. Tests
>> can be tuned to foreground them.
>
>Rob,
>
>Sorry, I can't be more precise. I sold my HDP2 at the time I
>purchased the SD722, and the impression of clarity and spaciousness
>was based on early impressions. It's about 4 months since I used the
>HDP2 and I don't trust my auditory memory sufficiently to expand
>beyond those impressions. As Raimund said it could be placebo
>effect, but I definitely wasn't expecting the improvement I heard.
>
>> Its my understanding that with a test like this one, mic self-noise
>> _remains_ the predominant factor. The "transparent" pres allow the
>> mics to shine through in nearly full glory. But, understandings
>> evolve
>
>The tik tests are predominantly about self-noise however I believe
>the methodology is highly flawed even for this purpose.
>
>The main concerns relate to non-repeatability of the "tik" test
>signal. This should be apparent to anyone who has looked closely a
>sonogram of one of these tests. Due to the mechanical nature of the
>sound source and imperfection of the mechanism a sequence of tiks
>will have non-identical spectral content. In Emil's comparison you
>can see the spectral content of the tiks varies across each clip,
>and none of the clips have a tik sequence which is identical.
I believe the two recorders ran at the same time. I believe clips 1/3
and 2/4 are recordings of the same ticks.
> This is problematic if you want to make comparisons of the tonal
>character or each preamp, as you have no fixed point of reference.
The clock ticks are not there for measurement in any way. They are
there because its easier to have something high in register for the
ears to recognize while attending to other quality details within the
setting and signals. These particular ticks are brighter and louder
than needed and not ideal. I personally don't need the clock tick
reference myself to make the judgements, but others report
frustration trying to "image" unknown room tone as the recorded
"subject." Outdoor presence when its really "quiet" works best for me.
>The non-repeatability has a flow on effect for level matching. There
>is no clear reference level, and matching by ear works to minimise
>any perceived difference between clips, regardless of actual
>difference.
You are right that "flow" is influential and often a desired quality
in a "match."
I don't use the ticks in making the matches. I first use a few RMS
measurements of room tone when the bulk of the impact usually is and
seek an average between these. Sometimes I use EQ to isolate bands
and see how this affects the levels required. Last, I tweak the
levels by ear until I notice I'm making changes of less than a dB.
The hardest challenge is the tonal differences. I feel its best to
include them in the judgement process as best as possible. When we
mix, we make adjustments in level that account for the influence of
tonal balance and vice versa. I similarly play with the levels and
tone of the samples to help make the qualities more tangible. I also
listen to them casually the way I might as if hearing it for the
first and only time. All of this comes down to few amplitude settings
that I usually sleep on.
>Again this is apparent in Emil's test - when using the steady 200hz
>tone that was present in all 4 clips as a reference it was clear
>that the clips for Recorder A were set roughly 1.2dB lower than
>those for Recorder B. Rectifying this changed the perception of
>which recorder had higher levels of self noise.
So we don't confuse anyone, recorders have input noise, mics have self-nois=
e.
>When you are comparing two recorders that have a theoretical
>difference of 0.25dB in a-weighted microphone/preamp noise (based on
>the geometrical addition of a-weighted preamp EIN and microphone
>noise output for a 8dB/20mV/Pa mic attached to DR-680 and SD702)
>1.2dB makes a real difference to how you assess the relative order
>of these preamps.
>
>Something like a common reference tone played at the start of each
>test recording -- -40dB, 1khz sine or similar -- would at least
>allow accurate level matching between various setups with-in a
>specific set of tests and remove the arbitrary gain changes
>resulting from perceptual matching from the equation.
There's one set-up. Just the mics were swapped to reveal any
significant discrepancies they might introduce.
I use pink noise as you describe sometimes. The lower registers of
"room tone" can also be used in a similar way when the recorders are
rolling at the same time. Many level references and measurements can
be incorporated but it doesn't solve the predicament that the pres
have different frequency response under 80 Hz. Its the way the pres
are-- the question is how to judge any differences in ways that can
translate into significant differences in field recordings. The
distinctions one may find through measurement may not translate very
well or at all.
>It's worth remembering that EIN is calculated from the measured
>level of the recorded noise plus the actual gain of the preamp, so
>to be really worthwhile these tik tests should account for actual
>difference in preamp gain only. So no perceptual matching allowed!
>Sorry ;)
If the perceptual tests we've been making over the last 6+ years
proved to conflict with results in the field or with each other, I
would have stopped making them. On the contrary, the results have
been very consistent-- even the tests made before Raimund started
providing us with measured input noise figures we could also take
into consideration.
> >I have no prejudice except that if a piece of equipment truly
> > produces important differences, we should be able to readily hear
>> them. There are many details to account for in a comparison test like
>> Emil has made and I'm pretty confident that his test conditions
>> present ample opportunity for several kinds of difference to surface.
>
>Even in these test there are clear differences - comparing clip 2
>and 4 there a greater emphasis on sub-200hz in clip 4 for example.
The greater low Hz response of the SD 702 is definitely there. This
difference could be quite glaring with a good subwoofers hooked-up.
Rob D.
>
>cheers
>Paul
>
>> I often sense phenomena that seem palpable that I can't adequately
>> describe. When I narrow my focus, my goal is to hear it, to dis-cover
>> it, if I can. If I can't, I can't. The opportunity usually resurfaces
>> if there is some _thing_ to it. Rob D.
>
>
--
|