[mythtv] No per-individual line-level adjust: bug or missing feature?

[f-myth-users@media.mit.edu]

    > Date: Sun, 15 Jan 2006 10:16:56 -0500
    > From: Daniel Kristjansson <danielk at cuymedia.net>

    > On Sat, 2006-01-14 at 22:45 -0500, f-myth-users at media.mit.edu wrote:
    > > Should I open a ticket on this [*], or will it be closed as "feature
    > > request without attached patch"?
    > It will be closed...

I was afraid of that.  (I'm still unclear on the rationale for not
just allowing feature requests in Trac.  Other projects do this all
the time.  They're marked feature; they don't (necessarily) get used
in milestones/scheduling; but at least that way they can be tracked
and various suggestions/snippets/whatever can all be kept organized
in one place, without a whole separate mechanism (wiki, archives that
don't keep the info together, etc) to do this instead.  I just don't
get it.)

    > >   If I hear nothing (or encouragement),
    > > I'll do the former; if I hear the latter, I'll add it to the wishlist
    > > at the new wiki.
    > Right place.

Okay, I'll do so.  Tnx.

    > > In short, it seems buggy to me that Myth assumes that all capture
    > > cards will agree on their audio level when capturing.  Apparently
    > This doesn't seem like a big problem on my machines. There is a slight
    > difference between the PVR-500 & PVR-250, but at a few dB it is hardly
    > noticeable.

Either you're lucky, or I'm unlucky, or both, but just because it's
not a big issue for you doesn't mean it isn't for some.  9dB isn't
really "a few"; it's pretty enormous.  There are reports in the ivtv
Trac about differences so large that listenable levels for one card
make the other card close to inaudible---that sounds like >12-15dB to
me.  (And remember there's lots of revision-to-revision variation as
well, and for all I know, even batch-to-batch; it sounds like it's
possible to get outliers in both directions fairly easily.)  Maybe I'm
misreading you here, but it sounds like you're trying to minimize or
pooh-pooh the problem, and I wish you were right---but unfortunately
some people get bitten.  And this should be a pretty easy fix.

    > > My suggestion is to put a per-card slider into mythtv-setup's screen
    > > for each capture card (where this is supported by the card; some might
    > > not).  This slider is used in addition to the master capture slider set
    > > in whichever recording profile is active.  [And see [+] for a further
    > > suggestion.]  The value should get stored in some appropriate table, 
    > > consulted when a capture commences, and be multiplied together with
    > > the master capture mixer when initing the card.  (A more user-friendly
    > > approach would be to be able to adjust this in real time, instead of
    > > via repeated trips to mythtv-setup, but that seems like a lot more work.)
    > > Presumably the slider should be an attenuator, like the rest of them
    > > in the UI, so 100% is unity & everything else is quieter, and should
    > > default to 100%.
    > Sounds like a good idea, it could be adjusted while recording just like
    > the recording picture attributes 'G'. I don't think 100% should be the
    > default though, you tend to get clipping if you go above 70% on any amp.

Remember that this figure gets multiplied by the master capture gain,
and -that- defaults to 90% (well, it did in my 0.18.1 installation,
anyway; I have no idea if this got changed since then).  Given that
the master defaults lower than 100%, I'd argue that every single
-other- slider that might affect this (card, channel, or input) should
be 100% if defaulted or not specified at all (e.g., if we go for a
scheme where "not in database" means "default", which seems reasonable
for, e.g., channels, where there might be hundreds and you only care
about a couple outliers).  Otherwise, we're talking about successive
attenuation at every step, and things will be inaudible by default
unless the user goes and fixes 'em all; that's a bad failure mode.

    > > split setups where the STB drives something else as well); and also
    > > that some individual channels run -very- hot---my own cable feed has
    > > mismatches of ~9dB across certain channels I watch, and this mismatch
    > > has been consistent across -years-.  If I could stop constantly
    > They are probably just using range compression on the loud station, this
    > is a technique used by pop bands and advertisers to sound louder without
    > actually increasing the peak volume (what you set with the volume knob.)
    > I'd love to have an audio filter that detected restricted range in the
    > audio and undid the damage as best it could (and reduced the loudness).
    > But using the volume control on these loud stations would work...

I really don't think what I'm talking about here is compression per
se; I used to work in the audio field and I'm pretty sensitive to
that.  Some channels (again, at least for me) just have their entire
range shifted down many dB compared to others, regardless of whatever
compression is going on, and I can verify this with my dB meter.
(Note also that I'm talking about the actual program part of the
broadcast, not the commercials, for those channels that have
commercials at all.  I know that the ads are a level nightmare wrt
compression and absolute level, and it isn't helped by numskulls at
the headend---one of the cable franchises around here had a problem
for a month or more where national ads were louder than the program
but still okay, but the really local come-to-our-drycleaning-outfit
ads were 20dB (really!) louder than -everything- else---just insane.)