[linux-audio-user] Fwd: [Jamin] Re: soft clip: Achieving Gain, inconsequential overloads

R Parker rtp405 at yahoo.com
Mon Apr 12 17:45:13 EDT 2004


Hi Jan,

I did some testing with normalization yesterday and
it's not what I need. Normalization, it appears to me,
is simply a destructive gain tool. I've always used
compressors for gain stages but normalization is
definitely cheaper on CPU cycles. I might experiment
with normalizing the new masters. I tried normalizing
one of the old masters and it didn't make much
difference. I've a feeling it won't do anything with
the new masters.

I am placing a single band compressor in front of
JAMin and using high cut filtering and these
strategies are helping to minimize the amount of
limiting.

Because the job is so demanding, it's a great learning
experience. I might actually be starting to enjoy this
but don't tell anyone. :)

ron

--- Jan Depner <eviltwin69 at cableone.net> wrote:
> Ron,
> 
> There's a command line tool out there called
> normalize that you might
> want to look at.  It's on the Loads of Linux Links
> page (that's easier
> to search for).
> 
> Jan
> 
> On Sun, 2004-04-11 at 21:03, R Parker wrote:
> > 
> > --- Mark Knecht <markknecht at comcast.net> wrote:
> > > On Sun, 2004-04-11 at 13:16, R Parker wrote:
> > > > Hi,
> > 
> > > Ron,
> > >    I'm not totally sure I'm in tune with the
> exact
> > > problem yet, but on
> > > my end today I'm doing some first mixes of a Big
> > > Band group I recorded
> > > Thursday and last evening. 8 horns, drums, bass,
> > > guitar and piano.
> > > Probably not the same sort of music you're
> working
> > > on. (Strange music
> > > for me too....jeez...I'm dancing close with a
> sexy
> > > brunette in 1940's
> > > New York...) ;-)
> > 
> > That sounds like a fun job. I don't work with live
> > horn sections very often but man they are great.
> >  
> > >    My first step, which doesn't effect sound if
> it's
> > > done right, was to
> > > simply normalize the tracks up a bit. (I told
> the
> > > normalizer to make the
> > > highest peak -1db)
> > 
> > I've never normalized a track. Can normalization
> hold
> > a -0.5db peak to -0.5 while it pushes a -12db
> valley
> > up to -6db?
> > 
> > If normalization behaves that way, it can help
> > "louden" a mix which would be appropriate for my
> > situation. That combined with cutting dominant
> > frequencies would give me the loudness I need.
> > 
> > I'll read up on normalization tonight. Do we have
> a
> > LADSPA normalization plugin or are they
> non-realtime
> > tools? I imagine you must produce a new normalized
> > file. Is that correct?
> > 
> > Until this job, I've never really needed to get
> every
> > bit of available gain from a mix so normalization
> has
> > never meant anything to me. Someone on this list
> tried
> > to explain it to me once but I'm to old and
> bullheaded
> > to learn anything new. That is, until my rear end
> is
> > hanging in the air.
> > 
> > >    I'm no expert on compressors, but I think a
> > > 4-band compressor is a
> > > really place to start for what you want to do,
> so
> > > Jamin should be
> > > helpful. My biggest recommendation would be to
> > > compress as little as
> > > possible. If I do too much it all sounds
> unnatural.
> > > In my case it's
> > > really only the band below 100Hz that's
> consistently
> > > ranging down. The
> > > other 3 are moving much less.
> > 
> > With the job I'm doing compression and limiting
> are
> > probably alot more extreme than what you're doing.
> > This is multitrack pop music. Reguardless, when I
> > reach for the last bit of gain, I'm limiting to
> damned
> > hard. If normalization does what I hope then it
> > combined with leveling the floor by cutting and
> > compressing dominant frequencies might enable me
> to
> > achieve greater "loudness" without the super hard
> > limiting. As I think about it, I'm not sure what
> else
> > normalization could be.
> > 
> > Good stuff, thanks for the help,
> > 
> > ron
> > 
> > > > 
> > > > > --- Steve Harris
> <S.W.Harris at ecs.soton.ac.uk>
> > > wrote:
> > > > > > On Sun, Apr 11, 2004 at 10:04:43 -0700, R
> > > Parker
> > > > > > wrote:
> > > > > > > Not sure but I probably mentioned
> something
> > > > > about
> > > > > > this
> > > > > > > several months ago. Steve, do you have a
> > > soft
> > > > > clip
> > > > > > > LADSPA plug? If yes, I think it should
> be a
> > > > > > candidate
> > > > > > > for adding to the output stage of JAMin.
> > > > > > 
> > > > > > There are some soft clip LADSPA plugins,
> but I
> > > > > dont
> > > > > > think any of them are
> > > > > > better than JAMin's "boost".
> > > >  
> > > > > The thing is my client has requested a %20
> > > increase
> > > > > in
> > > > > volume across the entire album. The first
> > > question
> > > > > is
> > > > > does he know what he's asking for. Well, he
> > > probably
> > > > > has a fair enough idea. He has compared the
> old
> > > > > master
> > > > > with stylistically similar million dollar
> > > masters
> > > > > and
> > > > > concluded the percentage to increase.
> > > > > 
> > > > > If anyone here has tricks for achieving
> gain, I
> > > need
> > > > > to hear about them. My bag ain't got alot in
> it.
> > > I
> > > > > basically cut dominant frequency either with
> > > > > compression or equalization and with the
> leveled
> > > off
> > > > > floor, achieve gain possible; input,
> compressor
> > > > > makeup, limiter input, boost and output. Of
> > > course
> > > > > then it's just a matter of tweaking ratios,
> > > > > thresholds, etc.
> > > > > 
> > > > > The trade off in general is less dynamics
> and
> > > more
> > > > > distortion with several deciables of
> increased
> > > > > volume.
> > > > > I love the work but this job is like a No
> Rules
> > > Cage
> > > > > Match. I might eventually win the title but
> it's
> > > a
> > > > > good thing I brought lunch and a first aid
> kit.
> > > > > 
> > > > > > I've heard that you can ignore peaks that
> only
> > > go
> > > > > > over 0dB for less than a
> > > > > > millisecond, but I've not experimented
> with
> > > it.
> > > 
> > > I think this is very specific to to where in the
> > > chain it happens as
> > > well as the exact system you're using. 
> > > 
> > > > > 
> > > > > If that's the case, then I might have this
> job
> > > in
> > > > > the
> > > > > bag. I'm watching overloads at the input
> > > channels of
> > > > > my mixer flicker on and off. I've been
> trying
> 
=== message truncated ===


__________________________________
Do you Yahoo!?
Yahoo! Tax Center - File online by April 15th
http://taxes.yahoo.com/filing.html



More information about the Linux-audio-user mailing list