quite clearly I don't agree with you. dbx does what you say is impossible. heh.

By doing a reasonable, though not perfect, reconstruction.

They know the exact input compressor settings, they know the exact output compressor settings. That also helps (loads).

Dolby C/B also work. They aren't perfect. They won't allow you to restore a tape that was encoded with something else.

so now you're using what I already stated as your own argument? heh. And while you're quoting yourself I suggest you go re-read what you wrote and what you were replying to. The guy was asking if it were possible to increase dynamics. Your reply was that it was not, and only after it was pointed out to you that you were wrong did you change your tune.

I wrote it wasn't possible to get the original back. You are right that I did not answer the original question, namely is it possible to make an expander (yes).

Since I already made this clarification a few pages up, I still fail to see why you started arguing against my point that it was impossible to get a perfect reconstruction, even going as far as giving flawed examples to try to illustrate you can invert a compressor perfectly.

basically because first you were stating something that was wrong. then you were stating something that relied on degree of accuracy to be wrong. and dbx doesn't give a reasonable reconstruction, for all intents and purposes it gives a perfect reconstruction, since it takes place in the analog world, where your accuracy argument is pretty much meaningless. "Perfectly" doesn't apply. You're talking about dealing with digital audio, which doesn't represent audio perfectly in the first place, what with only sampling the audio at certain distances apart in time. If you want to look at it from a integer math point of view, fine. But then it just comes down to how accurately you're doing your math. In which case I'm right in one instance, and you're right in another.

Of course I'm dealing with it from an integer point of view, since I was asked to make a plugin for FB2K, which is generally used to play back music from CDs.

Whatever is possible in the analogue domain is of no relevance to that.

...and foobar contains 64bit audio representation because...? heh.

The internal respresentation does not matter. 16 bits of resolution stored in 64 bits still remains 16 bits of resolution.

*sigh* You're right. You've never been wrong in your life. Level of data accuracy/resolution doesn't affect computed results at all. You said earlier that it does. But now you say it doesn't. So I guess it doesn't. This is my last post in the thread regarding my discussion with you Garf, so go ahead and get the last word in.

STOP FLAME

yeah, a little childish I guess. Sorry Garf, actually I'm annoyed at someone in the house right now, heh.

hi garf

i am using your fb compressor since you posted it. i really like it the way it is. it does what i want and better than many expensive sw's i use, sure, it's my opinion.

well, i have some questions for you. so, please, i would be very happy if you answer me.

first, i am using following parameters: peak limit 95%, release 50 ms, FSR 1.3, SRC 2.0 and input gain 3 db.

second, my mp3's are from net radio, most of them are dance or trance; i use 'mp3gain' normalization to max non-clip. after all, i convert to wav, using fb too.

questions: what mean 'fast and slow compression ratio'? how can i compare with attack/release? when do these parameters start to work?

again, thank you very much for this great piece of well done work!

The slow compressor is the one that slowly responds to changes in the overal loudness. The fast compressor responds almost instantaneously to short peaks and bursts.

Not sure if that helps you much.

no problems.

i will analyze what results it generates.

anyway, thank you for the reply.

Garf,
the slow compressor obviously uses some kind of absolute threshold, only kicking in if the peak is above some -20dB for a while. So if it is fed music with some very silent and some not-quite-so-silent passages, the former remain unchanged, while the latter are amplified. In this case, the compressor effectively does the opposite of what it's supposed to do, increasing dynamic range rather than compressing it.
I tried using the compressor for classical music, but this effect can be quite annoying. Very silent passages are just as hardly audible as before (they are actually unchanged, I visually verified this with an audio editor), plus there are now sudden, unexpected jumps in volume if the music gets just a tiny little bit louder.
Is this intended behaviour?

I cant seem to get the Dynamic compressor to work how I want it to. I want to basically have everything bought up to a 0dB level peak.

What seems to happen is with some bass boose in the eq which is ahead of the dynamics compressor in the DSP manager, is that when the bass kicks in, the compressor will turn down to prevent clipping as it should, but it never brings the level up again.

This is really obvuious in most of daft punk - discovery where the bass comes and goes all thru the first track. As soon as I touch any slider in the dynamics compressor screen, the level comes up again.

At the moment I have the compressor configured with
peak limit 100%
Release time 10ms (cant go any lower then that)
Fast ratio 40:1
Slow compressor 1:1
input gain 12dB

There is also a little gain on the EQ ahead of it to bring the bass up.

I want to get it so I can hear the whole of the songs the whole of the time, without the loud parts being audable in the next cubicle over. At the moment as soon as it attenuates something, thats all she wrote till I nudge one of the sliders in the DSP manager.

Garf,

do you have time to explain how the Dynamics Compressor works please?

It may sound like a stupid question, but just playing around with it doesn't make the algorithm clear at all.

I'm using 0.8 special, and (for example) Peak limit does this
0%=silence
1%=~6dB down
anything higher = no apparent effect.

I was thinking that the dynamic compressor would be the ideal thing to put at the end of the chain (adjusted so it's actually ineffective most of the time) to deal with albums that have been clipped by ReplayGain or (more likely) ReplayGain plus a positive pre-amp setting.

I was expecting to be able to set Peak Limit to 100%, Release time to something sensible (near-infinite would be nice, but quite fast would be good for some users), and put the compressors to 1:1. However, Release Time seems to be stuck at infinite, whatever the setting.

Can you shed any light on this?

Cheers,
David.

To put it another way: are me and richms doing something wrong, or has a bug crept in?

Cheers,
David.

The behaviour is intended, the goal is not to boost up background noise to extreme levels. Turn the input gain slider up, it should be enough to bring the quiet pieces into the amplification range (if not, I'll have to extend the range),

Peak limit determines the maximum output level ever used. I cannot reproduce your problem - it works as expected here.

For what you're doing, the Advanced Limiter would be much better though.

Uh? You mean as a peak normalizer? You'd have to disable all compression then. Not sure if this would work well.



This is the AGC. If you touch a slider, everything resets, and the AGC has to readjust. It doesn't turn the volume down to prevent clipping - it can compress after all. But it does correct everything to the same approximate level. Don't forget that if you have Input gain to 12dB and you click something, the first few seconds will go 12dB over peak scale!
But if I understand you correctly it seems that it doesn't properly adjust up again if the loud section is followed by a more quiet section? I certainly haven't noticed such a thing...

Thought so. Of course, this will only happen if the compression ratios are very high.

I tried this approach, but a) this does what the threshold is designed to prevent: increase background noise in completely silent passages (I have to use more than 10dB on the CD that I have problems with, and even then the very beginning remains unchanged), and b) it seems to me that the louder parts now get compressed more than I like. Don't know if this makes sense, I don't really know anything about dynamics compression.

Would it be possible to make it adjustable?

Yes, I was hoping that was the case!



I'll try re-installing it.




For me, I'd probably avoid both. But I was trying to find a "nice" solution for people who complain that ReplayGain makes things too quiet. I assumed they'd put the pre amp around +8dB. The behaviour I wanted was to leave anything which peaked below 0dB FS alone, but to compress anything above 0dB FS nicely. If you use the Advanced Limiter for this, you'll find that it make the audio unlistenable. It must be possible to get a better result, because it's quite easy to take a file with a ReplayGain of 0dB (i.e. a quiet track), and compress it so it sounds like a typical pop track with a ReplayGain of -8dB. That's effectively what someone who enables ReplayGain, and pushes the pre-amp to +8dB is wanting to do, but the advanced limiter just shreds the audio, and I can't make the DRC do the job either.


Like I said, I'll try re-installing it. It should be possible to make the dynamic compressor do this, shouldn't it? Or will other "unseen" parameters and processing prevent it from dealing usefully with stuff that is over 0dB FS on input?

Cheers,
David.

I'm seeing that too.

richms - does the "peak" slider work for you? if not, we could be seeing the same set of problems.

Cheers,
David.

Hello,

Quick question: I would like to achieve an effect "kind of like" Audiostocker for Winamp does, that is, set a whole bunch of songs from widely different sources and volumes in random, for background music while in the office, and not have to worry about the volume knob in my speakers. In this situation, I couldn't care less about the relative volumes of songs in their albums.

Any advice? I would think this plugin is the right one to achieve that, am I right?

That sounds more like trackgain's job to me...

To me it sounds like replaygain (trackgain) does what you want, but since I'm quite sure you know / have tried it already, I might have misunderstood what you want...

edit: oops ... (much) too slow

Yep, I thought so, but then I tried selecting all my playlist, right-clicking, Replaygain > Scan per-file track gain, and got a bunch of errors about not being able to update the info in the files... Any hints?

In any case, I would like to still use a dynamics compressor to avoid problems with songs that have big differences in volume (not hearing the quiet passages).

But of course, the thing about trackgain is going a bit OT for this thread...

Ok, I am stupid, I was completely sure I had already unset the read-only flag on the files on my hard drive, but it turns out I hadn't... Now the scan track-gain operation works fine... Not surprising...

If you have database enabled, RG info for files without tags that can store replaygain information (e.g. .wav files) will be stored in fb2k's database. The "error message" just tell you that the RG info isn't written to the file directly, but fb2k will still be able to use it.

edit: didn't see your last reply when writing this one...

Sorry about asking a stupid question but I want to listen to music in a noisy enviroment(for example doing the dishes while listening to music) and I want to hear detailes a little better, will the default setting do or could tweaking them be a good idea? Is the VLevel plugin better of worse than the Dynamic Compressor for the way I want to use it?

It's been a while!

I understand that:
The 'Fast Compression Ratio' has a fast attack.
The 'Slow Compression Ratio' kicks in after -24.3 dB and the attack speed is much slower.


What I still don't get:
The 'Fast Compression Ratio' will not work independantly of The 'Slow Compression Ratio'. How does this work?
And how does this dependancy apply to the 'Release Time' slider?

I guess for now I can ignore the 'Fast Compression Ratio' since it destorys transients but I'd still like to know how this plugin works.

I don't really understand the question, I fear. This plugin is really two compressors working alongside each other.

I use vlevel when compressing my music to vorbis to listen at work (I'm a projectionist, and the projectors make an surprising amount of amount of noise.) I also use it at home; It's nice 'cause you don't have to mess around with the volume nearly as much. It's really handy with some classical music that has what might be considered an "excessive" amount of dynamic range, with some parts whisper quiet, and other full-blast loud.

Sorry to revive such an old topic, but (I'm still using Foobar 0.8.3) I'm playin' around with the dynamics compressor and I'm seeing the same 'problem' as previous posters (oudalrich & 2Bdecided).

On a track which start with a low volume and the suddenly becomes much louder, I got the good behaviour: the first part volume is increased and when arriving to the louder part, the perceived volume is still the same.

On a track that starts with a louder part then a quieter one, the volume doesn't increase (or very few) when arriving to the quieter one.

I tried different settings, without success. Did I miss something or is it a limitation of this plugin?