Expander in the future?

I am in the process of transferring a number of LPs to CDs. I have an electronic expander (dbx II) in my system that I use to restore the dynamic range of LPs but recording would be an easier process (with probably a better quality result) if the expansion could be done in the digital realm. Any chance of an Audacity expander effect in the near future? I see that it was discussed here in 2010 but it appears it did not make it into 2.0.5. Or am I just missing how to do expansion with an existing effect?

You mean one of these:

That was designed for use with tape not vinyl. The idea is that you compress the audio before you record it, then expand it back to normal on playback. dbx (noise reduction) - Wikipedia

I’ve never heard of a DBX system for vinyl.

Normally one would use a phono-preamp (often incorporated in the phono stage of an amp) to restore the RIAA equalization

WC

I have not tried this, but [u]Chris’s Compressor[/u] can compress or expand.

I’m sure there are plenty of VST compressor/expander plug-ins if you search.

[u]GoldWave[/u] ($60 USD after free trial) includes a compressor/expander, but it’s a little screwy (sometimes it thinks it’s compressing when it’s expanding). There are expander presets including “Boost Loud Parts” or “Reduce Quiet Parts”, so it’s not too confusing is you start with one of the presets.

…(with probably a better quality result) if the expansion could be done in the digital realm.

In general yes… With a digital file, you can have an attack time of zero, or even negative (AKA “look ahead”) if you wish, and it’s a lot easier to “undo” and try a different setting.

With software it’s also easier to process it twice in case you want to upward-expand the peaks, and also downward-expand the quiet parts (or expand-down the noise). (Audacity does have a noisegate.)

However, it can be difficult to get the exact same results as you’re getting with a hardware/analog compressor expander. There are some (usually expensive) VST plug-ins that emulate some famous compressors, but I doubt you’ll find one that emulates the one you have.

My dbx unit is a 118, not the 128 pictured. It can expand or compress but in order to use it as a tape noise reduction system you would need to compress during record and then change the settings to expand during playback. Think of it as “the left side” of the 128 unit pictured. The function of the 118 was to deal with playback - compress signals if you wanted less dynamic range (like for background music or news programs on the radio) or expand signals if you wanted more dynamic range. Back in the day the dynamic range of vinyl LPs was less than that on the studio master tapes; the signal was compressed to deal with things like tape hiss and groove noise. One of the big selling features of CDs when they first came out was adding 10 dB or more of dynamic range compared to a vinyl LP.

Using a dynamic range expander set at something between 1.2 and 1.4 restored the dynamic range of a vinyl LP to what was possible on the studio tape. You can also set this unit for either threshold expansion (to deal with signals that had been limited like a radio station that ran their signal against their limiters) or linear expansion for the compressed signal seen on vinyl LPs and pre-recorded cassettes.

I looked at “Chris’s Dynamic Compressor plugin for Audacity” and it appears to be a compressor only. Am I missing something?

I looked at “Chris’s Dynamic Compressor plugin for Audacity” and it appears to be a compressor only. Am I missing something?

Compress Ratio can be a negative number like it says in the slider panel notes.

Koz

One of the things that sealed Dolby’s success is the inability to hear it while it was working. It was transparent. Just about all the other compander systems produced artifacts and odd sounds or would in some way fail to restore the original show at the end of the chain.

If you don’t have an iron grip on the compression process at the beginning, it’s almost impossible to restore the show in Post. You can simulate it, similar to the “Opera Concert Hall” experience in Windows 7 and Windows 8. It’s not a real concert hall and it’s not real expansion. It’s volume changes where the software (or hardware) thinks would be nice.

Koz

One of these?




There is a plug-in that you could try here:
dynamics.ny (2.84 KB)
This plug-in is very new, so no instructions yet.
Negative values for the “Expand / Compress” control give expansion. Positive values give compression, so that is the opposite way round to the DBX II 128.

Installation instructions are here: Missing features - Audacity Support

Yes, the dbx 118 pictured is my unit. And yes it is not as good as Dolby because as kozikowski said you don’t know how the compression was done so you can only guess at the expansion. The dbx unit either operates linearly (expands or compresses the signal above and below the threshold ) or expands/compresses only the signals above the threshold. This may not reflect the behavior of the compressor that was used in the recording process.

Note that the dbx unit expresses expansion and compression as ratios - a 1.4 expansion means that the peak signal is amplified 40%. This is a different measure than the -.5 to 1.25 compress ratio in the Chris’s Compressor plug-in. Can anybody compare the numbers such that an expansion ratio of 1.4 on the dbx relates to a compression ratio of -0.XX on the plug-in?

There’s nothing simple about Chris’s Compressor. It’s a dynamic, look-ahead compression manager he wrote so he could listen to opera in the car. He started with arpeggios and glissandos and worked down instead of starting with algorithms and subroutines and working up. So his compressions are very pleasant and easy on the ears, but have only the tiniest relationship to the original.

I have found that increasing the compression from the default 0.5 to 0.77 and the compressed sound is almost identical to the local radio station broadcasting the same show.

Illustration: Original show, 0.5 and 0.77.

http://kozco.com/tech/audacity/pix/ChrisCompressorBeforeandTwoAfters.png

Koz

No, because it is not a fixed ratio in Chris’s compressor. I presume that you determined that “between 1.2 and 1.4” produced the result that you most like by testing different settings. For other expanders you will need to use the same process to determine which settings you most like.

Like I said, I think it’s impossible to get the same exact results you are getting with your hardware expander. (with any software or any other hardware compressor/expander).

Even if you do get the same settings, you’d have to calibrate the levels & thresholds (Volts or dBV/dBm on the hardware unit and dBFS in software).

If you can’t get the sound you are looking for (by trial-and-error), Google “VST Expander Plug-In” (or give GoldWave a try).

Every compressor/expander is different, but “traditionally” you should have these settings for a compressor/expander:

Ratio - The amount of compression/expansion. Mathematically, 2:1 is compression and 1:2 is expansion. I assume 1.4 on the DBX unit means 1:1.4 for expansion and 1:4:1 for compression.

I don’t think a negative number makes any sense (mathematically), but 0.5:1 would be the same as 2:1 (compression) and 1:0.5 would be the same as 1:2 (expansion).

Threshold - The level at which compression kicks-in. Typically, you compress the peaks. i.e. If you set the threshold for -6dB, everything above -6dB is compressed and everything below -6dB is untouched.

With expansion, you may have a choice of expanding above the threshold (peak expansion) or below the threshold (noisegate or downward expansion).

Attack - How fast the compressor kicks-in after you cross the threshold.

Release - How long it takes to return to linear operation after you cross-back across the threshold.

Make-up Gain - An adjustment used to bring the peaks back-up to 0dB after compression. (It brings-up the overall level, so the recording is louder after compression).

With expansion, make-up gain could be less than one to bring the peaks back-down below clipping.

…There may be other settings or features, such as a hard or soft “knee”, which determines how the compressor/expander behaves near the threshold, or you might have a noise-floor setting (another threshold) so that boosting the volume with compression doesn’t boost the noise level during quiet parts, etc.


Using a dynamic range expander set at something between 1.2 and 1.4 restored the dynamic range of a vinyl LP to what was possible on the studio tape.

FYI - There is no “automatic” compression on LPs. Compression and limiting are up to the mastering engineer. If you get good results with 1.2 or 1.4, that’s good. But, you can’t “reverse” what was done because you don’t know what was done. You don’t know the ratio, threshold, attack, or release… Mastering engineers often use multi-band compression (more unknowns).

Compression anhd limiting on modern recordings is not done because of limitations of the medium (CD or MP3), it’s done for esthetic reasons (they think it sounds better, or at least it sells better if it’s constantly-loud). Often, a modern CD and LP comes from the same master, but the LP and MP3 will sometimes measure more dynamic than the digital version. (Both of these make some peaks higher and some lower, without increasing the perceived loudness of the peaks.)

You can also set this unit for either threshold expansion (to deal with signals that had been limited like a radio station…

Limiting presents another problem… There is no way to know if that 0dB peak was 0dB, +6dB, or +12dB before it was limited.

Yes, but I’m looking to go the other way - to expand the dynamic range of the music. I guess I will just have to experiment.

In the plug-n that I posted here Expander in the future? - #8 by steve a ratio of 2:1 is at a compression / expansion of about 25%. Try a setting of around -10% (10 % expansion).

I played around with Chris’s Dynamic Compressor and figured out that it is not by setting the compression to negative that you turn it into an expander but by setting the Floor to it’s minimum value. Then I set the Compress Ratio to 0.3 and it appeared (visual inspection of the waveform height) to behave about like the dbx II 118 expander set to 1.3. 1.3 was the consensus number for restoring the dynamic range of vinyl records produced in the 60s and 70s. I understand this is really just eyeballing but in essence it is no different than what many record producers did in those days. The quality of recordings was, to put it generously, quite variable so I don’t feel I’m way off base in adding one more recording engineer to the chain.

At any rate thanks to everyone for leading me to this. I hope my experience is helpful to anyone else who is going down this path.

One of the things that sealed Dolby’s success …

I’ll have to strongly disagree with you here. Dolby is/was successful because of one thing: MARKETING.

The dbx was a far superior system, but not first-to-market:
The compression.expansion cycle (they called it a “compander”) is “linear”. Dynamic range was compressed at a factor of 1.4 (for tapes), and expanded at the same rate. Now, you could expand it at 1.3 or even 1.5 and most people would probably be hard-pressed to notice that it was mis-calibrated.
In contrast Dolby “b” was a non-linear system, wherein the compression began only at two thresholds: above a certain frequency, and below a certain amplitude (volume) level. This means that Dolby “B” required careful calibration. If it was off you certainly could hear it!

(This is why most tape decks had a “MPX Filter” switch to filter out the 19kHz pilot signal in FM broadcasts that would play havoc with Dolby B but not dbx. This pilot signal is what turns on the “stereo” indicator on your tuner.)

One of the things that sealed Dolby’s success is the inability to hear it while it was working.

I am not sure what you are trying to say here. The internet is full of people who say something like “Dolby sounds like shit” because they fail to understand that it’s a two-way system that requires encoding AND decoding. "

While it is working" … if you had a three-head tape deck, you could monitor the signal on tape, but dolby decoding would be applied first. Same goes for dbx. So I can’t parse any sense from this statement.

If you don’t have an iron grip on the compression process at the beginning, it’s almost impossible to restore the show in Post.

The dbx units also had an “above threshold” switch that allowed you to peak-limit audio signals, but you would never use that for tape noise reduction. For the obvious reason that careful calibration of the “threshold” would be required to re-expand them.

It’s true that once people started using different compression for different frequency bands, the opportunity for re-expand dynamic range was pretty much impossible.

Having said all that, I too would like to see a flexible dynamic range tool for Audacity. I came here because I was looking for one :slight_smile:

What does “linear” mean in the context of dynamic compression? What is happening to the gain?

If you mean that the combined effect of “signal → compression → expansion => signal” is linear, then isn’t that the case with all balanced compander systems?

OK I explained that badly. Here’s another try:
If you take a graph, plot “Input Signal” on the horizontal axis, and “Gain” (EE term for amplification) on the vertical axis.

A regular amplifier is a straight horizontal line - same gain for any input level.

A dbx compressor would be represented by a downward sloping line : less gain for higher input levels.
A dbx expander would be expanded by an upward sloping line: more gain for higher input levels.
This is what I mean by “linear”: the line is straight.

For Dolby B, you will see a “dog-leg”.
During encoding, the line is at first downward sloping, then level.
During decoding, the line is first upward sloping, then straight.
This is because tape hiss is mainly apparent at low levels; remember we are concerned with signal-to-noise ratio.
Where the “knee” is, there must be careful calibration:
-to the tape recording/playback levels
-between recording and playback.
This is what I meant by non-linear.

For Dolby B, this is further complicated by frequency frequency: once the above knee is crossed, the higher frequencies (only) are compressed and expanded.

Both systems are/were effective. The dbx is so simple it could be used in an “outboard” device.
The Dolby system requires more complex electronics and calibration, thereby also violating the “KISS” principle…

And now you know more than you ever wanted about Dolby vs dbx. Sorry about that :slight_smile: