I know that completely undoing compression and limiting on cds like Californication is impossible, but what suggestions are out there that could improve the sound quality even a bit?

P.S. I'm using Audacity, so, maybe a little help there too.

Good luck. Maybe adjust the tone and bass a little, lol. Seriously there is not too much you can do besides that.

I haven't used Audacity before, but if it allows you to use plug-in's and you have some available, I have sometimes used a multiband compressor (like the waves C4) as an upward expander. But before I do, I will take the file (song) and destructively attenuate it about 6-8db. Allowing more headroom for your expansion. However doing this can be time consuming (but fun) and the results are not always as good as the original.

Sometimes I will look for an earlier release of a title before say..1992 when the loudness war on commercial CD's began. It's totally hit and miss. Some of the earlier releases can sound incredible or quite bright and flat due to the AD/DA technology at the time of its release. Hope this helps

Oh one other thing...if you don't have access to some higher end audio plug-in's, try some mild EQ, or better yet, I have found that upsampling to 24/96k, tends to warm things up a bit.

I am no engineer, but all you have is a 96Kb copy of a 24 Kb source. PLease be so kind to explain how it "warms things up a bit" and just what that means. Have you ABX'ed it? And, bad me, have you read the Terms of Service? Someone will jump on you for that claim of "warming."

<Satire>
I have heard that if you downsample from 96khz to 44.1khz, it puts everything "on ice".
Upsampling back to 96khz removes this ice.
Removing ice is analogous to "warming things up".
If you upsample too far, such as up to 384khz, things can get "too warm".
If sound becomes "too warm", you must use "cold digital gear" to compensate.

Similarly, if the sound becomes "too cold", you must either upsample further, or buy tube amplifiers in order to "warm things up". Both solutions combined tend to provide the best results.

Stereophile magazine in coordination with many audiophile hardware manufacturers have developed advanced algorithms to convert the relative apparent warmth of sound to an associated temperature.

The ideal temperature for the sound is calculated to be 310.15 Kelvin. This allows the sound a seamless transition to the cells of the cochlea. As the sound deviates from this ideal, problems arise and things begin to sound very bad, and may cause hearing damage. Therefore, many of the manufacturers have teamed up to graciously provide us with advanced devices to measure the sound-temperature for the purpose of making extreme profits preserving our hearing. After measurements have been taken, appropriate levels of upsampling and new hardware must be implemented in order to get sound that will synchronize properly with your ears via the ideal listening sound temperature.

As an added note---ABX tests are strongly discouraged, do to the very high probability of exposing yourself of sounds deviating from the ideal listening temperature. Please use a qualified testing device to calibrate temperature. Do not play games with your hearing!
</satire>

I almost missed the satire tag at first. Good job.

You could try running the stereo through Plogue Bidule, and by monitoring with RMS Buddy (shows average,peak, and ongoing dB level) on each channel,
adjusting the gain so there is no clipping, and recording to a new file it should help. Plogue also supports VSTs, so you can add/subtract or otherwise tailor to your needs, with some practice.
Although this site is about converting Stereo to Multichannel, check out dtsac3.com for some tutorials and
you can taylor the setups to your needs.
HTH.

boondocks

How can upsampling a signal to a higher resolution make music sound better when the source information is the same?

One might as well ask, how can a printed document with content at 300dpi be made to look better on printers capable of 600dpi? The answer in this case is that special software (eg, Resolution Enhancement Technology) interpolates which pixels on the edge of a character (for example) should be off and on, smoothing the edge that is printed to the finer granularity of 600dpi. Even though there is no more information in the source, the result is more pleasing at 600dpi.

Upsampling music from a lower resolution to a higher resolution should be able to improve sound in an analogous way. I have a dCS Elgar/Purcell combo that does this. My guess is that they interpolate the music signal by decomposing the music using fast fourier transforms which then allow them to 'fill in the finer granularity' intelligently.

In practice, I notice the difference most with CDs from the 1980s that were probably manufactured with lots of jitter.

This interpolation is exactly what oversampling DACs do with audio.



No, they "just" interpolate correctly and accurately.

There's no magic in dCS upconversion and DACs, they "just" do the job properly.

You can do the upconversion equally well on a PC with some free software. The DACs, however, are measurably in a class of their own.

Cheers,
David.

Does foobar's Output bit depth setting accomplish the same thing?

(edit) ...and if increasing bit-depth improves sound, why doesn't sample rate conversion?

I've had some success with redigitizing such CDs. First of all, I have an upsampling CD player (Yamaha). I play the CD at 2X upsampling (for some reason 4X offer the sonic improvement that 2X does) and through the player's analog outputs. I record it to my PC at 24/96. This tends to rebuild some of the transients that were lost with the brickwalling. Basically, I'm using the player's DAC to restore the waveforms. It's not perfect, but it does improve things quite a bit. I've also experimented with Cool Edit's Dynamic Processing plugin. I expand the frequencies above 1.5k or so and in some cases, this has really opened up the sound. It also helps to widen the stereo image somewhat, but only slightly.

The end result of this is frequently much more pleasing to the ear than the original.

Last summer I also reprocessed the entire Stadium Arcadium album by running each track through a CD de-emphasis filter and then an exciter plugin. This tended to tame the digital harshness and then restore some sparkle afterwards. You have to play around with some settings to get a good sound, but once you do, it's a lot easier on the ears than the commercial release.

Of course, then I heard the vinyl remaster of Stadium Arcadium this fall and I really understood just how good a modern recording can sound when they keep the compression freaks away from it!

I guess I fail to understand how DAC hardware can outperform any upconversion software algorithm when sampling rates are a digital property.

?!

The dCS Purcell is a hardware upsampler. You can do the same/similar job with foobar2k or SSRC.

The dCS Elgar is an exceptionally good DAC. It does the job of converting the digital values into an analogue waveform very well.

I don't think I understand what you were trying to say.

Cheers,
David.




Neither improves the sound quality (TOS #8!).

All modern DACs oversample. The resampling in software like CEP, fbk2 and SSRC is measurably superior that used in most oversampling DACs.

Cheers,
David.

Re-emphasizing my use of the word 'if'. I never made claims that it did in my experience, though I see how it could have been interpreted that way. Apologies for not being clear.

What I'm trying to understand is why bit-depth increasing is recommended in the wiki and upsampling is not. What benefit do I get from outputting CD rips at 24-bit/192kHz to my home theater speakers besides bragging rights that I'm able to do so?

Sorry, I thought you were saying that oversampling DACs outperformed software upsampling. Rereading your post, I realize that's not what you were saying.

No, it's me not being clear! I wasn't invoking TOS#8 at you. What I meant to say is that I wasn't making any claims that it sounded better, because I can't ABX it (I haven't actually tried for many years!), and the note in fb2k itself is probably trying to make the same point: if you think it sounds better, you're probably imagining it.


If you're playing back 16-bit 44.1kHz sampled content without any processing, and the DAC is ideal, then increasing the sample rate or bitdepth will make no difference to the output quality. If the DAC is non-ideal, then upsampling in software can improved the measured performance. If the DAC is terrible, then ABXing this imrpovement by listening is possible (especially with torture signals!).

Increasing the bitdepth is necessary to avoid a measurable decrease in quality if you're going to do any processing, e.g. EQ, ReplayGain, DSP, digital volume control, convolver etc etc. Most maths produces more bits, so it makes some sense to keep as many as possible. Otherwise you have to re-quantise the output of these stages back down to 16-bits, which is rather like dividing two by three, and then rounding to the nearest whole number! This is only ABXable on specific (mostly test) signals, but certainly improves the measured performance.

Cheers,
David.

Back to the original topic:

It looks like there are a number of plug-ins available for Audacity, and the LADSPA package has both gate and expander effects. I don't know enough theory to make sense of the "SE4" Effect parameters all too well, but it may be worth exploring for somebody who knows what they're doing.

You can't really "undo" compression on a mastered CD. However, you might find that you can color the sound in ways that please you more via various plug-ins for expansion, altered EQ, etc.

I don't believe you are getting any closer to the "original" sound (whatever that might be) but instead it will be screwed up in a nicer way. Just like several girlfriends I had many years ago...