So I guess CDDA is rough and not natural. I put an audio track ("Patience" by Guns N' Roses) into Audacity, made sure it was editing at 32-bit, upsampled it to 48 KHz, applied dynamic range compression, and then normalized the levels. Afte that I exported the track as a 24-bit WAVE.

My goal is to find a way to get 16-bit CDDA tracks to sound like they were taken from a 24-bit source.

So, is there really an audible difference between the two different bit depths? And, is this a good or retarded way of trying to make it sound more natural?

I'm using a ThinkPad T60 with a SoundMAX 1981HD. It supports up to 48 KHz at 24-bit.

Why?

Cheers, Slipstreem.

Ditto

Paul

Sounds like a recipe for squeezing most quality out of 8-bit.

If you want it to sound as close as is technically possible to the original master recording, leave it alone.

Cheers, Slipstreem.

In England

HERE HERE

me

THERE THERE

Paul

That doesn't make sense at all.


That should make it sound worse than the original but it is a matter of taste.


This sentence doesn't make any sense.


Not audible.


Retarded way.

You're not getting good responses, because both your goal, and your process are flawed. You can't improve the sound as it already exists on the CD. What you're doing can only make it sound worse, at almost twice the file size. And what the heck does "rough and unnatural" mean? You should read the terms of service on this board... we don't allow or tolerate unsubstantiated, unmeasurable claims.

Here, I'll help by specifically addressing each step:

1) Resampling to 48kHz is unnecessary. It increases the file size with no benefit. Actually, your sound card automatically resamples EVERYTHING to 48kHz on playback anyway, so all you're doing is wasting hard disk space. Resampling will not increase the frequency response or anything.

2) Don't apply dynamic range compression. A lot of playback devices can apply it on the fly, but I wouldn't alter your source tracks. Most modern recordings (not the GNR though) are over-compressed to begin with. Ideally you only want compression when you're a) listening at low levels because you don't want to wake people up, or b) in a noisy environment (car). You're removing dymanics! I thought you wanted to IMPROVE the sound!

3) Instead of normalizing and applying DRC as above, if you really want all your tracks the same volume (and want the changes made permanent, not just at playback) use WaveGain on your source waves. Then the volumes will actually sound the same REGARDLESS of whether the source is highly compressed or not. And it won't throw out dynamics like applying DRC will.

4) 24-bit output is pretty pointless... you're literally adding 8 zeros to the end of each 16-bit sample. Wasteful!

So my recomendation is to leave them alone, or if you really want the volumes equalized, use wavegain, output to 16-bit (and make sure you turn on dither/noise shaping.) I think applying gain adjustments at playback makes more sense though; you can do this by encoding your files to FLAC and using metaflac to tag them with replaygain tags. Most FLAC players will read these and adjust volume on the fly at playback. Plus you'll shrink the files by around 40% with no loss in sound quality at all.

If you're just burning them back to CD but want them normalized, go the wavegain route i mentioned.

Perhaps this wasn't what you meant, but the output itself is not useless. With 24-bits you can better realize volume control, for example.


lol.

{Perhaps this wasn't what you meant, but the output itself is not useless. With 24-bits you can better realize volume control, for example.}

Nope - volume control is done in the analog domain, bit depth has no effect on volume control.

Well, I guess that wraps up this little fantasy.

It's all part of the learning process.

Cheers, Slipstreem.

If anything that I really notice that makes all my music sound a lot better is using the "Noise Sharpening" component in the Foobar2000 player. It's found in foo_dsp_delta.dll. Maybe give it a try.

http://pelit.koillismaa.fi/plugins/show.php?id=92

I would be interested if someone could explain what technically goes on within this DSP? Is it some sort of dithering or what? If it's some sort of witchcraft, I don't want to know. It's better left unsolved.

I also use the vlevel DSP found here: http://vlevel.sourceforge.net/news/ Which helps brings out the quieter parts.

There is no fantasy. If Exponent always outputs unaltered 16-bits and uses a hardware volume knob, it's simply his choice.

If you were an android (with super hearing, like data from star trek) you could go in really close on a 16 bit waveform in adobe audition and hand correct the "blockyness" of it, I think. But you have to be an android and you would have to export it to a higher bitdepth.

This would make no sense as you would be correcting something that is already perfect. There is no point in conversion to 24 bits unless you decrease the sampling rate as well (you would be sacrificing frequency response to a better amplitude resolution). This is what delta-sigma converters do - they take a high-frequency, low (1-bit) resolution data and convert it (through filtering and decimation) to a high-resolution, low-frequency one.
You simply can't sqeeze out more information without "borrowing" it from somwhere else.

The "blockyness" of it?
I mean is 2^16 discrete volume levels (65536) blocky?
Is this a legitimate problem for some people? Are there really "Datas" out there?
Or are there people who wish to believe they are "Data"?

Concerning the frequency: my CD player's DA converter is supposed to do "4x oversampling", which means to me that it multiplies delta by 4, shaping a 176.4khz signal out of the cd's 44.1khz. Of course that won't squeeze any occult information out of it, so I did never understand the sense of that feature... I don't think that DSP engineers would be so stupid to develop such a complex feature if it were completely useless, though.

Put an oscilloscope on the output of a CD player and you will see no so-called “blockyness” there.

Paul

Using oversampling in the data converter simplifies the design of the reconstruction filter on the output - making the design of this filter much less critical to the performance of the converter.

As paulhoff pointed out, there is no blockyness on the output of a DAC. Despite what your audio editor might show, the output of a CD player/sound card/iPod is not "blocky".