I have read that a better quality recording from vinyl can be had if it is done at a higher rate and then converted down. Anyone have a procedure or a how to on that?

You must have a 24bits capable soundcard (not an Audigy 1), and specify 24 bits when you record the vinyl in CoolEdit, SoundForge, etc.
Once the recording is done, ask the program to convert your file from 24 to 16 bits.

However, I listened to 48 kHz 16 bits vs 96 kHz 24 bits recordings of vinyls and could not hear any difference, even without converting them. Furthermore, the background noise of a vinyl reduce the accuracy of the signal to about 12 or 13 bits to begin with. Thus I don't think it is worth.

I think that this is untrue. There is a tangible difference between CD-quality (44.1khz, 16bit) sound and 96/24 quality. Of course, this is still PCM data and doesn't compete with DSD, but that's neither here nor there. Also, the quality of the system from which you record directly determines the necessity of greater resolution. If you're just running a standard, consumer-level set of equipment, chances are that you're never going to be able to hear the higher quality that you'll get from higher resolution. However, in my knowledge in the past you do get a boost in quality while processing audio at higher resolutions (compared to CD-quality), and then down-sampling them. If you're not going to be processing/cleaning up the audio, however, there's really no need. Also as Pio pointed out, you'll need a sound card that truely resolves at 96khz/24bit, and for that you'll most likely want to go with either a pro sound card, or perhaps the M-Audio Revolution (think they have an even more "spiffed-up" card available, but I'm not sure). But, to summarize: if you're just transferring directly from vinyl to CD, don't worry about a higher sampling rate/word length. If you're going to be doing a good deal of complex audio processing (including rice-crispy noise removal), I'd suggest that you use a higher resolution for as long as you're actually processing the audio. As always, this is just my $0.02, and you're welcome to do whatever you feel or think is best.

Yes, I forgot about processing. In this case, it can be worth. It doesn't harm.

BTW, why cool edit cannot operate at 24-bit editing mode? It only supports 32. Is it for floating point reason?

Yes, it just automatically uses floating-point, but your sound card obviously will limit things to 24-bit resolution (or whatever your soundcard actually supports). Sound Forge supports 24bit recording/manipulation, I honestly couldn't tell you which is better without really looking into it, though. I'd think that 32bit takes longer to process, but of course you won't lose any quality there.

Thanks for the input everyone! I have an Audigy 2 although I don't seem to have software that directly relates to it. The sound level meter with the Audigy is almost useless. I'll give a look at Sound Forge although I do prefer CoolEdit and Audio cleaning lab.
I wonder if CoolEdit will process it at 96/24?

Yes, Indeed! but it will be edited in 32-bit mode. However, like what silver_cpu has said, there will be no quality loss just take some harddisk space. You can save it at 24-bit if you want.

I think there's no advantage at all in recording at 96 KHz, even for any kind of processing. To capture frequencies up to 48 KHz (fs/2) won't give any advantage in processing, unless you want to simulate some kind of nasty intermodulation effects happening with inaudible frequencies.

In case of vinyl, I don't think 24 bit are a real advantage over 16 bit, if you set properly the recording levels. However, 24 bit recording could allow you set the recording levels more loosely, because of the wider dynamic range to play with.

As to 24/96 vs 16/44.1, there has been quite a lot of discussion around that here. Still, I don't know of any properly executed ABX or double-blind tests that prove there's an audible difference.

I am just starting to work with it. Sounds good but is somewhat difficult to work with. It may be a matter of getting used to it.

There's a theoretical advantage to using 32-bit in Cool Edit, and a practical one.

The theoretical one is that, just maybe, you could do enough processing so that the dither noise (added repeatedley at the 16th bit) could become audible. You'd have to do a lot of processing of some very good vinyl, but many people claim to hear a difference. I'm not vonvinced. It's very slow using 32-bit compared to 16-bit.


The practical advantage is that the 32-bit processing in CE gives (at least) 24-bit accuracy, but with floating point processing. This means that, once recorded, you don't have to watch the levels while you're processing. Your sound card is limitted to fixed point (16 or 24 bits) so at play and record you have to be careful not to clip. But the actual 32-bit processing within CE cannot clip, and does not have a lower limit either. So you can neither distort the signal, nor push it down into the noise.

If, after processing the signal does go above 0dB FS - just reduce the level - the signal is still intact.

If, after processing, the signal is way too quiet - just increase the level - there's an (almost) infinite amount of low-level detail, all stored to 24-bit accuracy.


If you don't believe me, use the 32-bit mode to try boosting or cutting things by 200dB - they survive perfectly. Now try it at 16-bit



You may think that the added luxury of not having to "watch the meters" makes it worth using the slower 32-bit processing. Personally, I (usually) don't, though it's sometimes useful.

You still have to watch the meters on record, and if you play back audio that's over 0dB FS in the 32-bit domain, it'll still clip in the 16- or 24-bit domain of your sound card. It's just the processing that doesn't suffer, meaning you can always correct any clipping you create, without undo/redo.


Cheers,
David.

2bedecided, this is great information. I will have to study what you said.
To see if I got it right........I can record in CE at 32 bit without regards to volume. That is "max" out the recording slider and the sound will not clip. If recorded to high and in simple terms the "volume" of the wave can be adjusted with no harm to the music.

bigjohnson:


No. Your sound card will do the recording and it will only have 24-bit or 16-bit fixed-point with a fixed maximum positive and fixed maximum negative amplitude, so it can clip and cannot be recovered if it does. It's critical that it doesn't clip, so usually aim for peaks in the music to be about 2 dB below full scale, or preferably allow more headroom if you have a low-noise soundcard (use the record level meters). That's because the sound card operates on physical signals (electrical).

CoolEdit can only overcome the physical limitations once it's all in the software domain, and might then store it in 32 bit floating point representation, which allows you to exceed what would have been full scale while processing (which happens when filtering and amplifying or mix-pasting two loud sounds together for example). So once you have a good unclipped signal it will stay unclipped during processing. (Foobar2000 audio player uses floating point internally too, for the same reason)

You can also check you soundcard to see whether it has more noise when multiple inputs are enabled in the Mixer (e.g. microphone input, CD input, Auxilliary input) and can help by muting all except Line In.

Because you can save at 24-bit, you could also consider using Foobar2000 diskwriter to finally "master" your audio CD down to 44.1 kHz/16-bit by applying ReplayGain Album Gain with clipping prevention and doing the conversion down to 16-bit with your choice of dither (and resampling down to 44.1 kHz if necessary and converting mono to stereo if necessary for old mono records).

This is a valuable and good advice, FB2k quality in this respect is as good as anything you can use.

Thanks, I understand now. As to Fubar, I will give it a look for the cd portion. I have been using it for flac and such. It is a good all around player and compressor.