Optimum settings for recording LPs with further editing (sample rate and bit depth)

[Dan McDonald]

not to confuse the matter, but I found that I hadn't actually enabled the soundcard in the computer I was using, so it was running based on the motherboard's sound device. I enabled the soundcard and still got essentially the same thing.:
Then I tried about 2 minutes of an lp at various sampling rates and looked at the files in the spectrogram Here's what I get:
44.1 - audio maxes out at 22kHz as expected.
48 - audio maxes out at 24 (but there is still information above that, according to the image)
96 - same as 48.
192- also maxes out at 24kHz
Then when I tried the 'ez impulse' set on gentle scratch & click remover, I got these results:
44.1= 2627 total fixed
48= 3137 total fixed
96 = 152 total fixed
192= "error: This filter cannot be used at this sample rate (192kHz). you will need to resample ot less than 192kHz to use this filter."
I resampled to 96kHz and got this: 62 total fixed
when back to the file and resampled to 48kHz and it found 2673 total fixed.
I then took the 48kHz file and resampled up to 96kHz. It found 70 clicks.
For a check, I then went back to the original 48kHz file and re-ran the filter, and it found 3137, exactly what it had found before.

When I listened to them all, I thought the 48 sampling sounded the best in terms of click removal. I also tried it using the ezimpulse before and after the VPA (this was with a flat preamp) and thought that doing the de-clicking before the VPA sounded the best (the end product).

Anyway, I know this is not a great test because it's not adjusted for each individual recording technique, but I did think it was interesting because I got worse results when I recorded at 96 or 192 and sampled down to 48 than I did when I recorded at 48. I was using the 'master cd' setting for the downsampling.

Dan

[Craig Maier]

Hi,

Interesting test - - - I am still thinking about it. My primary confusion is the bandwidth brick wall occurring at 48 kHz with a 48 kHz sample rate. The Nyquist frequency for that should occur at no greater than 24 kHz. Can you double check that file using the spectrum analyzer? Or, maybe that file is really a 96 kHz file (maybe that is how the soundcard software driver works)?

It certainly adds some confusion to the picture.

Craig

[Dan McDonald]

I'm sorry Craig - that was a typo. The 48kHz should have been 24kHz. It was 24kHz for everything above 44.1 I just fixed the post so that it is correct now.

[Craig Maier]

Oh, good. Looks like your soundcard has the same bandwidth limitation as does Marc's. Makes more sense now. Does 48 kHz work better than 96 kHz if you re-adjust the controls optimally? Maybe 96 kHz actually works the best with some adjustments?

Craig

[Dan McDonald]

Hi Craig -

It could be - I was noticing distortion on the higher sampling rates when I just let it run. It may be that I need to back off and make it gentler at higher sampling rates.
It gets complicated very quickly! I will experiment a little. I was looking at soundcards and trying to figure out which ones actually have 192 or 96kHz capabilities. I've seen some that say they have but it's really difficult to know for certain without buying the card and experimenting, I guess. I wonder if its a limitation in the drivers for the soundcards or something where the card is actually capable of recording, but they haven't implemented the software portion to enable it.
You and others on here would have a better idea of the issues than I would. I would like to be able to have recording at the sampling rates that are claimed though!

Dan

[Craig Maier]

Quoting Dan:

"I would like to be able to have recording at the sampling rates that are claimed though!"

--------------------------------------------------------------------------

Dan -

I agree - - - I do not think it unreasonable to actually obtain a digitization of the analog signal at the rate advertised. In the case of the Diamond Cut Impulse routines, it is clear that the algorithm(s) will take advantage of ultrasonic signals if they are present in the digitization. But, if the soundcard is pretending to be digitizing at 96 kHz and only producing 24 kHz top end frequency response limit, then that is sort of mis-leading. I do not know how ones goes about assuring that they are getting the true Nyquist values associated with the soundcard sample rates before making the purchase. But, at least we know how to verify the performance once we have our cards.

Craig

[Dan McDonald]

Yes...and I will when I get the next one. Maybe a birthday present. Anyway, I just ran the analyses and found that there was a major difference in the performance of the narrow crackle filter. You could be much more aggressive with it at 96kHz - even more than at 48kHz - without producing distortion. There was soe advantage with the scratch and crackle filters also. Also the same with 192 downsampled to 96.
Unfortunately, I happened to have recorded a rock band with electric bass, electric piano and drums, and a kazoo solo, which none of the impulse filters liked very much! They all had to be turned completely off. The kazoo was very close to the mic, and you could literally hear the cellophane vibrating, so I was asking a lot of the filters, but I'll figure out how to deal with that.

I'll continue to experiment with non-kazoo records to see what I can come up with.

Anyway, the potential upshot is much like Marc said - more aggression without distortion with the higher sampling rates, even if the actual frequency response isn't where it should be.
Dan

[Marc Hildebrant]

Kazoo solo ?

Well, that's quite an instrument for a rock band. Next thing will be a jug and washboard !

Marc

[Dan McDonald]

Yes, kind of different - the Youngbloods. you may remember them. Their big hit was "Get Together." Their albums tended to be kind of jazzy with occasional folk/Americana stuff.

[Craig Maier]

Youngbloods - now that takes me back - - - 1968?? Anyway, just thinking about the construction of a Kazoo, it is almost the definition of an Impulse Generator. So, maybe the filters are doing what they are supposed to do??


Craig

ps - didn't Country Joe and the Fish use a Kazoo on their song "I Feel Like I'm Fixin' to Die, Blues Rag" around the same time? (I think that was the name of that sweet little lulaby)

[Marc Hildebrant]

Youngbloods...My favorite was "Grizzly Bear".

Country Joe and The Fish...now that brings back memories (Bad mainly) of the Vietnam War. Feel like I'm fixing to die rag with the line " Be the first one on your block to have your boy come home in a box !"

On a better note : I have been trying to find published data from Sound card manufactors about the frequency response of their cards. So far, the highest frequency is 22 KHz . The sample frequency was 48 KHz. No mention of using 96 KHz ?

Marc

[Dan McDonald]

yeah, I was trying to find info also. Most of the time they are talking about ability to play back high frequency sound. I did find one that said 192 kHz for playback and 96 for recording, but then I don't know how accurate that is.

[Dan McDonald]

It would be nice if we had a list of soundcards that actually gave true numbers for recording. Maybe if we can get members to run the tests and see if anyone gets 192 or 96 kHz recording?

I have a little handheld digital recorder that is supposed to record up to 96kHz samplin. I will try to record some music via line in with that and see what it does.

Dan

[Craig Maier]

Hi,

Here are two for the list that do sample at 96 KHz and brick wall at ~48 kHz (these are in my 5 yr old system):

M-Audio Delta AP (PCI based soundcard on Workstation Tower)

Realtek HD Pro (Motherboard Chipset on Workstation Tower)

Sigma Tel Audio (Chipset on Dell Laptop)

I will keep an eye open for others to test.

Craig

[Dan McDonald]

thanks Craig. Just tried my little m-audio handheld recorder, and it provided true recording using 96kkHz, with lp sound all the way out to 48kHz (and it looks like above that, but decreasing. I think that I'll probably just use it for transfers and forget about a new soundcard. I do all my editing on a different pc anyway, and it has no trouble handling high sampling rates during playback.

I will re-run some of the de-clicking to see what happens. with 'true' 48kHz frequencies.

Dan