sample rate & bit depth choice

[mikeebb]

First - thanks to Craig for his advice, and sorry about naming other names - won't happen again.

For economic reasons (can't afford to blow $couple hundred on a decent sound card right now), I have to use my motherboard audio. It's not bad (better than the cheap cards I had in past computers) - can go up to 96/24 for playing DVDs. However, 96/24 works for recording only if nothing else is running on the computer, and the file sizes are so big that I don't have enough processor and disk speed to crunch through them in a reasonable amount of time when filtering. So I've been using 48/24 with downsampling to 44/16 at the end, and it sounds pretty good to my less than Golden ears when working with 78s recorded at 45 then speed-converted in DC7 and from very used LPs. Would 96/24 be enough better that I should put up with the processing time and file size problems, given that the final product is a standard CD and that the audio is after all on the MB (and subject to its noise floor on top of the mixer currently used for its phono preamp)? If I could do 96/16, would that work better than 48/24?

Thanks.

[Craig Maier]

Hi,

For 78s, it is my humble opinion that it is more than sufficient to master with 16 bit depth and 44.1 kHz sample rates. All of our commercial releases were done that way, and they are well accepted by the consumers of our CD products.

At most, I would use 48 kHz and 16 bits. Keep in mind that 78s only have a useful bandwidth of around 10 kHz (I wrote an article about that here somewhere sometime back - - - record speed and bandwidth).

I just found it:

http://gator1104.hostgator.com/~edis...cord+bandwidth

That means that only 20 kHz is really needed, but 44.1 does the trick nicely and 48 kHz will provide a little more discrimination of the rise times of the clicks and ticks.

As for resolution, 24 bits is crazy for 78s since they only have around 35 dB at most of dynamic range. 16 bit give you 96 dB which is a ridiculous amount of head-room for a 78 RPM recording.

I hope that helps.

Craig

[Doug]

What about when recording 78s flat and adding equalization during the restoration process?
I have been recording them at 44.1kHz, 24/32bit (whichever is available), processing them with the same parameters before dropping back to 16bit following equalization.
Is that also overkill in terms of bit depth?

[Craig Maier]

Quoting Doug:

"What about when recording 78s flat and adding equalization during the restoration process?"

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

The Dynamic Range of a 78 is at most around 35 dB. If you record flat, that actually subtracts from the signals dynamic range because the bass is attenuated. Unlike the LP, there is no Rolloff, so no concern. If we were talking about an LP with Rolloff, that would add the requirement for another 20 dB of dynamic range (the opposite effect compared to Low Frequency turnover).

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

Quoting Doug:

I have been recording them at 44.1kHz, 24/32bit (whichever is available), processing them with the same parameters before dropping back to 16bit following equalization.
Is that also overkill in terms of bit depth?

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

It can't hurt, but I do not think that it is helping in any way. The Edison Lateral Cut Archive was transferred at 16 bit depths.



Craig

[Doug]

hmmm....might change my tact then!

[mikeebb]

Thanks! Going with 44/16 will definitely save some disk space and processing time for the 78s (piles of my Dad's from the 1940s and 50s, many German). I'll probably use 48/24 for LPs given your comment about the rolloff.

-mike

[Dan McDonald]

It seems to me that the software does take advantage of higher sampling rates when you're dealing with very scratchy material, so I prefer a higher sampling rate to start with, but it could be my imagination.

I typically use 24-bits when I know there is going to be a lot of work in cleanup (i.e., nasty stuff to start with, like really worn/heat deformed cylinders), but it's probably overkill.

I just transferred 4 lps for a friend of my son. He'd had them since the 70s. I used 24-bit, 96 kHz for transfers, but they were in such good shape, I could easily have done it with 16-bit, 44.1. I just figure that, as long as I have the space, and I don't know what I'll be getting into until I've already done the transfer, I just allow as much freedom as possible.

Dan

[Craig Maier]

Quoting Dan:

"It seems to me that the software does take advantage of higher sampling rates when you're dealing with very scratchy material, so I prefer a higher sampling rate to start with, but it could be my imagination."

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

Dan - my answer was focused in terms of 24 bit depths, and so that is what I was referring to and not sampling rate. I do understand a logical reason why using a higher sample rate may be helpful, especially with LPs, but less with 78s. As for higher bit depths with LPs, I think that the benefit is marginal since the dynamic range at best with an LP is going to be around 60 dB and then adding in the RIAA Rolloff slope, you pick up another 20 dB (actually, 19.6 dB) thus requiring only around 80 dB of dynamic range to capture the signal with a flat preamp.

Craig

ps - I guess you can say that level settings are less critical when using 24 bits.