Topic: COMP: Oade Warm Mod Marantz PMD661 vs Zoom F3

[Billy Mumphrey]

Picked up an extra pair of Line Audio CM4's locally, giving me a total of 4 of them as well as the opportunity to run a couple recorders head to head. I ran a pair of CM4's into a Marantz PMD661 with the Oade Warm modification (24/48) and right underneath those, the other pair of CM4's right into a Zoom F3 (32fp/48). All mics were at 90 degrees with 25.5cm of spacing in between capsules, using Movo windscreens. Pm for the results or I'll probably upload a spoiler txt file later on. No editing except some level adjustment to get them close. Dithered/noise shape -> 16/44.1 > flac.  I was very thorough in making sure to not get any of the files mixed up. 

https://samply.app/p/AQz3hJVGLYUqPV9nCZKo

Notes:
-I haven't had the practice of setting levels in well over a year due to 32fp. I start the PMD661 during soundcheck and the levels are spot on (peaks at -12), don't even need to change them haha.
-The above samples are from a favorite young local jazz band Daruma, who have wonderful original arrangements. There were 3 more bands that I would normally be totally down to record and preserve for all of time, but I just wasn't feeling it and skipped out after the opener. Felt (kinda) bad but I'm home early and I doubt anyone even noticed. I always try to record everyone, but I've been taping a lot lately and sometimes you gotta stay fresh.

Quick sound thoughts:
-flavors

EDIT: If the mics look a bit wonky in the pics, it's because they are. I took this during soundcheck and later on I adjusted all mics so they were identically spaced and angled.

[grawk]

First, thanks for doing this, I love equipment comparisons.

Second, why did you go to 44/16?  Why not leave them at 48/24?

[Billy Mumphrey]

First, thanks for doing this, I love equipment comparisons.

Second, why did you go to 44/16?  Why not leave them at 48/24?

You know, you're right actually, it's better to leave out any unnecessary steps or conversions. 99.99% of my recordings are for non-tapers and non-audiophiles so I'm just so accustomed to saving to a standard file size.

I went ahead and added NATIVE FORMAT files to the comp. The files labeled "native format" have not had any dither, noise shape, or resampling applied to them. Just a slight level adjustment to match each other, and that's it. I also took great care to not mix up any of these files, for it's actually fairly easy to do!

I've already had a couple people in PM's guess right. Good ears! I can BARELY tell in my car speakers, but it's pretty apparent on studio monitors, especially A-B ing.

[admkrk]

First, thanks for doing this, I love equipment comparisons.

Second, why did you go to 44/16?  Why not leave them at 48/24?

You know, you're right actually, it's better to leave out any unnecessary steps or conversions. ...

Technically speaking, a fair comparison would be to process both of them equally (or as equally as possible), so actually, no. Processing one and leaving the other is not a fair comparison. You did it right in the first place.

Just saying.

[grawk]

Going from 48/32 to 48/24 after setting levels isn’t a destructive process. Dithering to 16 and converting to 44 is.

[admkrk]

You are confusing sampling rate with frequency.

Going from 32 bit to 24 bit w/out dithering is no different than going from 24 bit to 16 bit w/out dithering.

https://www.izotope.com/en/learn/what-is-dithering-in-audio.html

OK, I just realized I am confusing terminology. For sample rate, I meant bit depth. Any time you lower that, you are introducing errors. The link I provided will make that clear.

Sorry for not being clear, I am a bit tired.

[grawk]

You are confusing sampling rate with frequency.

Going from 32 bit to 24 bit w/out dithering is no different than going from 24 bit to 16 bit w/out dithering.

https://www.izotope.com/en/learn/what-is-dithering-in-audio.html

I'm not confused.  32bit float and 24bit both have more headroom than either recorder is capable of utilizing.  You absolutely can just convert 32bit float to 24bit without needing to dither, because the noise floor for 24bit is -144db. On the off chance there was material down there that was truncated, it would never be heard.

[admkrk]

This has nothing to do with noise floor.

https://en.wikipedia.org/wiki/Quantization_(signal_processing)

[grawk]

https://taperssection.com/index.php?topic=153333.msg1940767#msg1940767

That summarizes what's going on pretty well.

For the purposes of recording what we record, there's no difference between 24bit fixed and 32bit float.  Once you normalize, your entire 32 bit datastream is inside 24bit, and the remaining bits are unused, because you're not scaling.

[admkrk]

... Just saying.

I don't think it really matters. Reason, a floating point number holds info like xxxxxx*10^yy (mantissa and exponent). 32 bit floating point is stored in a computer cpu as 23 or 24 bits of mantissa and the other bits are exponent. Saving to 24 bit is basically keep the mantissa and forget the exponent.  So it's not at all the same thing as going from 24bit to 16bit.  I expect those checkboxes in the software screen don't mean much at all in this case, but possibly make the software do a lot of exta math for nothing.

Although this is an opinion, just as mine is, this assumes all software manipulates bits in the same way. I have written programs that manipulate floating point numbers as fixed point and I can assure you that there are many ways to go about it. Tossing the exponent is not necessarily the way every program will handle it.

[spyder9]

My guess:

1 is 661

2 is F3

[Billy Mumphrey]

I went ahead and posted a spoiler txt file with the results.

For the "native format" files, I opened the Zoom F3 file (32FP 48kHz) in Reaper, made a small adjustment to volume, and rendered it as a 32FP 48kHz file (the dither buttons are greyed out, so there is no change to the bit depth or sample rate). For the PMD661 file (24-bit 48kHz) I opened it in Reaper, made a small adjustment to the volume, and rendered it as a 24-bit 48kHz file (the dither buttons are greyed out, so there is no change to the bit depth or sample rate). Then I flac'd them and uploaded them. Just wanted to be clear, since one is 32fp/48 and the other is 24/48.

I've always wondered how software is actually handling a 32fp file. Reading your guys discussion made me realize how little I know lol. However I find it pertinent to a comp with 32fp files tho and welcome all knowledge.

Next I would like to put a pair of CM4's and CM3's on a stand and run both of them into a Zoom F6. Maybe my 2nd pair of CM4's into an MP2 > F6, see how some transformers sound.

Spyder9 that's your warm mod 661, thank you for selling me that and an MP2.

For the record, I enjoy both of the recordings in the comp.

[spyder9]

Just read the results.  Wow! 

[fanofjam]

Caveat that I listened through my laptop sound card, so not a very good system.  I felt sample 2 was the slightly better sounding sample.  Seemed to have a notch better presence, which is something I appreciate in a jazzy recording with lots of hi-hat and subtle cymbal use.  Overall impression though is that the difference between the two is not very discernable and definitely not enough to say that one is the clear winner over the other...but then again like I said I didn't listen through a very good system.

Thanks for the comp.

[grawk]

Doug's a solid engineer, but the difference between different IC based preamps isn't gonna show up in a casual listening environment