ATRAC[3[plus]] Frequency Analysis

[Avrin]

There already were attempts to analyse frequency responses of various ATRAC bitrates using various material. Now I decided to put it to the heaviest possible test by using white noise. I tested both hardware and software encoding.

Equipment used:

1. eMachines D620 laptop PC running Windows XP SP3, SonicStage 4.3 "Ultimate", and Adobe Audition 1.5, for software encoding, uploading, and analysis.

2. Panasonic DVD-S75 for direct playback via optical out (no enhancers).

3. SONY MZ-RH1 for recording via optical in @ manual rec level 23, and for uploading.

The track used was 12 seconds of 44.1 kHz/16 bit stereo white noise generated in Adobe Audition, and burned to a CD-RW using EAC 0.99 prebeta 4.

Let the pictures speak for themselves:

PCM (green), Hi-SP (red), and Hi-LP (blue) recorded via optical in

PCM (green), ATRAC3plus 256 kbit/s (red), and ATRAC3plus 64 kbit/s (blue) ripped and encoded by SonicStage in High quality

ATRAC3plus 352 kbit/s (green), 192 kbit/s (red), and 48 kbit/s (blue) ripped and encoded by SonicStage in High quality

SP (green), LP2 (red), and LP4 (blue) recorded via optical in

ATRAC3 132 kbit/s (green), 105 kbit/s (red), and 66 kbit/s (blue) ripped and encoded by SonicStage in High quality

ATRAC3 132 kbit/s (green), 105 kbit/s (red), and 66 kbit/s (blue) ripped and encoded by SonicStage in Normal quality

Edited May 4, 2009 by Avrin

[sfbp]

Thanks. Why didn't I think of doing that???

The interesting thing to me is the comparison of LP2 (132kbps) on the last two. It appears to be way better on optical in than via SS. Is there any possibility this is a function of the setup? Of course it's obvious that the reverse effect applies on HiSP (SS) vs 256K optical recording. That doesn't make sense to me (yet, at least). I do know that the LP modes seem to deliver somewhat random quality depending on how they were arrived at. Weird, isn't it?

[Avrin]

The interesting thing to me is the comparison of LP2 (132kbps) on the last two. It appears to be way better on optical in than via SS. Is there any possibility this is a function of the setup?

I think that this is a function of the ATRAC Type-S chip in the RH1.

[sfbp]

Don't understand that. Maybe it's not what you meant to say? Type-S is playback-only, so it shouldn't affect what your Panasonic does, right?

[Avrin]

IIRC, Type-S is used for both playback and recording in LP modes, same as Type-R is used for playback and recording in SP.

[sfbp]

Hmm....

my understanding was based on the reference for the first deck that deployed it, the JB980. It is indeed possible that Type-S manages a better job on the encoding as well as on the playback. However until this moment I was under the clear impression that it is LP2/4 playback only.

I wonder if your experiment proves this impression to be simply wrong?

[greenmachine]

The cutoff frequency doesn't say much about the resulting sound quality, unless set way too low. Most adults can't hear beyond 15kHz anyway.

[Avrin]

I wonder if your experiment proves this impression to be simply wrong?

We need to obtain a frequency response of white noise encoded by a non-Type-S machine to check this. But I don't have one.

The cutoff frequency doesn't say much about the resulting sound quality, unless set way too low. Most adults can't hear beyond 15kHz anyway.

Exactly! Especially when listening through headphones. But when using good loudspeakers higher frequencies may create additional harmonics in the environment, which add naturality to sound.

One the other hand, the less information remains to encode after filtering out higher frequencies, the better the remaining frequencies may be encoded. But listening shows that software-encoded ATRAC3plus 256 kbit/s sounds quite a bit better than hardware-recorded Hi-SP. This may be simply because a PC has much more processing power than a Hi-MD unit, and uses more advanced encoding algorithms.

Edited May 1, 2009 by Avrin

[sfbp]

We need to obtain a frequency response of white noise encoded by a non-Type-S machine to check this. But I don't have one.

I do!

I'll attempt to match your diagrams in the morning. However I have only CoolEdit (Audition's ancestor) so no pretty colours.

[sfbp]

The cutoff frequency doesn't say much about the resulting sound quality, unless set way too low. Most adults can't hear beyond 15kHz anyway.

This (and Avrin's reply which I am too disorganised to quote in here) highlight something I stumbled upon about a month ago. Lossy codecs are great if and only if (IFF) you don't convert them, some conversions being worse than others. Not sure why, but please read on.

I have been recording stuff (classical music) from internet radio, which is generally transmitted at 64kbps. So it makes sense that I should be able to reencode and save at about that bit rate. Namely LP4.

Furthermore it sounds quite decent on the computer speakers. It sounds great in the car. To my surprise (I had played with ATRAC3 LP4 briefly a long while back and said "this is rubbish") it doesn't even sound any worse than the same thing recorded direct to much higher bitrates, when I play it back on the stereo. Somehow I assumed that maybe leaving lots of bits there would prevent further lossy conversions.

The trick seems to be to leave it completely alone at the LP4 bitrate. As you may know the LP4 encoding is quite different from LP2, so in retrospect I shouldnt expect to be able to go backwards or forwards (up- or down- convert) without making a mess, even though the underlying rate is about 64k off the radio.

This all started when I bought a 600D off someone who left their collection of downloads intact. That sounds good, thought I, and even though I erased them (wasn't my kind of music) the idea to try LP4 was at the back of my mind for a while.

Here's what I do currently -

1. record off radio (optical out to optical in) from realplayer to LP4 on std MD in JE640.

2. Mark the "songs" using a portable's track mark button as I listen to it again.

3. Label the names using Sonic Stage (or the PS/2 keyboard with the 640, but SS is easier as I can keep the playlist on the screen and type or cut-and-paste from it to the MD).

4. (optional) Label all the announcements with short names (a,b,c,d etc) and delete them using SS (remember the multiple-"Untitled"-delete problem).

(at this point I have a disk I can play in the car)

5. Upload the tracks I want "as is" ie LP4 at both ends, to SS, using the RH1.

6. Download the LP4 tracks (again, "as is") to a HiMD. This way I can get 32 hours on a single 1GB disc.

I think my previous impression (I won't use the word "observation", as the experience wasn't well controlled) of LP4 was based on being much less systematic than this.

These recordings once uploaded from the LP4 disk look like they are quite versatile.

And yes, the disk also sounds decent in the Type-S playback deck in my stereo. Not as good as CD->SP but that is a given, due to the original quality of the transmission over the internet. I was *really* surprised by some organ music recently.

So the comment about it not mattering much about frequencies above 15khz may be quite on point. I agree that there are some harmonics missing. But done as I describe there are none of the horrible artifacts which I am pretty sure I observed, when downconverting SP recordings I had made, to LP4. Maybe Sony even did their best to protect us from that because they didn't want us complaining, and felt that anything they *did* allow should have a good standard.

I wonder if some comments made here about :needing: PCM arise from a similar scenario.

[bluecrab]

We need to obtain a frequency response of white noise encoded by a non-Type-S machine to check this. But I don't have one.

Exactly! Especially when listening through headphones. But when using good loudspeakers higher frequencies may create additional harmonics in the environment, which add naturality to sound.

One the other hand, the less information remains to encode after filtering out higher frequencies, the better the remaining frequencies may be encoded. But listening shows that software-encoded ATRAC3plus 256 kbit/s sounds quite a bit better than hardware-recorded Hi-SP. This may be simply because a PC has much more processing power than a Hi-MD unit, and uses more advanced encoding algorithms.

I recently upgraded my integrated amplifier. It seems to me that LP2 MDs now sound either much better or much worse, depending on how good/bad they were to begin with. The good ones, though...really, it's easy to forget I'm listening to such highly compressed audio. I think the Type S helps, too. I am hearing stuff on LP2 that previously, I didn't notice on SP or in somes cases, perhaps even on CD. I'm old enough so that the high-frequency dropoff applies to me. In fact, I even have some diagnosed hearing loss. Almost all my MDs are recorded using standalone gear.

[garcou]

Good work avrin.

My hearing experience in term of audio quality is : PCM ~ atrac3+ 352kbps ripped with sonicstage in high quality (slow encoding) > optical atrac3+ 256kbps ~ atrac3+ 352kbps ripped with sonicstage in low quality (rapid encoding) > atrac3+ 256kbps ripped with sonicstage in high quality (slow encoding) > atrac3+ 256kbps ripped with sonicstage in low quality (rapid encoding) ~ optical atrac SP 292kbps

Edited May 1, 2009 by garcou

[sfbp]

OK, here's what I did to try your experiment, Avrin.

First I made a CD with 12 seconds of white noise. I made the amplitude "10" with Cooledit and I left the other parameters alone (500 uSec stereo separation I think).

Note that to make this CD playable, I was forced to add some other tracks after it. The computer was happy to play it but neither Sony CD player would recognise the disk with a single track.

Then I played back the 12 seconds of noise and recorded onto a std MD at the 3 different rates, a total of 13 times:

Setup A. Computer -> JE640 (via optical cable from sound card to optical in)

Setup B. Computer -> RH1 (via optical cable), AGC (gave slightly lower amplitudes)

Setup C. MXD-D400 deck CD -> MD (normal speed dubbing)

Setup D. CE375 CD changer -> MXD-D400 (optical cable)

Setup E. CE375 CD changer -> JE630 (optical cable)

I did NOT try the D400 in high-speed dubbing mode.

1. Setup A, SP

2. Setup A, LP2

3. Setup A, LP4

4. Setup B, SP

5. Setup B, LP2

6. Setup B, LP4

7. Setup C, SP

8. Setup C, LP2

9. Setup C, LP4

10. Setup D, SP

11. Setup D, LP2

12. Setup D, LP4

13. Setup E, SP (that's all the JE630 has)

I would be willing to post the .OMA files if someone would like them.

I labelled these tracks on the MD and uploaded them from RH1 to PC via USB, concluding with automatic generation of .WAV files. I then opened each .WAV file in CoolEdit, selected the portion of it with the sound (only) and applied Frequency Analysis followed by Scan. The results were as follows (allowing for slight differences in recorded amplitude especially in Setup .

(One mea culpa, I imported the SP tracks at 256K rather than 1411K. Apparently MDLP tracks will be imported "as is" regardless.)

i. All the SP tracks looked like this, pretty good up to 20kHz, as advertised.

ii. All the LP2 tracks looked like this. As you can see the response is flat up to 16kHz then drops off fairly smoothly to 18kHz below which it is essentially silent.

iii. All the LP4 tracks EXCEPT the JE640 looked like this. The response up to 14kHz shows this characteristic bumpy pattern (an artifact?) and then a very fast drop.

iv. The LP4 track from the JE640 looked like this. Same as iii. except that it drops much sooner after 12kHz, with the attenuation being substantially complete at 14kHz.

For some reason the encoding of LP4 appears not as good on the JE640.

I would have liked to find a way to test Type-S playback directly. However my MXD-D400 is not easily routed to the PC.

Oh yes, one more thing - I used the same MD for all the tests, a TDK PoSHE gold 80m.

[Avrin]

Added some really strange stuff to the very end of my initial analysis.

Edited May 4, 2009 by Avrin

[sfbp]

Added some really strange stuff to the very end of my initial analysis.

Do you mean the last picture? I don't even recognise Normal quality vs high quality, maybe you or garcou should explain.

I can maybe imagine something like what I described in my post about LP4 in this thread, that saving at higher bitrate (upconversion) may make sound worse - as you said, the lower the cutoff, the better the SQ of what remains?

[Avrin]

Normal (Faster) and High are the values of the Recording Quality parameter in SonicStage. You may also set the quality via the registry. That's what the "CDRecordMode" DWORD parameter of the

[HKEY_CURRENT_USER\Software\Sony Corporation\SonicStage\CDRipper] key does - the value of 1 corresponds to Normal (Faster), and the value of 2 corresponds to High.

Minidisc Simple Burner uses Normal quality only.

Edited May 4, 2009 by Avrin

[sfbp]

Sure, I can edit the registry. Sigh.

But

1. your experiment seems to indicate "High" doesn't improve anything

2. I can't find the parameter you refer to, unless it's arrived at by some combination of other choices. Heeelp

[Avrin]

Just press the "Briefcase" button on the CD-ripping screen in SonicStage. You'll see the Recording Quality drop-down box.

[sfbp]

Oh, importing CD's. I don't often do that, and never noticed that menu when I did.

So what is your explanation? Many people here seem to swear by SB, so if it uses the lower quality, this is all very interesting.

Something else odd, I noticed. LP2 is fine on all NetMD. LP4 (and this is probably why I never used it before) sounds bad on the MZ-N910 (one of the last NetMD models?) but much better on a HiMD player such as the NH700. I did an A-B comparison, same disk, same headphones/remote, and the difference was dramatic.

[sfbp]

Something else - I've just done a little study. It started out as a project to pack all my Bach sacred choral music (I have about 14 CD's all by approximately the same set of forces - Leonhardt/Harnoncourt and some others occasionally) from the TelDec set, onto a single MD. I wish I could afford the entire set, but life would be sooo boring if we could afford everything we wanted.

There appears to be no way to record directly from CD onto "HiMD" (ie std md disk in HiMD mode and 1GB disk) in anything other than PCM, HiSP (256k) and HiLP (64k). I had noticed that stuff that I recorded using my JE640 MDLP deck, at LP2 and even LP4 sounded quite good after uploading and saving to HiMD format, leaving the codec as ATRAC3 (132k and 66k respectively). So I thought, let's take CD's, upload the lot as PCM (1411k) and then use SonicStage to output at LP rates to the HiMD. Naturally I picked LP2 since this would just about include all of the content I wanted (limit is 16 hours 30 minutes). So I did this.

I was very disappointed in the result. The sound is mushy and unacceptable, probably explaining why a couple of years ago when I first got my RH1, I discarded MDLP in favour of Atrac3+ for saving large amounts of music.

So now I got creative, and decided to record from CD to a standard MD at X1 (times 1) using the MXD-D400 which features MDLP rates, direct to LP4 - this being the closest I could find to the recordings I made from internet radio that sounded so good at this rather low bit rate.

So I stuck 5 CD's in the changer, connected up the optical out to the D400's optical in. Presto, everything except titles, of course. Now that's only 1/3 of what was on the 1GB disk, but so what? I can always upload and save to that later, since as far as I can tell SonicStage is smart enough to leave MDLP well alone unless you tell it not to. This would give me a massive 32 hours 50 minutes if I really wanted it, on a single MD.

Imagine my surprise when I found out that the same pieces recorded on the D400 at lousy bit rate (66kbps) sounded BETTER than than same thing recorded from library-PCM to LP2 (132kbps after conversion) over USB.

I'm willing to post samples if you like but I refrained initially from doing so because I feel sure that I'll end up posting something that doesn't reflect the difference I am hearing. Maybe someone can tell me what to post. If it won't fit here, I can easily put it on a webpage, but if this turned out to be right we would probably want it available here on the MDCF pages.

Oh yes, the comparison was made in EH70 using RM-35ELK and standard short-lead headphones from Sony. I swapped the disks in and out many times before believing this result. I'm sure there are many variations on what to test but this on its own should be enough to "prove" the point.

Maybe I do want to know how to use Simple Burner to burn Atrac3 onto HiMD!! Can it be done???

[Avrin]

Simple Burner can only burn ATRAC3 (LP2 and LP4) to conventional 80 min. discs in standard MD mode. But SonicStage can transfer 66, 105, and 132 kbit/s ATRAC3 in Hi-MD mode. Here's a way to make SonicStage rip CDs directly to 66 and 105 kbit/s: http://forums.minidisc.org/index.php?showtopic=15928 - I have just added quality selection there.

File conversion in SonicStage and music transfer in Simple Burner are always done in Normal Quality.

Edited May 25, 2009 by Avrin

[sfbp]

Thanks. I will try repeating the transfer using High quality 66kbps and see if it matches the D400's result. Will changing that registry setting also alter what LP2 transfers do? Seems so weird to me that LP2 sounds bad.......

1/2 hour later: Got it. Finally LP4 sounds as good as what I made at x1 and LP2 sounds way better than that. I'm sure this was the basis on which I junked LP2 as a format when initially using SS and the RH1. After all Sony should be happy that I used a higher bitrate - more MD media sales

So the step which is critical is the importing step. What strikes me as weird is that importing the same disk at 1411kbps (which doesn't have a quality setting at all) and then converting **that** yields such an awful result.

[Avrin]

That registry settings only apply to 66 and 105 kbit/s. LP2 (132 kbit/s) is configurable from the SonicStage interface.

[sfbp]

I just double-checked - the WAV import still is bad. Does this mean that you cannot import WAV files at highest quality? Very subtle way to restrict us.....

At a stroke this explains so many of the oddities I have seen over the last 2 years since I got serious with USB and MD. There is no way to get the highest quality off CD except by making a PCM file yourself using some other tool (eg recording from optical and capturing into .WAV). If you do that, then you actually have master-quality files in the PC. Otherwise, not.

[Avrin]

The best program to get bit-perfect copies of your CDs as WAV files on your hard drive is EAC. But you need to set it up according to the parameters of your CD drive.

[sfbp]

Hey - I'm not obsessive about this. Acceptable is good enough. Thanks for the tip, maybe I will chase that program down. But I am certain that x1 always worked (and RH1 uploads are identical) into the sound card optical in. It just means this is useless for CD ripping. I can anyway handle by throwing them onto D400, just doesn't directly get me onto HiMD.

But when I can get a better result by importing to 66kbps than by importing 1411kbps and downconverting in software, something is terribly terribly wrong. Finally I'm mad at Sony - for crippling something so darned obvious.

[sfbp]

Postscript: I finally ripped 16 hours of CD's again at 132kbps and it sounds absolutely wonderful.

WHY OH WHY would Sony say that ripping as WAV is "highest quality" and then do this to us?

I also wonder if you folks who listen to PCM exclusively (Wiz and others) actually do their ripping some other way. As far as I can see, importing at WAV using SonicStagemeans you get garbage.

Do tell

[sinus]

I also wonder if you folks who listen to PCM exclusively (Wiz and others) actually do their ripping some other way. As far as I can see, importing at WAV using SonicStagemeans you get garbage.

Do tell

I don't have the feeling but on the other hand I normally record real-time.

[Avrin]

Recording to PCM in realtime via optical in does not produce bit-perfect copies, as may be seen from the first two screenshots in the first message here.

Edited May 26, 2009 by Avrin

[sfbp]

I *do* recall noticing that (before I finally learned how to make RH1 work at high speed, thanks to you and your Russian chums) there seemed to be no way to get good music into SonicStage. I noticed it as soon as I tried it, back on my first portable, in 2005. So I went back to my original methodology:

- record onto SP minidisc

- x1 playback from there into TOSlink

- save using CoolEdit (later WaveRec)

- load into SonicStage from WAV files

- convert to MD compressed formats.

SS makes it very difficult with all the copies of files lying around and it's very easy to get rubbish. Sometimes it even picks the bad copy from its store of "optimized files".

Once I found my way through the maze, I never bothered much again, until very recently when someone (maybe you) made me realise that the same MD (80 mins) could store the same format (LP2) more efficiently. It always bothered me that stuff that I uploaded never sounded as good. Now I think I understand why, and as of yesterday I can finally achieve what I first tried to do in 2007 after getting the RH1. As I pointed out, Sony directly or indirectly sold me a lot of media since then - neat conspiracy, huh?

[kino170878]

I have a feeling that your bad results regarding WAV files is due to effectively converting the files twice - first to WAV and then again to Atrac. If you keep WAV as WAV (PCM) without doing additional conversion then the file should sound near perfect.

Incidentally I have read that converting bit rates in this manner will break gapless playback. You have to keep to the original bit rate (and possibly even rip the CD in one go) while keeping tracks 'as is' if you want to preserve the full dynamic range in Sonicstage and on MD. Any deviation from this procedure - whatever the bit rate chosen - can result in problems.

[sfbp]

I have a feeling that your bad results regarding WAV files is due to effectively converting the files twice - first to WAV and then again to Atrac. If you keep WAV as WAV (PCM) without doing additional conversion then the file should sound near perfect.

Incidentally I have read that converting bit rates in this manner will break gapless playback. You have to keep to the original bit rate (and possibly even rip the CD in one go) while keeping tracks 'as is' if you want to preserve the full dynamic range in Sonicstage and on MD. Any deviation from this procedure - whatever the bit rate chosen - can result in problems.

Right, so the observations about gapless playback (which I never cared about) fit nicely. What you're really saying is that the fast rip is good enough to the ears until you try to convert it to lower bit rate, at which point it screws up. IOW the fact that certain people find 1411kbps PCM the only "good" format is really an artifact of Sony's little trick. They're being tricked, as I was. If you do it right, you really only need a much lower bitrate, and the 292kbps of SP is easily good enough for listening on speakers. For portable listening, one can go to LP2 (or as I have recently been astonished to find, LP4) and the quality is amazingly good, once you get rid of the "double conversion" problem.

It also explains the odd glitches that everyone including me has seen usually at the beginning of HiMD tracks, up to 1/2 second of audio. That's essentially a "feature" not a bug of this diabolical scheme to prevent us from actually copying, whilst preserving the illusion of quality.

I for one don't actually care if my copies are bit perfect. All I ever wanted was something that sounded good enough.

I also noticed that high-speed copying on the MXD-D400 resulted in garbage on one particular disk. The CD in question happened to be a re-release of a CD, produced by ArkivMusic after the fact. It looks like such disks are like the Sony rips - not to be relied upon.

When I did it again at slow copying speed, it was fine. That in turn suggests to me that 3/4***(see below) of the data in PCM is undetectable by the human ear, which is why SP sounds "good enough". If you follow my rather twisted logic.

Cheers

added:

*** 3/4 referring to the fact that you can copy at x4 without evident losses. This corresponds to 1411/4 = 352kbps, the highest rate offered in HiMD. However for CD->SP it's actually very close to 4/5 (1411/5 = 282, but if you allow for 1024 bits/k maybe this is even closer to the 292 of SP).

[Avrin]

Ripping and recording quality effectilvely depends on the manufacturing quality of the CD in question. Some CDs can be ripped at higher speeds with bit-perfect results, while some require slower speeds or even multiple passes with error recovery (that's what EAC does). And some cannot be ripped without errors at all.

Edited May 26, 2009 by Avrin

[sfbp]

My results seem quite consistent with an across-the-board degradation introduced by Sony. Perhaps they will say that management was pushing for high rip speeds or something. The fact is, when the same disk is ripped to one of Sony's proprietary codecs, rather than PCM which is only a small step away from what's on the CD, the sound is wonderful.

The other oddity is that they never pushed a data rate around 128kbps on HiMD. I am finding the LP2 sound is just perfect for portable listening (it always seemed it on NetMD recorders) with the Ipod-like ability to hold a huge amount on a single 1GB disk. Yet noone at Sony appears to have anticipated that anyone would need to put LP2 onto HiMD as a matter of routine, and that rate is not there in Simple Burner.

(or did you tell us all a way to do this, and I missed it?)

I recall writing a post saying that the Type-S support for Atrac3 was perhaps missing from HiMD. That is clearly wrong - the difference is in the initial upload. Using AAL as the target on the PC, LP2 always sounds great. I haven't tried LP4, but the non-transcoded uploads of things actually *recorded* at LP4 sound very good, when played back under HiMD.

[Avrin]

The other oddity is that they never pushed a data rate around 128kbps on HiMD. I am finding the LP2 sound is just perfect for portable listening (it always seemed it on NetMD recorders) with the Ipod-like ability to hold a huge amount on a single 1GB disk. Yet noone at Sony appears to have anticipated that anyone would need to put LP2 onto HiMD as a matter of routine, and that rate is not there in Simple Burner.

(or did you tell us all a way to do this, and I missed it?)

No, there is no way to use LP2 or LP4 for Hi-MD in SimpleBurner. Bitrate lists (PCM/Hi-SP/Hi-LP/48 kbps for Hi-MD, and LP2/LP4 for NetMD) are hard-coded into the program, and the bitrate registry setting is just a number from the corresponding list. You can't use these bitrates for a different mode, and you can't enable any other bitrate.

[Strungup]

Avrin , thx for noticing in your first post , that the RH1 had a slightly lower input/output ratio ........... The fact that someone else noticed this solved an issue for me . I have noticed , that stuff I recorded on my R50 always sounded " Hotter " in fact pegging the meters of everything else I have . My Denon DMD-D50 in fact will drop out at times from some of my Earlier LP discs from my R700 ( I think that is what I recorded on ??? ) I thought something was wrong with the deck at first , ....... but I began watching the meter more closely during playback , and noticed the dropouts coincided exactly with " OVER" lighting up on the meters ..............

This kinda sux ......... anyway I have a Teac M5 MarkII showing up later this week , so I will see if there is a difference there .

But this is disconcerting as that Audio Levels , SHOULD be consisitant from within one companies gear , and especially with Recording gear .

I wonder if you have a way , to measure the Levels to see if the input / output is DB accurate ? I appreciate the Freq response test , very telling indeed .

But DB accuracy is also important , as that when your recording from Mic , or Mixing Board etc , what you se on the meter should be exactly what you are able to get into the Computer , That way you get the proper mental picture of what editing your going to do before you even start . If the levels are not correct , then you have to go about that first , before any serious editing like effects or ........ well you get my point .

I have an Older Denon SP only deck that is solid as a rock , and outputs a very solid Optical signal as well . But it is SP only .

I didnt pay attention to the levels until you mentioned that . I thought something was wrong with the other Denon . but it isnt , it is Levels are wrong .............. I have a LOT of discs in LP mode .......... So maybe I will sell this Denon for someone who doesnt mind a lower level .