Monday, June 20, 2005

HD5's Got A Secret

Having been voted the best sounding MP3 player must be a publicity coup...

...but, there's a rather calculated side to this apparent victory.

Take a look at the graph below*.


This reading was taken with all EQ turned off on the HD5. It's a shame I can't scale these graphs more to indicate the sort of impact it might have on hearing, but while for example the iRiver H320 has basically a flat response in the 20-100hz range, the NW-HD5 has a (very audibly noticeable) shaped boost. To implement a curve like this, you don't do it by accident or by inappropriate component choices. It's engineered.

The NW-HD5 therefore makes a mockery out of any listening tests with 'flat EQ settings'. I'm not trying to debate whether the bass boost makes things nicer, it is simply that like-for-like comparison must be just that, and the Sony tuning of the 'flat' sound is... well, for want of a simpler word, cheating. For example, would the HD5 still sound better in that "test" if the Cowon X5 was given a slight boost in the lows? You know, I think it might lose. Would slipping in '02' bass boost on the H320 make the iRiver sound 'better' to the test panel? Quite possibly. The big difference between the Sony and the rest is that the other players don't play this trick, so comparing with no EQ is a reliable test of sound quality for all but the Sony. Even the iPod with the now infamous fall-off has a flat response when loaded with high-impedance phones.

Although I make no definitive judgements which I'm prepared to publish here about the sound quality of the HD5, it is perhaps not quite as clean as at least two of the players on that comparison list. I put it to the CNET staff that they were fooled by the bass boost of the HD5... frankly, I'm quite surprised that no-one picked it up. I did within 10 seconds of hearing a track I was familiar with on the HD5, not as an improvement in quality but as simply a boosted bass.


*Usual disclaimers apply... not guaranteed accurate, author indemnified against inaccuracies, yada yada yada. ATRAC3+ 256K & 320K LAME MP3 codecs, loaded with 16 ohms impedance and adjusted for typical listening levels.

12 comments:

Anonymous said...

In engineering circles, a 0.25dB deviation in an amplifier-transducer transfer function does not qualify as "bass boost". Have you taken into account the typical characteristic impedance and reactance of the test load over the audible frequency range wrt a typical 30mW driving source as the likely culprit? The only "secret" here is, perhaps, the questionable training and qualifications of the person performing the test.

A little information can be a dangerous thing...

bangraman said...

In hearing circles, the mid-bass increase is noticeable, as is the increase at the lower bass. I trust my ears first, then go for testing.

This is only the second time that I've posted RMAA results. I don't pretend to be an expert at this, but I do offset my lack of experience through some dilligence. A little information can be a dangerous thing indeed, so I've done as much as I can to make sure this doesn't fall down under such problems. As with the reviews I post up, there's usually more to what I'm writing than sitting down with something for half an hour and then writing based on that.

I was aware that measuring a range of devices with similar power outputs versus the lower (although not that much, I think) output of the HD5 might yield skewing results. And yes, I was briefly concerned about the characteristic impedance of the test load. So I expanded the test to devices ranging from 1.5mw up to 55mw rated output into the test load. Firstly, the test load has been previously measured by other parties to have a fairly consistent impedance throughput the frequency range, although those tests are now offline. Secondly, test devices outputting 1.5mw and 2mw respectively did show a flat response with with the nominal 16 ohm test load and a nominal 70 ohm test load.

There were quite a few which were affected by a bass roll-off due to the nature of the load, but the HD5 is the only player to behave like this.

"an engineer", may I suggest that you correlate the changes in the HD5's response as shown to actual heard experience and see if you don't think it makes a difference to even the average joe's perception in a psychoacoustic sense. This is not a challenge to your presumable expertise in this field, rather that I would like to know if there is any other reason why the HD5 measures different as in the graph, when even other Sony players players measure flat under the same conditions.

The differences may be vague but hearing preferences are often based on vague things. I hear a difference and it is noticeable in a way that would affect preferences. I personally didn't get the impression of higher quality... just different. It took me a couple more seconds to work out that it was a bass increase.

I'll try and post up more results later.

Anonymous said...

In human circles, Sony HD5 produces some boom tunes which cause the eardrums within me to resonate with glee. My "secret" is that org*sm is nearly always imminent.

bangraman said...

Ah, but whose to say the iRiver with subtle bass boost will not give you even more glee? I'm sort of regretting posting this now as people seem to keep interpreting it as 'additional bass = bad'. That's not the point.

Anonymous said...

I rather the bass boost be separate and the "No EQ" or "Flat EQ" option be as flat as possible. I love bass, but there are times you want a completely flat EQ; adding a bit of bass on to the flat EQ and then having a seperate bass boost function too is pointless.

bangraman said...

Indeed, anonymous.

The 'secret' (artistic license in the title) is that there is a shaped curve on the 'supposedly flat' setting. It may be very subtle but it may have an effect on relative perception especially to audiophiles, and it may even influence untrained ears in a non-quantifiable way.

I'll stress that I didn't go out to 'reveal' anything from the outset. It's just the HD5 sounded different to me in a way that was divorced from sound quality, so I turned to RMAA for some visible results.

Every other player I measured showed a flat response at the non EQ setting, or measured with a bass falloff due to a non-optimal choice of dc blocking capacitor for the test load. Even the players which had a falloff eventually measured flat when loaded with the 'optimum' impedance load... but not beyond.

The Sony is arguably a less clean source than the majority of the other players listed in the CNET comparison, so in terms of sound quality it may not able to deliver as good a result if it was on a level playing field. But the subtle flavouring may be an influencing factor for testers and would-be buyers.

Obviously I can't say with any authority that the sound was shaped to do this, but it kind of smells that way where I'm standing.

Anonymous said...

he is a secret close your eyes and trust your ears muice is ment to be heard and felt and studied thru o scopes and spectrum analyzer. did it every occur sony did in fact engineer the product to sound "the sony way" because most consumers buy on what looks better what sounds better after all sony is in it for the dollars

Anonymous said...

harp all you want. i had an 20gb ipod. i now have an hd5. after the eq adjustments, the hd5 sounds better. show me whatever graphs you want, but at the end of the day, i'm on subway going home after another day at work and my tunes sound fuller.

and the battery lasts three times longer. and when it finally dies, i won't be asked to pony up another $100 and send it to apple.

screw ipods. they're there for people who were looking for an mp3 player two years ago or don't do their research before they buy electronics. let them experience the hassle from apple when their recharagable battery finally dies. go ahead, ask them why they didn't put a battery door on the thing. i dare you.

Anonymous said...

It probably is engineered in and that's what bangraman is pointing out. No this isn't fair when trying to do a comparison at flat settings. Nor is it something you would want for a studio setup.
However that doesn't make it a bad player. It most likely is there to sell more. Personally I'd rather adjust those settings myself though. There are times when I want to hear something the way it is intended, not the way sony thinks we should.

Anonymous said...

Well, I do own a HD5 and I can confirm that this Walkman ist sounding "spectecular". Fuller, richer, nicer and stronger than all other players. If you could label this performance "accurate" depends on you.

What I noticed:
Using lower ATRAC bit rates blows the sound. Even using 256Kbit makes me feel this: slightly more bass "woom" and stronger than on the source material - even having all the EQs off. Recently I ripped my favourite CDs again by using ATRAC3Plus 352Kbit - wow: Now it sounds SPECTECULAR AND ACCURATE. Very transparent, somehow soft and tenderly, especially the higher tones but yet strong and deepest black in the bass. Try it... You won't achive this using any other codec or player. BUT: Please try it with high-quality phones.
Headphones I have used:
- Sony MDR V700 DJ
- Grado Reference Series RS-1

Anonymous said...

Good lord, what are we complaining about?
To the human ear, this is a perfectly flat curve. A zero point three dB deviation does not constitute a peak, simply because it is inaudible, period.

If you've been hearing a peak, it must have been at least 10 times, i.e. +3dB. Possibly +2dB for a trained audiobuff with a sensitive ear, but definitely not + 0,3dB!

If you trust your ears first, then go for testing, AND YOU COME UP WITH THIS GRAPH to show your point, you should definitely question either your perception at listening time, or your testing.

bangraman said...

I think it's an intersting point. Technically speaking I'd agree. However so many people who claim the Sony has higher sound quality make reference to the 'warmer/fuller sound' of the Sony that the psychoacoustic elements of the change are perhaps worth considering.