HIFIMAN Susvara Headphone Review

[Coverpage]

How is this to be achieved? Each member has to invest in expensive testing equipment and then learn how to operate it? You start.

I think you missed the point. They don't need to measure again. They just have to deduce if Susvara sounds worse than the IEM.

If they think that it isn't true, then they won't believe that the methodology in ASR accurate.

All they have to do is to listen to both Susvara (in-store) and the IEM.

 

[Brubaker]

I am not at all surprised by these results. I don't see many differences between the various Hifiman products. Hifiman's commercial policy is clear... the quality is very similar and the profits for the top models are disproportionate in relation to the quality and results. Moreover, they sell headphones for 6000 dollars that are difficult to drive with very low sensitivity and with a lot of problems related to the resonances of the structure, lack in bass and which require equalization like almost all cheap headphones. I believe that Hifiman reaches the maximum price/performance level with the Sundara and Edition XS models , 300/400€. It is clear that the owners of the Susvara will not agree having purchased it at a high price, as will the reviewers who manage to write pages and pages of words about their "sensations" but without even a measurement.... it makes me laugh.. Well done ASR, keep it up!

 

[caught gesture]

I think you missed the point. They don't need to measure again. They just have to deduce if Susvara sounds worse than the IEM.

If they think that it isn't true, then they won't believe that the methodology in ASR accurate.

All they have to do is to listen to both Susvara (in-store) and the IEM.

You‘re missing the point of ASR if you think that bias trumps measurements.

 

[amirm]

The reason is that the transient response (at least perceived to me) makes it sound more "real." When you listen to jazz or acoustic tracks, the percussive sounds sound more impactful and realistic.

This is just an audiophile myth. Transient response is directly proportional to bandwidth:

Digital audio already puts a hard cap on bandwidth (22 kHz for CD rate) and hence transient response. If you examine the frequency response and you see the transducer covering that range, then there is no issue regarding transient response. Subjective comments about this and that headphone, speaker, IEM being faster or slower is just nonsense, layman speak. It has no place in an informed discussion.

To make it quick, you can either have a light diaphragm, strong magnets, or both.

Nope. Just no.

 

[Robbo99999]

A bit of background. I met the designer at Cambridge AES conference in 2008. And listened to what was one of the most incredible things I had heard. He had a 5 channel system and you would put on and take off the headphone and could not tell the difference whether you were listening to speakers or headphones! The designer by the way was the same person who came up with DTS compression algorithm for movie sound. So we had a lot in common and had a great chat about the industry.

So yes, would be great to test one although it would not fit the style of these reviews. It is like testing room EQ or something like that.

Yes, I can see how it wouldn't fit into the normal style of review, but it would be great to see it tested/reviewed, but might need some thought on how to approach it. Nevertheless, a very interesting product to find out about - Smyth Realiser.

 

[Coverpage]

You‘re missing the point of ASR if you think that bias trumps measurements.

The point of the measurement is about finding out which headphones or iem is better. This has been clarified. If the user find that the headphone that is measured worst sounds better, there’s two possible deduction. The model isn’t good enough, or they’re the minority.

This is science.

What do you mean by bias trumps measurements, that’s a random jumble of words.

 

[amirm]

So these are the things that should give us the "detail" of the recordings, and should be visible in FR plots if the system is simpler and LTI.

Nope. If non-linearities are sufficiently small we can treat the system as linear. Such is the case here with these transducers when driven at lower volumes. 1% THD isn't going to change the frequency response.

In addition, we are NOT trying to model the whole system using frequency response. For that reason, the measurement need not be invertible as to require it being an LTI. We show the measurements for humans to interpret.

 

[Soria Moria]

I find it unlikely because of this.

The person I replied to mentioned the price of the IEM as something that matters. Whether an IEM is $20 or $2 000 it still bypasses the pinna.

 

[fivepast8]

This is just an audiophile myth. Transient response is directly proportional to bandwidth:

View attachment 337171

Digital audio already puts a hard cap on bandwidth (22 kHz for CD rate) and hence transient response. If you examine the frequency response and you see the transducer covering that range, then there is no issue regarding transient response. Subjective comments about this and that headphone, speaker, IEM being faster or slower is just nonsense, layman speak. It has no place in an informed discussion.

Nope. Just no.

Hello Amir, thank you for your wonderful work.
I admire your patience and willingness to engage with your audience and to "try" and teach them basics of electro-magnetic system design. Such patience eludes me.
Have a wonderful holiday. It's bed time for you, unless you are on vacation somewhere!
Take Care

 

[Coverpage]

Nope. If non-linearities are sufficiently small we can treat the system as linear. Such is the case here with these transducers when driven at lower volumes. 1% THD isn't going to change the frequency response.

In addition, we are NOT trying to model the whole system using frequency response. For that reason, the measurement need not be invertible as to require it being an LTI. We show the measurements for humans to interpret.

I’m on my mobile so I’ll reply to this less verbose one first.

Lower THD does not mean system will then be somehow close to LTI (which as a statement make no sense). You derive THD from a sine sweep which exercise is assuming the system is LTI for you to derive the FR

 

[Sokel]

Bittersweet after Christmas story alert.

I own some 40 yo orthodynamics,I was 14 when I got them, audio-technica ATH-M2 which I still use even at this very moment.
If you ask me about how it sounds to me I will tell you they have the best high on the universe (or multiverse)

The truth?What I hear in there is my youth,countless plays,exercises and practice,millions of goosebumps for years plus the fact that I had to save my school money for a month or two to get them so their price was an equivalent of at least north 4 figure-south 5 figure to my own income percentage.

How can I not be emotional about them?Yep,they are not going anywhere below 50Hz or so and when I swap them for the LCD-x is like entering another world but ask me which ones I would burn if I was threatened to do so .

There's nothing like emotion but true emotion can admit it's own weaknesses and often enjoy them and laugh with them.

 

[Turambar]

Just a small abstract of my point:

If we base our headphone purchase solely on these measurements, the Susvara loses out to IEMs that may be 10s of dollars.

Assuming the goal of your measurement is to deduce how well a headphone sounds, then the above statement must be true.

We can verify this through a survey. But anyone who has heard the IEM and a Susvara knows that, putting price aside, the Susvara is perceived to be better sounding.


So either the goal is not clear, or the methodology is not good enough.

You keep repeating this message time and time again, assuming something must be wrong because Susvara (6k) loses out to IEMs that costs less than 100 dollars.

But look at it the other way around. Maybe the problem is that the Susvara, and other headphones / pieces of audio equipment, are grossly overpriced and we should not be surprised that the measured performance of some of these headphones are rather poor, and can be bettered by items costing much, much less.

I never listened to the Susvara, so no personal opinion to share, I can't be part of your projected survey. But reading Amir's review, he is doing his usual stuff, with the same set of metrics, he listens to the headphones and then he shares his conclusion. Like with many other headphones. Nothing more, nothing less.

 

[caught gesture]

What do you mean by bias trumps measurements, that’s a random jumble of words.

Bias, the subjective listening experience in the store and conclusions drawn. Trumps, surpass (something) by saying or doing something better. Measurements, the process of associating numbers with physical quantities and phenomena.

 

[Chagall]

The person I replied to mentioned the price of the IEM as something that matters. Whether an IEM is $20 or $2 000 it still bypasses the pinna.

Sorry, I just have a problem with such a blanket statement ("IEM better than YXZ") because it won't be true for everyone - and just wanted to point out one instance of why that can happen regardless of bias, price, or feelings.

Yep, the price does not matter but is the main reason that Susvara is crucified. As I said before, I don't own one and have no feelings either way.

 

[usern]

Transient response is directly proportional to bandwidth:

View attachment 337171

Yes. You may have a headphone with wide bandwidth and source (DAC + AMP) with wide bandwidth (above 20 kHz) and scope will show nice steep, vertical, "fast" graph, but your ears filter it into "slower", less steep slope anyway.

 

[threni]

You keep repeating this message time and time again, assuming something must be wrong because Susvara (6k) loses out to IEMs that costs less than 100 dollars.

But look at it the other way around. Maybe the problem is that the Susvara, and other headphones / pieces of audio equipment, are grossly overpriced and we should not be surprised that the measured performance of some of these headphones are rather poor, and can be bettered by items costing much, much less.

It's an odd argument, isn't it? Does stuff sound better because it's expensive? So buy a cheap pair of headphones and list them on ebay for £20,000,000 for an instant audio improvement! Is everything priced according to its quality? Then don't read reviews/look at measurements - just buy the headphones (presumably this applies to all audio equipment) you can afford - obviously all headphones at a given price sound identical.

 

[usern]

I just listened to this headphone a few days ago and my oh my do I love how it sounds. But hey, placebo is one hell of a drug I guess?

You are missing the point if you think you HAVE to like Harman target and dislike deviations. You may just like the tuning (+weird phase effects it has) and that's it.

 

[IAtaman]

You think these two are in the ballpark??? Response at 20 Hz is lacking by about 7 dB. Where the heck do you get "2 dB" from?

The research offers a way to predict a headphone's preference based on its tonal balance that involves calculating standard deviation and tilt. Oratory1990 does these calculations. According to his calculations predicted preference scores of these two headphones are in the same ballpark with 81 and 84 points respectively.

Hifiman Susvara.pdf

Shared with Dropbox

www.dropbox.com

Sennheiser HD650.pdf

Shared with Dropbox

www.dropbox.com

Data says measurements between 50 Hz to 10K Hz correlate with preference. 7dB difference at 20Hz is not relevant.

I mentioned HD600 and Stealth in connection with tonal accuracy, not in connection with distortion.

 

[Coverpage]

Bias, the subjective listening experience in the store and conclusions drawn. Trumps, surpass (something) by saying or doing something better. Measurements, the process of associating numbers with physical quantities and phenomena.

It’s the opposite in this case.

We are trying to create a model for which headphones sound better.

What sound better to a population are the observations, the measurements is part of the model. If the model doesn’t agree with the observation then it’s a bad model.

 

[IAtaman]

Because of this:

See any distortion at up to 110 dBSPL above 600 Hz? You don't. Yes, it also has a hard limiting but up to that point its level of distortion is similar to audio electronics. No way can you compare it to Susvara with its clear distortion spikes.

And keep in mind that Stax did NOT get a better review score without EQ:



Postman panther is the same rating. But add some EQ to it and stay below its hard ceiling and the Stax puts a smile on your face.

The point is not to compare the distortion characteristics of Stax to that of Susvara. Height of the distortion is not relevant as long as they are not audible. Are you are evaluating sound quality of headphones or their "engineering excellence". In any case you have recommended much worse performing headphones as I mentioned, and you kindly ignored.

Nevertheless.. the point is Susvara very likely has no audible distortion at 94db SPL, but distorts at higher levels. And so does Stax. When I ask you why you think Stax is a wonderful headphone and highly recommended for low to medium level listening but Susvara is not, are you telling me you are not recommending Susvara because its 94 db SPL distortion is not clean enough? What about all the other headphones you recommended that had similar distortion characteristics, like HE6 or Aeon Noire? Or are you saying your not recommended rating is not based on distortion measurements but on your listening tests?