[Theogenes]

This may seem a bit off-base, and if so, somebody lemme know and I'll quietly slink back to the shadows . I've just finished reading this thread (lot of great information, especially in the first 10 or so pages), and as it seems there are several people here who work in audio reproduction professionally in some capacity, I'm curious if said people frequent sites like gearslutz as well, and whether they feel the information they find in professional forums like GS is generally better than, worse than, or on par with the information found here (meaning Head-Fi specifically). 
 
I realize this is a horrifically overbroad question, but I'm hoping someone could peer through my wording to the actual meaning I'm trying to get at. For example, I have a strong interest in the Lynx Hilo, and have found a lot more usable information on gearslutz-- specifically, the audio files that compare A/D/A converters like the Hilo by running samples through a full ADA conversion 10 times and null testing to determine the impact of the conversions, which I found to be quite eye-opening and very helpful. 
 
Any opinions would be graciously accepted!

 

[bigshot]

I usually have more confidence in individual posters who have proved they know what they're talking about than web forums or brand names.
 
That said, testing generation loss is a great idea. When I've done it myself, I've been amazed how little degradation there is.

 

[Theogenes]

That makes more sense, I 'spose! And yeah, that was kind of the gist of the 'whoa' moment for me with the ADA conversions... I was frankly blown away by it. I had originally strayed away from the more scientific-leaning portion of HF simply because what I heard just did not seem to match up with the FR graphs I saw with a few headphones, which caused me to think that the quantification of audio was far behind its actual reproduction-- and then I heard that, which prompted me to wonder how incredibly far we've gone down the path towards audible perfection. I'm still not certain where I fall, but the more I read and listen, the more I find myself gravitating towards the Sound Sciencers. We'll see what happens in another 10 years, I guess
 
Btw, thank you for your reply!! If you don't mind my asking, is there anybody in particular you hold in high regard for gear reviews? 

 

[bigshot]

I took an AAC 256 file and encoded it to WAV then back to AAC 256 ten times. It still sounded good. So much for worrying about lossy!
 
When it comes to equipment, I usually don't listen to anyone's review. I research it myself. But for audio theories the "regulars" here in this group humble me with the depth of their knowledge.
 
We are lucky to have Ethan Winer post here on occasion. He really knows his stuff!

 

[Joe Bloggs]

The state of the art in headphones is in some way very advanced and in others not. Distortion characteristics of even cheap headphones can be impeccable, these days it is the frequency response that affects listeners' impressions the most. And here's where things are backward: there is no agreement on what constitutes a good frequency response for headphones or even how they should be measured. There are some good emerging standards but if you see an FR chart for a pair of phones and there is no info on how the measurement was made or what compensation curve was used, it's as good as reading tea leaves for fortune telling...

 

[bigshot]

Frequency response is a lot easier to correct than distortion.

 

[Theogenes]

In reading a bit more about the Smyth Realiser, it seems (if all the stratospheric praise can be believed) that essentially anything can be corrected for, if you only have the right corrections-- like an individualized HRTF. Given the increase in customizable products in the headphone world in the last couple of years (V-Moda, COP, etc), it seems reasonable to hope that a more cost-effective (and widely implementable) version of a device like the Realiser would become available at some point and really start to put the differences between different transducer technologies to the test. After all, if a DT770 can really be made to sound like a setup costing 50x more, you'd have to question the need to spend 1.5k on a pair of 'better' headphones. 
 
Although I still want to hear an HD800 one day, regardless .

 

[vwinter]

When you see a CSD graph, is this done with a sine sweep?

I ask because if say music is playing with many parts of the frequency response being used within very close proximity in the time domain, doesn't this affect the amount of decay allowed for a specific frequency?

Which would cause.... the decay shown on the CSD to be realized only when a specific frequency has a chance to sustain?

If someone who knows what they're talking about could help m wrap my head around this or direct me to where this question is appropriate, I'd really appreciate it.

 

[vwinter]

Edit: mobile site induced duplicate post.

 

[Joe Bloggs]

Quote:

When you see a CSD graph, is this done with a sine sweep?

I ask because if say music is playing with many parts of the frequency response being used within very close proximity in the time domain, doesn't this affect the amount of decay allowed for a specific frequency?

Which would cause.... the decay shown on the CSD to be realized only when a specific frequency has a chance to sustain?

If someone who knows what they're talking about could help m wrap my head around this or direct me to where this question is appropriate, I'd really appreciate it.

 
A CSD graph is theoretically an analysis of an impulse response.  An impulse is a click, the input going from zero to full scale in one sample then going back to zero again in the next sample.  This signal contains all frequencies at the same time--to reproduce an impulse perfectly a transducer would have to be able to reproduce all frequencies with the same magnitude response and have no phase shifts.
 
In reality, an impulse is a very quiet signal such that capturing it with a microphone leads to problems with signal to noise ratio.  Thus the impulse response can be calculated from recordings of other kinds of signals, including maximum length sequences and... sine sweeps.  An explanation of how this is done is beyond my pay grade

 

[vwinter]

A CSD graph is theoretically an analysis of an impulse response.  An impulse is a click, the input going from zero to full scale in one sample then going back to zero again in the next sample.  This signal contains all frequencies at the same time--to reproduce an impulse perfectly a transducer would have to be able to reproduce all frequencies with the same magnitude response and have no phase shifts.

In reality, an impulse is a very quiet signal such that capturing it with a microphone leads to problems with signal to noise ratio.  Thus the impulse response can be calculated from recordings of other kinds of signals, including maximum length sequences and... sine sweeps.  An explanation of how this is done is beyond my pay grade

Haha, thanks for the info! Every bit helps towards better understanding. I'll probably need to go over this a few more times and look some stuff up before it registers anyway lol.

 

[stv014]

Quote:

When you see a CSD graph, is this done with a sine sweep?

I ask because if say music is playing with many parts of the frequency response being used within very close proximity in the time domain, doesn't this affect the amount of decay allowed for a specific frequency?

Which would cause.... the decay shown on the CSD to be realized only when a specific frequency has a chance to sustain?

 
A CSD is generated from an impulse response, as already explained above. You can in theory measure the impulse response with anything for which a practically usable inverse filter can be created if you treat the measurement signal itself as an impulse response. The headphone is assumed to be (at least close to) linear time invariant, that is, have low enough non-linear distortion and not change much if measured repeatedly, so that the test signal can be something other than an actual impulse to allow for a better signal to noise ratio, while the measured impulse response will still be about the same as with an impulse.
 
A maximum length sequence (MLS) is a commonly used signal for measuring impulse response. It is generated with a linear feedback shift register (a kind of simple pseudo-random number generator) that creates a binary sequence (there are only two possible levels in the output, so it can be the loudest for a given peak amplitude) that loops every 2^N-1 samples, where N is the length of the shift register in bits. This sequence has a perfect "white" spectrum, that is, all harmonics of Fs / length (but not DC) have the same magnitude; if you convolve this with a reverse MLS, the result is a train of impulses. The advantage compared to recording actual impulses is that an MLS is much "louder" for a given peak amplitude, and it will improve the signal to noise ratio proportionally to the square root of the length. Also, it does not have low frequency content at a high amplitude (like a sine sweep), which reduces the distortion of dynamic transducers.

 

[uchihaitachi]

[size=medium] Is there a way of replicating the 'tube sound' with solid state amps? Say how could one seek to reproduce even-order distortion, by adding overtones that are richer than odd-order distortion? [/size]

 

[autoteleology]

I wouldn't think so.

 

[Theogenes]

Quote:

[size=medium] Is there a way of replicating the 'tube sound' with solid state amps? Say how could one seek to reproduce even-order distortion, by adding overtones that are richer than odd-order distortion? [/size]

 
Your question made me think about references I've heard to Bob Carver doing essentially what you're asking, so I finally took the time to look it up and found the article here. I'm not entirely sure what to make of the article, but it definitely seems to indicate that whatever comprises the "tube sound" can be indistinguishably emulated by a solid state design. 
 
Also, one of the footnotes now unmasks the high-end tube amp of the comparison as a former Conrad-Johnson flagship. Interesting reading for sure!