Seeking Piano Lovers

But how does piano sound now?

Pianos are sounding very much better. They sound at least equal to the sound of the DAC8. At the same time just about every other instrument including human voices sound better. As I’ve said, in really surprised about the bass. One of my favourite pieces of music is Saint Saen’s so called “Organ” Symphony. The big treat is in how low the organs dive. This DAC does those bass pipes justice. Also base heavy music like Massive Attack, ‘Unfinished’ is pretty spectacular.

Of course, this is a perfect time of the year, I can listen to music all day long. Of course, there are limits, my daughter is doing her PhD work and her cousin is doing her undergraduate work so, I’m not able to open up the volume much. When they go for their obligatory afternoon run . . .

Lovin’ It.

Great to hear. Thanks!

With Snowmass I don’t miss any bass detail with DSD Sr vs Benchmark DAC 3. The bass is phenomenal - totally accurate and not bloated.

I suspect it is your hearing adjustment. A more accurate bass can be less resonant and sound less boomy or attention grabbing. However, after a time you start to hear details in the bass that are only apparent with a tight clean bass response. Revel is one of the very very few speakers with an accurate bass response. Most speakers have a hump in the bass which tends toward impressive one note sound. Most tube power amplifiers will tend to make the bass overly warm even with the most accurate designs like ARC (unless you carefully choose the tubes).

I find a tube preamp with SS power amp can sweeten the bass up just enough and give it just that little bit of AC/DC punch without becoming to overbearing. Your room will be a big factor too - some rooms give way too much bass from room modes.

My room heavily favours the thinner lower mid range of the DSD Sr.vs Benchmark DAC 3. DSD Sr really is a superb sounding DAC - very smooth without loss of detail.

1 Like

Okay, I’m not sure how to express how much I’m in love with this DAC. In the past, I’ve devices that got me listening to my music in a new way but never like this. It probably not a very elegant way of expressing the experience but, it’s as if the DAC is sucking every last bit of playability from my recordings.

I’ve, for the first time, experienced DSD in my home and, I’m afraid, there is no turning back. What an incredible experience to hear Bizet - ‘L’Arlesienne and Faure - Masques et Bergamasques (L’Orchestre de La Suisse Romande: Pentatone) in DSD - freakin’ amazing!

1 Like

Wonderful!

Are you now pleased with the piano sound?

Yes, I’m pleased with it. I’m pretty well pleased with the whole presentation. I’m a died in the wool bullcrap smeller. I don’t buy into the next best thing syndrome. For example, I’ve held back on MQA for some time - waiting until I’ve got a sense that it represents a real improvement. I felt that way with DSD too. Now that I’ve heard what DSD can do, through this DAC I think that it’s very impressive indeed - I hear the difference as music that is more convincing, more present, more exciting and far more euphonic.

Though it’s only the first unfold, through Tidal, I hear some of what MQA promises but (a bit the same with DSD) I’ve had to think about how much of the improvement is due to the DAC and how much to MQA. Comparing the MQA first unfold through my Audio Research DAC8 to the first unfold through the Directstream, I’d say, with the highest degree of confidence that a majority of the great increase in sound quality is due to the Directstream.

Anyway, this is a long answer to your question. Yes, piano music is far more convincing than I’ve ever heard it, in my house. The sound quality is hugely satisfying no matter the file type. I can hear felt-bound hammers on strings.

Great to hear!

If you use the Bridge II you can get a complete unfolding of MQA.

That said, MQA just adds distortion which you will easily hear on piano. Transients are broadened and lower in frequency due to apodizing. And the phase distortion from minimum phase filters wrecks the imaging - leaving a more spacious diffuse clouded image or a hole between the speakers.

Of course many are predisposed to like the distortion as MQA marketing say that it is an improvement.

You are right to be cautious about the next best thing since sliced bread.

Did you listen to Dave Grusin “Homage to Duke”? Be prepared to pick your jaw up off the floor if you do.

As you stated and as nearly everyone concurs, the PS Audio DSD Sr is a superb DAC that deserves its Stereophile Class A+ rating in every way!

The claimed benefit of a minimum-phase apodizing reconstruction filter is to avoid the pre-ringing of a linear phase digital filter. As I understand MQA, it also seeks to eliminate the pre-ringing encoded by the ADC’s linear-phase anti-alias filter.

I also do not like the sound of MQA, but do not understand how “[t]ransients are . . . lower in frequency” as a result of the application of a minimum phase filter. Can you explain?

The claimed benefit of eliminating audible pre-ringing is a lie because pre-ringing of an anti-alias filter isn’t audible provided it is designed correctly. The ringing occurs at 20KHz and above. There is no benefit to fixing what you can’t hear.

Pre-ringing is indeed a problem for filters that operate WITHIN the audible range - so this is of concern for sharp EQ filters and high pass filters operating at AUDIBLE frequencies. Mix and mastering engineers need to worry about pre-ringing. It is of ZERO concern for digital audio playback.

So proper anti-alias filtering pre-ringing causes no audible distortion in playback.

Unfortunately, eliminating pre-ringing by delaying high frequencies with a minimum phase filter actually adds audible phase distortion. It alters the musical waveform by changing the timing of the arrival of various frequencies. High frequencies get delayed a lot compared to low frequencies. This is really destructive to fidelity of the original audio signal.

Apodizing filters by definition are gentle filters that “remove the foot” - this is done by weighting coefficients or “windowing” as it is commonly called. The effect of apodizing is to reduce the sharpness of transients (reduce peaks and broaden the width of peaks). It is a form of audio compression like CD Loudness wars. Music that has been apodized will have a higher average loudness because peaks are less sharp relative to average energy. You can see this on this plot from MQA, the blue peak is wider than the original audio red peak. They neglect to mention that the larger red skirt of the industry reference original linear phase audio is completely inaudible as it occurs at such high frequencies. The way the peak transient energy has been altered by MQA is highly audible.

image

Yes, I am familiar with this. I am however focusing on your statement “[t]ransients are . . . lower in frequency.”

I assume what you actually mean is that the character of transients is changed due to phase distortion, more so as frequency increases. That is, no frequency is lowered, only the timing.

I’m not trying to start an argument, but … there are some folks who would disagree with this statement. Lots of people heard (or thought they heard) a change using a super tweeter which operates beyond human hearing. Personally, I don’t think I can hear much above 10k, but just the same there may be something to eliminating the ringing that doesn’t make sense in terms of our hearing.

Using your comment as a springboard, this reminds me of the discussion on pro audio boards of whether there is any benefit to recording in higher than 44.1/16 resolution when the target format is Redbook.

Similarly, why offer playback higher than 44.1 when we cannot hear the higher frequencies captured by a quicker rate? Yet many hear the improvements wrought by high-resolution audio.

Perhaps more intriguing, 44.1/24 sounds better than 44.1/16, and the improvement can be greater than increasing the sample rate.

Apodizing softens the filter shape. Instead of a brick wall with a foot you have a gentle transition. It reduces ringing effects at the expense of spectral resolution. A broader smooth peak naturally contains lower frequencies than a sharp peak.

If apodizing is done at 96 KHz, I doubt a smooth vs sharp filter does much to alter the audible sound at 20 KHz and below. However, redbook CD files represents an awful lot of digital music and around 22.05 KHz the difference between an apodizing filter and a sharp brick wall non-apodizing filter is likely to become audible due to the smooth filter affecting audible frequencies.

Agreed. If you believe we can hear things above 20 KHz then ringing might be a problem.

Bear in mind that audio signals should be properly filtered at A to D stage so as not to contain lots of energy above 20KHz - so that takes care of all ringing except errors like an individual sample of digital audio at high level (impulse response).

You can make an impulse response to test a D to A and observe the characteristic impulse response with ringing - Stereophile do this on every DAC review - however it is a mistake to assume that this type test signal is ever supposed to happen with recorded music. Recorded music should not contain impulse responses. Properly recorded music should be free of powerful ultra HF energy that would make a DAC brick wall filter “ring”.

Again, the character of transients are changed due to phase distortion, but no frequency is lowered, only delayed.

No, this is incorrect. In graphs of frequency v. time, if the height of the peaks are the same they possess the same highest frequency. A broad peak simply means the highest frequency existed over a greater period of time.

A loss of resolution is fundamental to apodization. This is a fundamental feature of apodization in optics or electrical signals or signal spectral analysis using FFT

Here is a mathematical description of the effects of an apodizing filter or function (as a filter is called mathematically)

http://mathworld.wolfram.com/ApodizationFunction.html

In the reference “Uniform” is a non-apodizing function and of course it has the sharpest resolution. All other apodizing functions show reduced high frequency content resulting in a signal that has a higher proportion of lower frequencies (lower amplitude and broader FWHM).

We continue to fail to communicate.

You claimed transients are lowered in frequency by MQA. I have maintained no frequency is lowered, rather frequencies are time delayed.

Now you are indicating high frequencies are removed. Even if true, this does not lower any given frequency of the input, it merely filers some higher frequencies out.

As a separate matter, I am not convinced that “loss of resolution” equals “loss of high frequency content,” (although it may). Apodization does decrease resolution as a result of longer and higher level post-ringing. But this loss of resolution is not loss of high frequency content, rather it is temporal blurring.

Agreed. It is a complex subject. I don’t think I have been clear enough. My apologies for that.