Thread: Has anyone here heard SACD?

Originally Posted by millionrainbows

The most important consideration is how the files are played back.

Hilarious.

You are not going to get a lot of artists in DSD, it's usually reserved for what I'd call "legacy" artists, your Norah Jones, your Steely Dans, your, you know, Pink Floyds ... that crew.

All of which were recorded, in analog, twenty to forty years ago!

Originally Posted by bfBrian

Even though we only have a sample every 22.7 microseconds

Having done considerable digital editing, I'd like to emphasize just how SHORT 22.7 millionths of a second is. When playing back a digital file, several thousand samples have to go by before it's even perceived as sound. When editing LP tics and pops, the real sharp ones are maybe 10-12 samples. Handclaps and drum strikes are hundreds of samples. When zooming in to the individual sample level, the screen won't go any farther than about 24 samples, because going beyond that resolution is unnecessary.

Originally Posted by Rmathuln

I have owned other players capable of reading the SACD layer, usually higher end BluRay players. On those units I can hardly hear any difference between the CD layer and the SACD layer. But the Denon is a totally different story.

Give me the title of an SACD where you can hear the difference. I have acquired a half dozen SACDs over the years, and on my Onkyo C-S5VL SACD player I swear I can't hear any difference.

Originally Posted by Rmathuln

I play SACD on a Denon DVD 5910ci, a beast of a unit the was built for discriminating ear, whether it be DVDs or SACD. It has 5 discrete Burr Brown DAC chips, massive isolated power supplies, and other uncommon high-end design features.

I no longer use it for video, but it still plays SACD extremely well.

I have owned other players capable of reading the SACD layer, usually higher end BluRay players. On those units I can hardly hear any difference between the CD layer and the SACD layer. But the Denon is a totally different story.

SACD licensing does not allow a player to output the SACD (DSD) digital signal. The player has to convert it to analogue. That precludes using an external DAC. Getting the high quality DSD audio available on SACD media is only likely to happen with more expensive units. Most multi format units will not likely cut the mustard.

The licensing restrictions, which Sony implemented with goal to prevent digital sharing and piracy, are largely responsible in my opinion for SACD never becoming a mainstream media format.

That hasn't been the case for some time. About 10 years ago I started ripping the DSD layers of my SACDs using a hacked Sony PS-3. Much was written about this. After my PS-3 died, I switched to my Oppo 105, which can do the same thing. But more significantly, you can now purchase and download DSD files from a number of e-vendors. In either case you can take the DSD file and convert it using an external DAC.

I believe I can hear a slight improvement when I compare the redbook and DSD layers of an SACD, but most of the SACDs I have bought in recent years have also been remastered, and that makes a greater difference.

Originally Posted by NoCoPilot

Give me the title of an SACD where you can hear the difference. I have acquired a half dozen SACDs over the years, and on my Onkyo C-S5VL SACD player I swear I can't hear any difference.

Just recently I pulled out the Litton Prokofiev Scythian Suite. Put it CD player first, which feed optical digital output to Cambridge Audio DAC Magic Plus.

Switched to the Denon and found smoother top end and better defined sound stage in SACD stereo playback.



I hear great differences with just about every BIS SACD.
Older recordings benefit too to my hear, like the RCA Living Stereo SACDs, though not as dramatically.

I have also made a similar comparison with the sound from the Cambridge DAC and the Denon using redbook CDs. Found no discernable difference. The improvement comes from the Denon reproduction of the hi-res SACD layer.

I have a few Japanese made SACDs not available internationally. Those sound unbelievable when played in hi-res SACD with the Denon. Favorites include Kubelik's BSO Ma Vlast, the Szell Sibelius Sym. #2 from Tokyo, and the Blomstedt Dresden Beethoven Sym. set.

But those cost a fortune! Three current offerings I would love that way are the Martinon Ravel set, Kempe's R. Strauss, and Schmidt-Isserstedt's Beethoven 9 Symphonies. But all told I would have to lay out almost $600 for 18 SACDs. Not worth it.

$58.84 +$3.99 shipping at Amazon. I'm not that rich.

Originally Posted by jegreenwood

but most of the SACDs I have bought in recent years have also been remastered, and that makes a greater difference.

I have a number of SACDs and DVD audios, too. In my experience, the format makes very little difference. I'd much prefer to stream some well recorded music from youtube, rather than listen to a substandard recording/mix from any high quality format. Also, the mutli-channel format doesn't do much for me. I typically find the stereo reproduction to be more authentic sounding than the surround mix, even though I've set up my receiver for time alignment of all the channels, etc. Surround effects can be fun for a while, but for critical listening I still prefer stereo.

Originally Posted by NoCoPilot

Not only that, but with a properly-matched pair of ADCs and DACs (analog-to-digital converter to get the music into the digital format, and digital-to-analog converter to turn it back into sound waves), there is one -- and only one -- waveform that matches both converters. In other words, you'll get out of the DAC exactly what you put into the ADC. If there's a 7μs intra-aural delay in the original signal, it'll be there in the output. If it wasn't there, you can't add it.

If there is a 7μs event that comes and goes within 7 μs, it's quite correct that sampling every 22.7μs won't capture it. It would "fall between the samples."

But the Nyquist-Shannon sampling theorem states that for digital recording the highest captured audio frequency will be half the sampling frequency, so a 44.1kHz sample rate captures frequencies up to 22kHz.

And that 44.1kHz sample rate translates to the 22.7μs sampling interval.

Now, a 7μs event would be roughly 1/3 of the 22.7μs sampling interval, meaning its frequency is around THREE TIMES the Nyquist frequency of 22kHz, or up around 66kHz. I don't know about you, but my hearing doesn't go up that far.

Dogs' hearing goes up to around 45kHz. Cats to around 64kHz. Bats can go up to 200kHz ... maybe bats can hear a 7μs event.

Originally Posted by bfBrian

The sampling interval doesn't have the implications that you associate with it. Consider a continuous waveform like that of the voltage changes associated with the transducer element in a microphone. If we sample that waveform discretely every 22.7 microseconds starting at time 0, then we get a set of samples, let's think of them as S_0. Now let's sample that same waveform again, at the same interval of 22.7 microseconds, but let's start at a time of 5 microseconds instead of time 0. Let's think of those samples as S_5. Even though we only have a sample every 22.7 microseconds, the samples in set S_5 and set S_0 are different. S_5 captures the 5 microsecond delay, and it will be present at playback for us to perceive.

Originally Posted by NoCoPilot

If there is a 7μs event that comes and goes within 7 μs, it's quite correct that sampling every 22.7μs won't capture it. It would "fall between the samples."

Originally Posted by Baron Scarpia

The assumption that a digital audio system is not capable of reproducing a delay shorter than the sample time is incorrect. A delay of one sample period would imply the exact same waveform is recorded and appears one sample later. If the waveform is delayed by a fraction of a sample period the waveform would be sampled at different instants along its time course and different values would be obtained, producing a delayed waveform.

We're all saying the same thing. You guys just said it much better than me.

Originally Posted by Joe B

I'm not an audio or electrical engineer, but logic tells me that 32/384 must have a lot of information in it to be able to generate a 128DSD recording which sounds superior to a standard red book. I would love to be able to directly compare 128DSD (native recording) to 32/384 (native recording).

"Superior" in what way?