There is so much more going on with these different formats than dynamic range. As has been pointed out, the mastering of the album will determine what amount of dynamic range it will have more than will the format. There’s clearly enough dynamic range in CDs, just not the clarity/transparency that is available in other formats. So, I don’t get the obsession with measuring or ensuring “I can make use of it all.” But, this is just me. I’m focused on the resulting end sound quality. If I can get that, I’m not too interested in the nits of it.
On the other hand, I’m currently chasing a new shiny object in the PCM world. That is 32-bit wav files at either 352.8k or 384k. My listening comparisons so far suggest these sound noticeably better than 24-bit DXD (on a same recording to same recording comparison). What I think I’m hearing is a clarity/transparency that approaches that of DSD256. But this exploration is just all about listening comparisons. I don’t really care a whit about the technology except to understand what may be going on given what I’m hearing. More to come sometime down the road if I come to any final conclusions regarding this new shiny object.
That’s exactly my point and opinion. The rest is the essential thing, but the so often proclaimed (not by you) higher dynamic range not.
This was a recent question of me to Paul in Paul’s Post’s (unfortunately without answer), when he stated the opposite (hardly a difference). I asked how he could compare if the DS DAC just plays back 24bit files (as far as I understand).
Yes, appears to. The working theory, as provided by my DAC’s designer is that:
The 24 bit file has been truncated more severely than the 32 bit file and then there is the question if they were dithered and how severely and with what kind of algorithm. At any rate whatever has been done to the 32 bit file has been less severe than on the 24 bit file even if different algorithms have been used. Hence you hear a difference.
One wonderful thing about 24 bit and higher formats is the tremendous headroom they provide when recording.
Recording in 96/24, for example, allows me to set levels so that peaks are at -12dB for safety headroom while still capturing the full dynamic range of the orchestra I am recording. We are spoiled.
Yes I understand this from the recording perspective.I finally get, that the dynamic range of the music above noise floor is generally quite limited, but the headroom may be needed for the high spl‘s anyway.
Although a hall is a generally different kind of fish…do you know which max spl is usually measured at an audience position in a symphony concert?
By a quick search I just found this, which noted 50dB dynamic range in a Mozart symphony, but at least 96dB max spl, so a fat choir or another very loud piece could probably lead to 110 or a few more dB max spl (still with a dynamic range of only 60+ dB then). It was taken 14ft from the 1st violin, so quite 1st to 3rd row, maybe lower spl‘s further back.
What are you (or a commercial mastering engineer) doing with a recording which is using the headroom, to prepare it for release? Compress it to make it practically usable or publish the full dynamic range anyway?
Which DAC? How do you send the 32 bit audio data to it? And do you know whether your files are 32-bit integer or 32-bit float? There’s a fascinating new development in audio recording/production using 32-bit float formats but AFAIK that has to be converted to an integer format with some chosen 0dBfs reference point.
These links will tell you more than I know, and the Sound Liaison link includes a sample file. As I said, I’m just chasing this new shiny object at this point.
PCM: all sample rates up to 384kHz and word lengths up to 24 bits
As is completely normal, something has to fix the floating-point sample space to a specific reference level and transcribe the sample values to an integer format for D-to-A decoding. It’s likely JRiver doing that, though sending 32-bit integers. The DAC is then truncating to 24 bits. DACs usually don’t have the capability to dither at that point, but perhaps the MPD is sophisticated enough to take care of that.
Bottom line is, I’m doubtful that you’re hearing anything better than you could get from downloading the 24-bit versions directly from the website. But if you are, it’s because you prefer the 32-to-24 bit conversion that either JRiver or the DAC is doing over the one that the label offers.
If the peaks are at -12 dB full scale, one increases the overall level by 12dB before release. It is always best to provide the end user with the best S/N you can.
Classical recordings typically use little or no compression.
Given you could record a timpani closer up with 140dB impulse…where would be your limit and why (or not)?
Another interesting question would be:
In close mic‘ing, how is dynamic contrast designed (as it can’t be natural when not using few far away mics only)?
Interesting. Beyond my ken, but I’m trying to learn a bit. I asked the designer of my DAC about the 32-bit sound difference I’m hearing and he replied:
Most likely the difference between 24 and 32 bit files is not caused by the MPD-8, but rather in the way these files have been created. The 24 bit file has been truncated more severely than the 32 bit file and then there is the question if they were dithered and how severely and with what kind of algorithm. At any rate whatever has been done to the 32 bit file has been less severe than on the 24 bit file even if different algorithms have been used. Hence you hear a difference.
I’ll have to follow up with a query about what the DAC does with the incoming 32-bit data if the manual says “word lengths up to 24 bits”. I hadn’t gone back to the manual to see this, so thanks!
Sure. And staying limited to the dynamic range topic for the moment, what we hear is not the dynamic contrast at a normal listening position, but the one at the mic positions combined into a new “whole”. It would be interesting to know how much the two have to do with each other. The less and further away mics, the more I guess.
In multimic’s recordings there is typically a separate mic on the tymps. This is to capture the initial attack. This is a sound the listener expects. While mics have a maximum SPL as to how loud they record without distorting, there plenty which can handle 140dB.
Dynamic range diminishes with distance Mixing in a bit of a close mic’d instrument into the feed from the main stereo pair can add a sense of realism.
Yes, that was why I wondered how much the dynamics on recordings have to do with the real dynamics perceived in the audience. My impression from my classical live concert experiences is, little and the live concert is partly much less impressive soundwise than a recording (with blatant counterexamples).
I love it all as it is, but I also love to have the discussion focused on the real topic. And e.g. regarding dynamics, what we demand in our recording/playback scenario is quite away from the live experience (except for the one standing besides the timpani or maybe the conductor and musicians). I say it’s “better”, but therefore not realistic.