Can you hear bit rate?

@erik_squires 

DSD is a very weird duck ... If you had DSD and wanted to EQ it, you'd have to either convert to PCM, or convert to analog, EQ, and then return to DSD, not really a very good process overall for purity

DSD is a brilliant format for playback, but a difficult format for editing, because each bit is relative to the running total!  PCM assigns an absolute level to each sample.

For this reason, Philips and Merging Technologies created the DXD format to facilitate editing and merging of DSD streams.  From AI:

"DXD, or Digital eXtreme Definition, is a digital audio format that records in PCM at a 24-bit depth and a 352.8 kHz sampling rate. It is used primarily for post-processing high-resolution Direct Stream Digital (DSD) recordings because it provides a convenient, less destructive method for editing while maintaining high quality"

The Norwegian company 2l.no takes this much further and sometimes uses floating point number representations.  Their archives are in DSD1024 (from memory) from which all other digital formats they support can be recreated.  For immersive sound they use 9.1 channels.

I note that from the earliest days of CDs, Philips players used four-times oversampling so they could use much gentler low-pass filters.  Their early machines did not bother with the two least-significant bits, effectively giving 14-bit resolution.  They sounded better than contemporary CD players.

I doubt that any 24-bit audio files really make use of all 24-bits.  And I for one can definitely tell the difference between native DSD from a SACD and the same down-sampled to CD quality!

By way of analogy, the difference in the appearance of an 8 bit digital photograph compered to its 16 bit version is pretty much impossible to discern. But if you're going to do any post-processing in Photoshop (playing with contrast, levels curves etc.) the extra bit-depth is very important. The same is true for recording sound that will be digitally manipulated in post, which now is pretty much everything, which is why even cheap consumer home studio gear records never records at anything less than 24/96. I agree that this matters far more on the upstream production side than on consumer format end. My Meridian CD player doesn't make me thirst for higher bit depth streaming options- but, that said, I'm just happier with my vinyl rig, which lays bare all the physics of sound reproduction instead of shrouding it in a black box full of ones and zeros.

A number of years ago, I tried a not particularly rigorous experiment with various bit depths and sample rates. It involved taking an excellent sounding 24/192 FLAC file and converting it down to 16/44 and a couple of points in between. On listening back with a highly resolving replay system, the CD quality file was just as enjoyable an experience as the original high res version. It appeared that not much was lost in the conversion. I came to the conclusion that the mastering of the album and quality of the streamer and DAC, etc. was far more important than than the resolution of the file.

@erik_squires wrote:

Can you hear bit rate?

Reliably and with my current source (i.e.: streaming device), no. Usually there's more than bit (and sample) rate to determine the sonic outcome of a file, so the basis for comparison is hardly one to one in terms of isolating the bit-rate itself. 

And to be honest, I don't care (any more): if it sounds great, it sounds great, and bit or sample rate in itself isn't a reliable marker in my experience. 

Next year I'll be upgrading to a new streamer, and whether that will be able to make a difference with regard to discerning the audible impact of changing bit or sample rate, remains to be seen/heard - assuming I would be able to make a reliable comparison in the first place. 

Just to be clear, I used 'bit rate' specifically and not 'bit depth' for shorthand, which I guess wasn't fully clear. 

Bit depth = a power of 2 number, usually from 16 to 24 in audio, describing the bits per sample. Synonymous with sample size. 

Bit rate = bit depth * number of channels * sample rate. 

So, apologies, I meant to ask: Can you hear differences as the sample rate or bit depth changes?