[skamp]

16 bits * 44100 samples per second * 2 channels = 1411200 bits per second

 

[reginalb]

Quote:

 
There is a remarkable resemblance to full-resolution at that bit rate, to be sure.  However, at least for me, longer listening (esp. on a familiar system when the subject is selecting the music and volume) reveals that differences can be heard up to 320kbps.  Double blind methodology matters -- to do it well is seemingly impossible, and lengthy.  When DBT does in fact detect a difference, you have a strong positive result.  However, when it does not reveal a difference, by definition ("null hypothesis" testing), nothing is proven except that the difference is not large enough to be obvious under those exact conditions. 
 
hile there is certainly excessive self-fulfilling or placebo effects in the high-end world, it is not universal.   I have been an AES member and even published a few times, and over the decades I've studied Sean Olive, Floyd Toole, and others, and been friendly with John Curl and John Atkinson.  And spent some time with Mikey, and a dozen manufacturers and journalists.  A major mentor was Richard Heyser.  The ones who know what they are doing do not often fall prey to that.  At least speaking for myself (acoustician and audio consultant for a time, who has studied with James Boyk at Caltech, Isadore Rudick and others at UCLA, and was offered the job of technical director at Stereophile in 1988), I tend to be hesitant to conclude any differences exist unless I can double check myself (like an informal version of the formal replication required in science journals.)  Turns out, BTW, that my wife hears differences even faster than I.  I could tell stories!....  

 
 
  My personal testing with ABX testing in Foobar is that, if I use a song from HD Tracks, convert it to 320k MP3 with dbPoweramp using the LAME codec, I can match the files correctly around 50% of the time (randomly) with most tracks. I think that what a lot of the double blind tests miss is this: the differences between 320k MP3 and an FLAC file are subtle, and some music doesn't have enough definition for it to be detected. But if I listen very closely, there are parts of certain tracks that give it away. I have found that on those tracks where I am confident that I have noticed a distinct difference, I nearly always get it right, it's just that most tracks I don't pick up the difference. Basically: on the tracks where it seems I am just guessing, I know I am guessing, because they sound the same to me, but there are tracks that don't sound the same to me, and I nearly always get those tracks right. If I didn't discriminate, my total results would be a little over 50% correct, in the neighborhood of 55%. Of course, that still might be a statistically significant variance from 50/50, if n is large enough. 
 
  If i wanted to spend the time, I might be able to find a part of more tracks that sound different, but I have done enough testing myself to be confident. And this isn't with super equipment, either. It's just my laptop with some AKG Q701s fed by a Hifiman 101. Any way, that is my take on lossless vs. lossy. In regards to this program, I will echo what many others have said, that what matters far more, is the mastering. And if he is able to push better work in the studio when the files are produced, then this will be a win for us. Give me a 320k MP3 produced correctly any day, and I will be a happy camper.

 

[Stoney]

Quote:

 
16/44.1 WAV or AIFF from CD rip is effective 1411kbps, no redundancy there in the meaning of error correction.

 
AIFF is 1411kbps, as you say.  Apple Lossless is variable bit rate, about 800-900kbps.  By definition, any higher bit rate than a lossless format is "inefficient" for some reason or another.  So it is not a good comparison quantitatively to mp3 which is intended to be ultimately efficient. 
 
But, perhaps I'm recalling poorly when I said that AIFF is what is exactly what is in the CD pit by pit.  I do know that the CD pits include additional bits and spatial arrangement of the data to better cope with scratches, dust, laser wavelength, etc.  So, accepting the math, that implies that the bit stream from the CD itself is higher than 1411bps, some of which is not necessary for the music encoding alone: 
 
 
http://en.wikipedia.org/wiki/Eight-to-fourteen_modulation
 
Under EFM rules, the data to be stored is first broken into 8-bit blocks (bytes). Each 8-bit block is translated into a corresponding 14-bit codeword using a lookup table. ... Thus, in the final analysis, 17 bits of disc space are needed to encode 8 bits of data.  

 

http://www.webopedia.com/TERM/E/Eight_to_Fourteen_Modulation.html

Eight-to-Fourteen Modulation, or EFM as it is abbreviated, is an encoding technique used by CDs and provides a way of countering errors by encoding abyte into 2 bytes. Using EFM the data is broken into 8-bit blocks (bytes). Each 8-bit block is translated into a corresponding 14-bit codeword using a predefined lookup table. The 14-bit codeword are chosen so that binary ones are always separated by a minimum of two and a maximum of ten binary zeroes. 

 

[Stoney]

Quote:

some music doesn't have enough definition for it to be detected. But if I listen very closely, there are parts of certain tracks that give it away.
 
 ...
 
And if he is able to push better work in the studio when the files are produced, then this will be a win for us. Give me a 320k MP3 produced correctly any day, and I will be a happy camper.

 
Agreed, on all points.  
 
I'm now listening to a CD on my SCD-XA5400ES ($1500) through a $600 tube amp, and comparing with the same disc at 320kbps on my iPad 3G with an Arrow 4G amp ($300), same headphones (HD650).  The latter is far more detailed, revealing, and fascinating (Joe Morello, drums, "Going Places"), with immense detail on each percussion strike, skins tuned to tones, suble variations on cymbal strikes....  The equipment differences, and terrific recording/mastering in this case swamp out the subtle if detectable at all differences in bit rate.  I do most listening this way now.  

 

[goodvibes]

I don't doubt your result but you guys keep blaming the wrong thing. You can't make absolute format evaluations on a PC or $600 tube amp. It does mean that for many, HiDef may be a waste but things like this were never meant for everybody.

 

[Stoney]

Please don't overgeneralize the points made recently, which may be a digression from the main point.    No blaming or overgeneralization intended.

 

[goodvibes]

Quote:

 
AIFF is 1411kbps, as you say.  Apple Lossless is variable bit rate, about 800-900kbps.  By definition, any higher bit rate than a lossless format is "inefficient" for some reason or another.  So it is not a good comparison quantitatively to mp3 which is intended to be ultimately efficient. 
 
But, perhaps I'm recalling poorly when I said that AIFF is what is exactly what is in the CD pit by pit.  I do know that the CD pits include additional bits and spatial arrangement of the data to better cope with scratches, dust, laser wavelength, etc.  So, accepting the math, that implies that the bit stream from the CD itself is higher than 1411bps, some of which is not necessary for the music encoding alone: 
 
 
http://en.wikipedia.org/wiki/Eight-to-fourteen_modulation
 
Under EFM rules, the data to be stored is first broken into 8-bit blocks (bytes). Each 8-bit block is translated into a corresponding 14-bit codeword using a lookup table. ... Thus, in the final analysis, 17 bits of disc space are needed to encode 8 bits of data.  

 

http://www.webopedia.com/TERM/E/Eight_to_Fourteen_Modulation.html

Eight-to-Fourteen Modulation, or EFM as it is abbreviated, is an encoding technique used by CDs and provides a way of countering errors by encoding abyte into 2 bytes. Using EFM the data is broken into 8-bit blocks (bytes). Each 8-bit block is translated into a corresponding 14-bit codeword using a predefined lookup table. The 14-bit codeword are chosen so that binary ones are always separated by a minimum of two and a maximum of ten binary zeroes. 

 
 
No difference between aiff and wav other than how the blocks are organized and how the meta data is stored. ALAC and FLAC are more effecient in space utilization due to effectively being zipped versions of a music file but they're less effecient in playback due to needing the additional step of unzipping before constructing a stream.

 

[doublea71]

Take it to another thread already - I want to read about PONO or little toy choo-choo trains....
 

 

[1502]

Good to see that some artists these days actually care about music.

 

[tds101]

Hey, their off topic banter is more action than this thread has seen in awhile,...enjoy it.

 

[Sebhelyesfarku]

Quote:

Originally Posted by Stoney /img/forum/go_quote.gif
 
But, perhaps I'm recalling poorly when I said that AIFF is what is exactly what is in the CD pit by pit.

 
Yes you are. It isn't. Just 16/44.1 PCM data comes off the CD.

 

[Stoney]

Quote:

 
Yes you are. It isn't. Just 16/44.1 PCM data comes off the CD.

 
You must be right, and I'm thinking of the pit encoding ON the disc, which is not what the transport outputs OFF the disc.  Thanks--this is concise: http://makbit.com/articles/cd-overview.pdf. 

 

[doublea71]

Quote:

Hey, their off topic banter is more action than this thread has seen in awhile,...enjoy it.

Umm, no. It isn't enjoyable at all to see that new posts have been made and none of them have anything to do with the PONO.

 

[skamp]

Yes. This thread needs more free PONO.

 

[goodvibes]

I also wish that HiDef debates would stay away but it's paramount to what the PONO is about.