Is Amarra bit-perfect?

[cfmsp]

 

Is Amarra bit-perfect?

 

Below is the text of an email from Jon Reichbach of Sonic Studio which was posted on another forum in response to the question "Is Amarra bit-perfect?"

 

"In almost all Digital Audio Processing systems there is some amount of processing.

 

For example, every 24 bit integer PCM sample is most likely converted to a float and then back to a sample in the audio driver/hardware.

Other types of processing could include Gain, EQ, and Dither. Amarra does not process the data in anyway unless the user specifies it.

 

There are times though when some processing is important. For example, when using the digital gain in Amarra one would want to apply dither as it will improve the sound. Likewise, if the EQ can help with the bass response then it should be used. The fact is that most modern day Digital Audio system perform some manipulation of the audio, even for simple conversion.

 

Regarding bit perfect, the HDCD light will show that the HDCD sync bit is being read correctly as detailed in the HDCD specification This bit is verified every 10 frames (1/3 of a second of audio). It does not guarantee that all the bits are OK, nor does it guarantee that every bit is correct only that the HDCD sync mark is valid. On the other hand, this is a very good indication that the audio is correct but is not an indication of full data integrity. Only with a valid digital NULL test where both signals are played in synch with one out of phase can a full bit test be done.

 

The most important part of this is how it sounds."

 

 

enjoy

 

 

 

 

 

[The Computer Audiophile]

I agree with him 100% on this.

 

[flatmap]

I think he is saying that (as long as the user doesn't specify EQ or digital gain) then 1.) Amarra only processes the signal in that it converts from PCM integer data to floating point and then back to integer. And apparently, 2.) Amarra always does this type of conversion even when no EQ is specified? Do I have that right?

 

[souptin]

I think this is a very good question, for all computer systems not just Amarra.

 

The interview with Johnathan Reichbach is also interesting on this very point (no doubt most people have already listened, but if not, or if you want to listen again, it's still up on the home page)

http://www.computeraudiophile.com/content/CA-Podcast-Episode-2-Jonathan-Reichbach-Interview-Amarra

I got the impression that when asked 'the question' his reply was, not vague exactly, but as if he didn't feel that it was quite the all important requirement for best sound quality. Certainly his software allows for some EQ adjustment without any penalty to quality, which is very nice once you get over the drilled-in audiophile reluctance to use it.

 

I feel that part of the appeal of 'bit perfection' to those of us trying to build ourselves a nice system is that it implies that if you can establish that you have 'bit perfect' at some point in your chain, then anything before that point (usually computer setup) is irrelevant. Also, since we're all human, perhaps the use of the word 'perfect' helps a lot with that impression.

 

At the end of the day, though, it's not the actual bits that we're listening to, it's an analog interpretation. After all I imagine only a few people can look at a musical score and mentally hear how it will sound, and even fewer would look at a steam of data (Matrix-style!) in preference to hearing it.

 

[cfmsp]

 

 

Chris,

 

I don't disagree with anything Jon says, but I'll admit to assuming that the question can be answered either yes or no, and that the absence of an unqualified yes seems to imply a reluctance to admit no. Not that there's anything wrong with a 'no', necessarily.

 

My impression from Jon's response is that he's asking for a "license" to process the bits in a manner that gives the best results. Fair enough, I guess. He'd probably never run into the audiophile dictum of bit-perfect-ness before, after all, as is well known, recording engineers routinely and intentionally process the sound using his software. If it's good enough for award-winning recording engineers, why shouldn't it be good enough for audiophiles?

 

While admitting that it is the final result (i.e. the sound) that matters, the stakes are too high with regard to our collective understanding of how computer audio works to just accept the result without understanding what/how/when processing is being applied that might alter the 'bits'. Wouldn't you agree?

 

I mean, after all, the $64,000 question still remains - all other things being equal, how can two players sound differently if each of them replay the bits exactly as contained in the file?

 

Regarding Jon's comments on the HDCD flag, I don't think it reasonable to accept that files played through Amarra are bit perfect simply because the HDCD light 'goes on' during playback. Jon even says as much. I believe that the HDCD light is a good indicator for discovering when bits are no longer bit perfect "unintentionally", but with the kind of care with the data which Amarra claim (and no doubt, employ), one could easily process the data and then restore the HDCD flag, indeed not to do so would be considered a flaw in the software.

 

just my two cents,

 

clay

 

 

 

 

 

[cfmsp]

 

 

To your question, flatmap, here is more info from Amarra on the basic processing you ask about. This is an excerpt from the Why Amarra pdf available on their website.

 

 

"Once the audio is in memory, it needs to be converted to a format the computer can "understand". Current audio

formats for uncompressed audio use a 16 or 24 bit sample and this must be converted to the IEEE floating point

architecture in use by most computers today. This conversion can introduce noise into the audio signal. Should we

truncate, round, scale or perform some combination to achieve the best sounding result? This conversion occurs on

input from the disc and on output to the hardware interface. Based on our experience, we sometimes find that

textbook math does not mean the best sounding math."

 

One question is - if every software player goes through this conversion (and I have no reason to suspect that they don't), how do they get back to bit-perfect-ness (as often as they do), unless by following the 'textbook' math?

 

enjoy