Tír Na HiFi

sima66 wrote:Thomas, would that CRC or SHA (no idea what they mean) change when you copy the song from one place to another?
I'm assuming NOT, but the sounds degrades with every single copy!
Maybe that will be your answer.

Sima66, what I understand from the discussion here and around the BHE (Bug Head Emporer) rewrite files sound different. If it would be right that sound degrades with every copy the would also something be changed with the bits, if it is not the process of reading and working with the file itself. And yes, just if you just copy a file CRC etc. shouldn't not be changed if nothing changed with the content of the file.

I have teste BHE's rewrite around a year ago and with the whole System in RAM and no SSD attached anymore the rewrite did not change sq in my system. But knowledge has improved ;)

I am just stumpling about the same result in this context as elsewhere mentioned around ripping. While writing here, ripping in RAMOS is finished, going to compare the files now.

Let's see what happens ;)
Thomas

P.S. CRC etc. are checksums build from file allowing to verify that the content is the same, working quite fast and easy.

Thomas is MQNRewrite working for you in RAMOS?

sebna wrote:Thomas is MQNRewrite working for you in RAMOS?

Hi Seb,

No, I did not even have the time to test the latest MQn. I will try to find some time. In general, I would not see many reasons why it shouldn't work in RAM if it is working when normally booted. One could be, that there is not enough memory left in RAM. For testing you should be able to use a file with a size which fits in your existing environment. I would start very small as I have no clue often the file is rewritten ;)

Seb, you have for sure seen the remarks on my experience with BHE rewrite in RAM?

I changed my speakers and realized my Plans of an active concept with KEF LS50 and two Subs. Lot of measurements with Acourate but I am very happy with the results. At the moment I am focussed on Server 2016 TP4 testing with individual optimization and this is running perfectly in RAM. I just stumbled about the similarity of the argumentation between rewriting and ripping in R2 Core Mode that says there is different noise left in the respective files.

Cheers
Thomas

I do not write off anything before I try it especially that I can think about at least two aspects of why it can work even tough of course we do not change the logical content of file hence checksums will check-up perfectly each time (however Gordon's rewrite is modifying files by removing last few bytes as he says, so in this case checksum would not be the same).

But it is worth to remember that CRC check checks only logical content of data without telling us anything about physical properties of how this mathematical concept, we call data, is stored and as such, how easy or difficult, depending on quality of "saved state" the reading of data will be and how much electric noise will it cost to retrieve it from physical storage potentially causing us loss of sound quality. Logical part, the data, is purely theoretical concept of 0 and 1 which are absolute but those 0 and 1 are stored in physical word using physical and as such analogue methods and those methods are far from perfect, leaving room for rewrite.

To give easiest example of what I am talking about is removing data from HDD. You can remove the file, write same space over with another file but still be able to recover the original file... in digital world that should not be possible but reality where the digital concept is still stored in physical analogue world it is just reality. If you want to get rid of data permanently you have to delete it with at least of 3 passes over sectors in question, but really 7 should give you a peace of mind :)

I also tested BHE rewrite and I though it has been making a difference but I did not like the difference in most cases :)

CRC and the other checksum tools have been first ideas. I thought checking the whole binary content would be to complex at the beginning. In between I recognized a tool making differences visible per single byte als well as an overview. Interesting results but allow me some time to review and verify.

If one would be able to share, maybe via pn, an original file and after it is rewritten with Mqnrewrite with me, I could use these files as an example here. I compare at the moment the results of the different ripping scenarios.

Take care
Thomas

there were 2 incorrect settings, so have uploaded a new set of files.

all the files required for MQn are in mqninstall.zip

uploaded control v97, loader v82 and play v224. Also rewrite v14.

control v99, loader v84 and rewrite v15 use sse4 instructions rather than avx, think there's better detail with these.

sbgk wrote:uploaded control v97, loader v82 and play v224. Also rewrite v14.
control v99, loader v84 and rewrite v15 use sse4 instructions rather than avx, think there's better detail with these.

My heart skipped a beat reading "sse4" and I immediately downloaded the archive. It's too late to try anything now, but would love to do so tomorrow. The file names all still have "haswell" in them though. Does this mean that I still won't be able to use play v224 on my AMD CPU?

I've been cherishing the latest sse4 versions from mid December. Despite the fact that they're 16/44 and sse4, they do sound better than my previous reference, the 3.39 avx hires. I can't wait to try the new stuff. I've even been thinking of finally getting on board with a Haswell upgrade, but I just can't afford that at the moment.

Still enjoying the fruit of your labour of love each day. It's very much appreciated.

satshanti wrote:

sbgk wrote:uploaded control v97, loader v82 and play v224. Also rewrite v14.
control v99, loader v84 and rewrite v15 use sse4 instructions rather than avx, think there's better detail with these.

My heart skipped a beat reading "sse4" and I immediately downloaded the archive. It's too late to try anything now, but would love to do so tomorrow. The file names all still have "haswell" in them though. Does this mean that I still won't be able to use play v224 on my AMD CPU?

I've been cherishing the latest sse4 versions from mid December. Despite the fact that they're 16/44 and sse4, they do sound better than my previous reference, the 3.39 avx hires. I can't wait to try the new stuff. I've even been thinking of finally getting on board with a Haswell upgrade, but I just can't afford that at the moment.

Still enjoying the fruit of your labour of love each day. It's very much appreciated.

might work, don't know, they are still compiled with AVX settings, but my code is no longer using AVX2 instructions

uploaded rewrite v17 and control v101 after having a rethink of how the wav file is read/written. Sounds great.

sbgk wrote:uploaded rewrite v17 and control v101 after having a rethink of how the wav file is read/written. Sounds great.

On some wav-files Rewrite v17 doesn't finish. I have to kill the proces and than the batch file will continu with the remainder of the selection. The file it hangs on is written into out, but is 0 kB in size. Repeated tries on the same file all hang.

Trying v16 on the same problematic files runs without an issue.

Most files are ok, but some are a problem for v17

Cheers