Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Poor Performance Of Md Compression (Read 14773 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Poor Performance Of Md Compression

I just downloaded the MD castanets samples at http://www.pcabx.com/product/mds-jb920/index.htm, and realized that they show lots of pre-echo (16/16 ABX very easily).

I guess ATRAC compression is not on par even with MP3 compression? Maybe this is an old ATRAC version?

edit: I put wrongly the PCAVTech link, now I've posted the direct PCABX link.

Poor Performance Of Md Compression

Reply #1 –
That's about it, ATRAC just doesn't compare with MP3.

Even Sony acknowledges that.

I think you'll like the sleek new MD Net Walkman or whatever they call it.

Loads MD data discs and plays at least MP3 and WMA from what I recall.

Poor Performance Of Md Compression

Reply #2 –
Quote
That's about it, ATRAC just doesn't compare with MP3.

Even Sony acknowledges that.

I think you'll like the sleek new MD Net Walkman or whatever they call it.

Loads MD data discs and plays at least MP3 and WMA from what I recall.

As far I know NetMD does nothing more than convert the MP3/WMA/WAV into ATRAC3 LP2/LP4 format (by software) and then transfers the data onto the MD... So still it uses the ATRAC format.

Poor Performance Of Md Compression

Reply #3 –
That is true the NetMD's transcode the MP3 and WMA data into their own ATRAC format.

Hmm, wonder how hard will it be for Sony to create a file system for the MD (they already have a file system, they use MD as data storage mediums in Japan ). That way we can store files as they are on the MD, instead of doing a lossy transcode.

Laters
AgentMil
-=MusePack... Living Audio Compression=-

Honda - The Power of Dreams

Poor Performance Of Md Compression

Reply #4 –
The castanet clips sounded pretty close to the original clip.. It is OK for me..
I wonder if you are using ATRAC3 format? Because ATRAC3 & ATRAC2 does not switch to short-block... as compared to ATRAC1. I think it is OK because most music clips would not have transients as severe as castanets. Most attacks are rather "mild".. 

This sample sounded very much like ATRAC2 quality... because I have heard of ATRAC2 encoded castanets clip.. however I cannot be sure.. Have you try the Open MG JukeBox software?

wkw

Poor Performance Of Md Compression

Reply #5 –
About the ATRAC3: I know someone who has a JVC MD Walkman that seems to have ATRAC4.5. Is this possible? Sony still uses ATRAC3, isn't that right?

Poor Performance Of Md Compression

Reply #6 –
A moment...
There is no ATRAC2 codec.

ATRAC (or ATRAC1) have many revision : v1, v2, v3, v4 (and 4.5 ? - not sure), type R and finally TypeS.
ATRAC3 is better known as LP : LP2 (132 kb/s in pure stero CBR) & LP4 (66 kb/s in joint-stereo CBR too).

Poor Performance Of Md Compression

Reply #7 –
I've done some research at Google, and seems that the MDS JB920 minidisc model uses ATRAC 4.5, which, talking about MD machines, seems to be inferior only to ATRAC Type-R, which seems to offer some advantages at bit allocation.

Poor Performance Of Md Compression

Reply #8 –
Quote
I've done some research at Google, and seems that the MDS JB920 minidisc model uses ATRAC 4.5, which, talking about MD machines, seems to be inferior only to ATRAC Type-R, which seems to offer some advantages at bit allocation.

You have forgotten ATRAC-S :

Quote
ATRAC Type-S: "Type-S" apparently refers to a new chip containing an improved version of the ATRAC3 codec as well as the ATRAC1 Type-R codec; the combination of the two is called DSP Type-S. From Sony's information sheet: "The two new products MDS-JB980 and MDS-JE780 in the press announcement at CeBIT both adopt the newly developed DSP (Digital Signal Processing) IC chip "CXD2664R." The CXD2664R chip has two DSP functions; one is ATRAC DSP and the other is ATRAC3 DSP. The ATRAC DSP, the TYPE-R algorithm which is the version of the highest grade, is adopted, and the highest sound quality can be obtained at the time of recording. And the ATRAC3 DSP, the algorithm of a newly developed DSP is adopted. It can reproduce the track recorded in MDLP mode in very high sound quality at the time of reproduction. This works in the same way for the disc recorded by other MDLP products. Sony calls the DSP functions of the CXD2664R chip "ATRAC/ATRAC3 DSP TYPE-S". In other words, "ATRAC/ATRAC3 DSP TYPE-S" is a combination of "ATRAC DSP (TYPE-R highest Version)" and "ATRAC3 DSP (newly developed)."


<source>

Poor Performance Of Md Compression

Reply #9 –
Yes there is ATRAC2. Only if you are a developer and have access to Sony's specs.. I think they later decided not to release ATRAC2.. ATRAC2 is very similar to ATRAC3 with minor differences.

wkw

Poor Performance Of Md Compression

Reply #10 –
Quote
You have forgotten ATRAC-S :

Mmm... but ATRAC-S still uses ATRAC-R for high quality compression, doesn't it? Because, as I understand, ATRAC3 is aimed to lower quality low bitrate compression...

Poor Performance Of Md Compression

Reply #11 –
ATRAC3 is targeted at the MP3 / AAC competition.

Poor Performance Of Md Compression

Reply #12 –
I understand it like you. So it mean taht ATRAC 4.5 < ATRAC-R & ATRAC-S 

Poor Performance Of Md Compression

Reply #13 –
Hmm, lots of people think (and hear) MD to be superior to mp3... wonder if they're using Xing or such crap... but no, some of these people seem to be using LAME... 

Poor Performance Of Md Compression

Reply #14 –
Yes. Did you know that « ATRAC3 in LP2 is 132 kbps and sounds as good as(or better than) 320 kbps mp3 »    You can find this here.

Such comments are very common. There are a lot of myths around MiniDISC. You can easily find people in MD forum who claim that MD is superior to CD (it's logical : CD was launched on 1981 - MD on 1992  )

Poor Performance Of Md Compression

Reply #15 –
I own a MiniDisc recorder with ATRAC1 Type-R and ATRAC3 (MDLP) and can say that the algorithm sounds realitivly good, though I like MPC encodings better. ATRAC1 (the most recent version being Type-R) isn't all that impressive considering that it takes 292kbps to do what it does. On some samples where a bell is tolling, I've heard strange crackling noise in the background, and in classical pieces with strong brass passages, I can sometimes detect pre-echo. It seems to compare to mp2 more than mp3 as far as quality at a certain bitrate (though comparisions stop there since ATRAC is a MDCT codec and MP2 is subband). The newer ATRAC3 (not to be confused with ATRAC1 v3.0) does pretty well for the bitrate (132kbps); definitely better than a 128kbps MP3 but to claim it is "as good as a 320kbps mp3" is ridiculous. At 66kbps I'd say that it sounds better than WMA or AAC but not Ogg or MP3Pro. I own both MD and mp3, and like MD for portable use while exrercising, due to the great (60-hour) battery life and cheap media. For more serious listening (while sitting at my desk, etc), I stick to LAME -ape encodings on CD-R's. MD is probably going by the wayside anyway due to Sony's restrictions on the format.

Poor Performance Of Md Compression

Reply #16 –
Quote
The newer ATRAC3 (not to be confused with ATRAC1 v3.0) does pretty well for the bitrate (132kbps); definitely better than a 128kbps MP3.

Did you a complete test ?
I DL OpenMG Jukebox v.2.2 in order to do a fair comparison between the real performance of ATRAC3 (not the ACM codec) at 132 kb/s (pure stereo CBR) and LAME (ABR 132 & joint-stereo). I never did a complete test, as ff123 ones (cf this one and this one) ; i have only encoded a few classical tracks, and listened to it, in order to detect the most obvious artifact. I found some awful one. The LAME encoding didn't produce these artifacts.
I would not conclued that LAME is better than ATRAC3, because I must before did the same thing with LAME (listen to the most obvious artifacts - and them check the ATRAC3 files). And if I remember perfectly, ABX test are not easy to do : ATRAC3 files sound louder than a decoded LAME file. But I'm not sure - I did that test at the beggining of the year.

ATRAC3 files produced by OpenMG didn't convinced me. But I should try again... I'm really interested by your opinion, if you did listening tests.

Poor Performance Of Md Compression

Reply #17 –
In my experience, ATRAC3 seems to be a bit less percpetually annoying than WMA and MP3 at the same bitrate (i.e. the artifacts aren't as salient). However, this scientist thinks a good old-fashioned ABX test is in order here. The only problem is that OpenMG files can't be decompressed to .wav, so I may have to make a digital recording with the MD in LP2, then port it back to the computer. I can think of a few samples from CD's, with heavy brass, crisp hi-hats, classical guitar, etc. that would be good for comparison as Ogg, WMA9, ATRAC3, MP3, AAC and MPC files. If and when I get around to doing such a test, I'll be sure to post the resulting p-values and my interpretations (gotta love SPSS 10.1!).

Poor Performance Of Md Compression

Reply #18 –
Quote
The only problem is that OpenMG files can't be decompressed to .wav, so I may have to make a digital recording with the MD in LP2, then port it back to the computer.

Because I have no MiniDisc, I encoded the files in ATRAC3 via OpenMG, play (decompress) them with the same soft, and did an acquisition with TotalRecorder. Maybe compression/decompression is better (or worst) with a MD unit.

Poor Performance Of Md Compression

Reply #19 –
ATRAC3 shows up a codec within Windows on my machine. I wonder if this acm (wav-header) version is as good as the one in OpenMG?

Poor Performance Of Md Compression

Reply #20 –
I have no trust in this ACM codec. My goal was to test the ATRAC files that are played by a MD unit. OpenMG encoding are the « official » NetMD ones.

Poor Performance Of Md Compression

Reply #21 –
I'll try the Total Recorder trick, since with NetMD, all files are encoded on the machine, not on the MD recorder itself.

Poor Performance Of Md Compression

Reply #22 –
Actually, the ATRAC3 algorithm is very flexible. Its specs does not specified any psychoacoustic model nor even the quantization strategy. As it is licensed to many-many other developers, it is natural that there any varying qualities just as MP3 is... I think there is even room for improvements in the OPEN MG sofware..

Poor Performance Of Md Compression

Reply #23 –
?? Then what can you put in the decoding hardware for ATRAC3? A multipurpose CPU? 

Poor Performance Of Md Compression

Reply #24 –
Quote
I understand it like you. So it mean taht ATRAC 4.5 < ATRAC-R & ATRAC-S 

I think the ATRAC I was talking about is something like ATRAC1 v4.5, because the MD Walkman is not capable of reading/writing in LP-mode (thus ATRAC3).