Jump to content
Sony Insider Forums
Sign in to follow this  
Serge Smirnoff

Hi-SP at SoundExpert testing service.

Rate this topic

Recommended Posts

Hi,

I've just added ATRAC3plus (Hi-SP) to SoundExpert audio quality testing service (www.soundexpert.info) in “Coders 256 kbit/s” group:

ATRAC3plus CBR@256.9 (Hi-SP) - ATRAC3plus (used in Hi-MD and Sony's solid-state and CD walkman products), 256.9 kbit/s FBR

CODER: SonicStage 3.3

- 44100 Hz Stereo

DECODER: SonicStage 3.3

- "Create an Audio CD"

It sounds surprisingly good in comparison with mp3 256. The character of artifacts is quite similar to mp2 one. So it starts with 15.38 points. You are welcome to take part in testing.

Share this post


Link to post
Share on other sites

It sounds surprisingly good in comparison with mp3 256. The character of artifacts is quite similar to mp2 one. So it starts with 15.38 points. You are welcome to take part in testing.

It sounds surprisingly good compared to the source as well, only some high frequency is lacking compared to the original in your test file - and on my computer speakers that's not audible. Had to play the samples on my hi-fi system...

Share this post


Link to post
Share on other sites

Hi,

I've just added ATRAC3plus (Hi-SP) to SoundExpert audio quality testing service (www.soundexpert.info) in “Coders 256 kbit/s” group:

ATRAC3plus CBR@256.9 (Hi-SP) - ATRAC3plus (used in Hi-MD and Sony's solid-state and CD walkman products), 256.9 kbit/s FBR

CODER: SonicStage 3.3

- 44100 Hz Stereo

DECODER: SonicStage 3.3

- "Create an Audio CD"

It sounds surprisingly good in comparison with mp3 256. The character of artifacts is quite similar to mp2 one. So it starts with 15.38 points. You are welcome to take part in testing.

I'm just in process of taking the sound test.

I did 4 of them , 5 more to go.

The last test /sound was incredible, it is sound of sharp bell like instrument, (i don't know how it is called) but the sound is incredible {at least to me it is} ,

It is very healing sound.

I do believe that future medicine will be based mostly on sound and colour lights.

I would like to find title of that CD with "bells" , and with classical music part also.

-I would love to buy it.

Back to listening, .....the quality of the original source and the replicas are really fantastically close.

Thank You Serge Smirnoff for introducing this test.

I will talk to friends (and strangers) about it.

Cheeers, .......

Woow , ....now this next (5th.) classical tune is just fenomenal :)

Edited by Human

Share this post


Link to post
Share on other sites

I'm just in process of taking the sound test.

I did 4 of them , 5 more to go.

It sounds surprisingly good compared to the source as well, only some high frequency is lacking compared to the original in your test file - and on my computer speakers that's not audible. Had to play the samples on my hi-fi system...

Thank you for the response, guys. I have to beg a pardon for not explaining the essence of SoundExpert project as I see some sort of misunderstanding in the replies. Now I’ll try to correct my mistake.

SoundExpert testing engine is completely blind. So when somebody downloads a test file he doesn’t know what codec it belongs to. That is why the resulting ratings are objective and unbiased. If testers know the codec they grade everybody will grade “five” their favorite ones and “one” for others. It’s worth to say that the value of such ratings would be insignificant.

Downloading a test file you get randomly one of the samples of some codec. Now there are 67 codecs rotated in the system. After you have graded the sample SoundExpert automatically “unblindes” the codec of that test file and new rating of that codec appears immediately on corresponding page.

I would like to find title of that CD with "bells" , and with classical music part also.

-I would love to buy it.

For testing 9 different short sound samples are used:

Bach.J.S, "Oster-Oratorium, BWV 24"

Bass (SQAM)

Castanets (SQAM)

French Male Speech (SQAM)

Glockenspiel (SQAM)

Harpsichord (SQAM)

Lo-Fi Analog Tape Recording

Mike Oldfield, "Music From The Balcony"

Quartet (SQAM)

SQAM means Sound Quality Assessment Material – CD published by European Broadcasting Union (EBU) for sound quality evaluations specially. All samples are hi quality recordings made for the purpose.

"Bells" are probably Glockenspiel. So unfortunately it is not a part of some music work - just a sound sample from SQAM.

If you have any questions, please, feel free…..

Share this post


Link to post
Share on other sites

Thank you for the response, guys. I have to beg a pardon for not explaining the essence of SoundExpert project as I see some sort of misunderstanding in the replies. Now I’ll try to correct my mistake.

SoundExpert testing engine is completely blind. So when somebody downloads a test file he doesn’t know what codec it belongs to. That is why the resulting ratings are objective and unbiased. If testers know the codec they grade everybody will grade “five” their favorite ones and “one” for others. It’s worth to say that the value of such ratings would be insignificant.

Downloading a test file you get randomly one of the samples of some codec. Now there are 67 codecs rotated in the system. After you have graded the sample SoundExpert automatically “unblindes” the codec of that test file and new rating of that codec appears immediately on corresponding page.

For testing 9 different short sound samples are used:

Bach.J.S, "Oster-Oratorium, BWV 24"

Bass (SQAM)

Castanets (SQAM)

French Male Speech (SQAM)

Glockenspiel (SQAM)

Harpsichord (SQAM)

Lo-Fi Analog Tape Recording

Mike Oldfield, "Music From The Balcony"

Quartet (SQAM)

SQAM means Sound Quality Assessment Material – CD published by European Broadcasting Union (EBU) for sound quality evaluations specially. All samples are hi quality recordings made for the purpose.

"Bells" are probably Glockenspiel. So unfortunately it is not a part of some music work - just a sound sample from SQAM.

If you have any questions, please, feel free…..

Thank You Serge ,

after listening to first 4 samples the difference in sound quality in my opinion was very close (and i had to listen 5 - 8 times to get a clue), but few next samples were so different that after first 'listening' it was obvious the level of difference between them.

Thank You for posting the titles, some of the CDs will be part of my collection, just by judging by such short sample I already like to buy it.

Actually this test is good so called "coincidence" because today I went to theSource store checking out Panasonic SL-CT820 D.Sound CD-MP3 Player , and to Sony store asking about the Sony's DNE20 MP3/ATRAC3™ CD Walkman, - sales rep at Sony was telling me that when they were listening in the sound room to MP3 and ATRAC3 recorded in the same kbit/s ( i did not ask what level it was) - but He said that it was very-VERY, noticeable difference in sound quality - and ATRAC3 was definite winner.

Also He told me that DNE20 MP3/ATRAC3 CD player has special HD Amp, that is quite better then regular Sony Digital Amp.

I think for now I will get that Sony CD player, and keep my eyes on Hi-MD unit, perhaps RH10,-( because it looks I can not get NH900) , unless some people will report significant mechanical problems with that RH10 model.

So I understand that from the tests You've done the ATRAC3 is also wining, - in Your opinion.

I wonder if the samples that I marked as 5 - if they are by ATRAC3 - I guess You don't have option that we can access our test to see the formats played and our markings..

Thanks for that test == it is GREAT fun, ....

Cheeers,....

.

Share this post


Link to post
Share on other sites

If the two samples are imperceptible from each other, why do we have to choose the degraded sample, which is impossible in this case? Does a random choice have an influence on the results?

I've just compared a few and i like the idea. Keep up the good work.

Share this post


Link to post
Share on other sites

If the two samples are imperceptible from each other, why do we have to choose the degraded sample, which is impossible in this case? Does a random choice have an influence on the results?

I've just compared a few and i like the idea. Keep up the good work.

I was thinking also the same, about choosing one as degraded sample - even when i was going to give them 5 mark.

But now i mark only as 5 (no degrading choice), if they sound extremely close or the same.

Only below 5, I mark one of them as degraded sample - I think this is the idea [??] that 5 gets no degrading choice.

P.S. ...greenmachine ...looking cool in blue :)

Share this post


Link to post
Share on other sites

If you don't specify a degraded sample (at rating 5), your rating will not be accepted, this is why i was asking.

Is there a specific reason why there is an absence of medium bitrates (~128-192)?

Share this post


Link to post
Share on other sites

So I understand that from the tests You've done the ATRAC3 is also wining, - in Your opinion.

The ratings will change slightly as new participants will add their grades but the difference between ATRAC3plus (256) and mp3 (256) is too big in order they swap their places in future.

I wonder if the samples that I marked as 5 - if they are by ATRAC3 - I guess You don't have option that we can access our test to see the formats played and our markings..

Right you are, there is no such access. But I can say that for the last two hours there were no ATRAC3plus(256) samples downloaded ;)

If you don't specify a degraded sample (at rating 5), your rating will not be accepted, this is why i was asking.

No, if you grade "5" it doesn't matter which sample you marked as "degraded". Your mark will be accepted anyway. Well, I will think how to make it more clear in future.

Is there a specific reason why there is an absence of medium bitrates (~128-192)?

Just shortage of time. New testing methodology which allows to test reliably such high bit rates was introduced only this summer (you can look at news on the site). 128 group will be created till the end of this year.

Share this post


Link to post
Share on other sites

The ratings will change slightly as new participants will add their grades but the difference between ATRAC3plus (256) and mp3 (256) is too big in order they swap their places in future.

Right you are, there is no such access. But I can say that for the last two hours there were no ATRAC3plus(256) samples downloaded ;)

Wooow - !!! , no ATRAC3 in last 2 hours, - I was positive that the every close sounds were from ATRAC3 - Wooow = so the conclusion is that MP3 could be so good, - Wooow !

Serge, - since english is not my first language - Please explain bit more this statement of Yours:

>" The ratings will change slightly as new participants will add their grades but the difference between ATRAC3plus (256) and mp3 (256) is too big in order they swap their places in future."<

Thanks, .....

.

Edited by Human

Share this post


Link to post
Share on other sites

Wooow - !!! , no ATRAC3 in last 2 hours, - I was positive that the every close sounds were from ATRAC3 - Wooow = so the conclusion is that MP3 could be so good, - Wooow !

Actually, when you download a test file you can get a sample of any codec rotated at SoundExpert (now 67 codecs, as I've sad already).

Serge, - since english is not my first language - Please explain bit more this statement of Yours:

>" The ratings will change slightly as new participants will add their grades but the difference between ATRAC3plus (256) and mp3 (256) is too big in order they swap their places in future."<

This is not necessarily your fault – my English is not perfect as well.

I mean, both ratings in 256 group will change slightly after some time but this change will not be dramatic, superiority of ATRAC3plus (256) over mp3 (CBR,256) will remain for sure.

Share this post


Link to post
Share on other sites

Actually, when you download a test file you can get a sample of any codec rotated at SoundExpert (now 67 codecs, as I've sad already).

This is not necessarily your fault – my English is not perfect as well.

I mean, both ratings in 256 group will change slightly after some time but this change will not be dramatic, superiority of ATRAC3plus (256) over mp3 (CBR,256) will remain for sure.

Aha , - now i understand - Thanks :)

As to the ATRAC3 and MP3, - do You think it would be possible to download and hear the same sound sample with MP3 and Atrac3 side by side ?

So after all ATRAC 3 ...is the king :)

...but sounds that MP3 is also capable of great sound quality judging by some of the samples - ( Ref: to last 2 hours of downloads). If the really good ones were indeed MP3 sound samples?? :)

For now I must leave, see You in few hours, Thanks for very interesting experience

Cheeeers,...

Share this post


Link to post
Share on other sites

The quality of a mp3 will strongly vary depending on the encoder and settings used. Fraunhofer's encoder with disabled m/s joint stereo and disabled low pass filter (?) might not be the optimal solution for highest quility. Using Lame 3.97b1 "-V 0", "--abr 256" or - if it has to be cbr - "-b 256" could improve the situation and surpass A3+ 256, although i doubt i could abx at such bitrates. Conclusion: You can't generally say mp3 sounds worse than A3+ at a specific bitrate unless you specify encoders and settings each time.

Share this post


Link to post
Share on other sites

The quality of a mp3 will strongly vary depending on the encoder and settings used. Fraunhofer's encoder with disabled m/s joint stereo and disabled low pass filter (?) might not be the optimal solution for highest quility. Using Lame 3.97b1 "-V 0", "--abr 256" or - if it has to be cbr - "-b 256" could improve the situation and surpass A3+ 256, although i doubt i could abx at such bitrates.

Lame 3.97b1 -V 0 is in "224" group of SoundExpert. Bit rate range where mp3 realy shines is 128-192. The higher, the less efficient this format.

Conclusion: You can't generally say mp3 sounds worse than A3+ at a specific bitrate unless you specify encoders and settings each time.

Agree. But the gap between this two formats at SoundExpert seems to me too big that any other mp3 encoder could shorten it substantialy. Anyway, time and tests will show. I plan to add Lame 320CBR soon. It could be revealing in the sence of highest possible mp3 quality.

Share this post


Link to post
Share on other sites

As to the ATRAC3 and MP3, - do You think it would be possible to download and hear the same sound sample with MP3 and Atrac3 side by side ?

NO with high probability. But if you want to compare directly these two formats you may encode some sound excerpt with Lame and SonicStage by yourself and listen them. :rolleyes:

So after all ATRAC 3 ...is the king :)

Well, among two formats and at 256 kbit/s - probably yes B)

...but sounds that MP3 is also capable of great sound quality...

Definitely, at 128-192 kbit/s it could outperform ATRAC3plus. We’ll see soon.

Share this post


Link to post
Share on other sites

Right now, I always get the same zip file (and yes, I did report the results, did cut-and-paste the file name, for testing purpose tried the same file a second time which wasn't then accepted due to it being the second time...)

Share this post


Link to post
Share on other sites

Right now, I always get the same zip file (and yes, I did report the results, did cut-and-paste the file name, for testing purpose tried the same file a second time which wasn't then accepted due to it being the second time...)

What programm do you use for downloading?

Share this post


Link to post
Share on other sites

Right now, I always get the same zip file (and yes, I did report the results, did cut-and-paste the file name, for testing purpose tried the same file a second time which wasn't then accepted due to it being the second time...)

I had the same problem few times , - the file was "stubborn" and i could not get rid of it - in order to download next one, - but by clicking few times on X, and deleting , and I found that mostly by leaving alone for 30 -60 -90 seconds it seems that solved the blockade , and the next file was downloading OK.

Edited by Human

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this  

×