воскресенье, 28 Сентябрь 2008

Nero AAC Encoder 1.3.3.0

A few days ago the new version of free Nero AAC Encoder (1.3.3.0) became available. According to its change log there are no significant improvements affecting quality of stereo encoding though. But taking into account that previous version (1.1.34.2) didn’t participate in SE testing at all and any version of this encoder never tested at SE at low bit rates where it usually shows the best efficiency we decided to perform extensive testing of the new encoder starting from low bit rates (thank you, Lance Penguin for this proposal). First four settings added are VBR: -q 0.15, -q 0.20, -q 0.25, -q 0.35. They dropped into the sections respectively: 32 kbit/s, 48 kbit/s, 64 kbit/s, 96 kbit/s.  We intentionally did not tune resulting bitrates with q – parameter but exercised commonly used anchor values instead. Fine-tuned contenders can be added later if needed. As usually first reliable results will appear in a week or two. Thank you all for participation in listening tests and Vito Marolda personally for donation.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 12:52 AM in SoundExpert news

среда, 23 Июль 2008

Lame 3.98 (-V 0) added

Next setting of new Lame mp3 encoder (3.98) to be tested at SE is “–V 0” – the highest quality VBR mode of the encoder. Resulting in 231.9 kbit/s on SE test files it fell into 224 kbit/s group. As usually first reliable results are expected in a week or two. But you can help SE to improve some of its green (grey) ratings faster - just take part in testing … if you want.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 11:42 PM in SoundExpert news

воскресенье, 6 Июль 2008

Lame 3.98 in 128 kbit/s group

We must thank developers of Lame mp3 encoder for their time and efforts - Lame 3.98 released to public. In order to compare the new encoder not only with other encoders but also with previous Lame versions SE will add new Lame with the settings which are the same as or similar to the ones already used in Lame SE testing. It will depend on resulting bit rates produced by new Lame on SE test samples, though.

First setting is "-V5" - joint stereo variable bit rate mode resulting in 131.3 kbit/s on nine SE test samples. So it perfectly fits 128 kbit/s group at SE.  Previous version (Lame 3.97b2) when using this setting (-V5) showed lower bit rate - 112.7 kbit/s and its comparison with other contenders producing around 128 kbit/s output was not very honest to be true. Now with 131.3 kbit/s the competition will be more fair.

Next setting planned for testing is "-V0" - highest quality VBR mode. It was also used for testing of Lame 3.97b1 at SE. If you have some other suggestions, please, let us know.

Thanks for participation in testing.
The People from SE

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 4:16 AM in SoundExpert news

воскресенье, 29 Июнь 2008

Bluetooth audio quality (A2DP)

Bluetooth audio devices become more and more popular today – stereo headsets and headphones, wireless audio transmitters/receivers, cell phones and mp3 players with Bluetooth Advanced Audio Distribution Profile (A2DP). Headphones and headsets seems benefit the most from Bluetooth (BT) audio. To be honest first BT stereo headphones were of poor sound quality – dirty sound with annoying “hiss” on background. Definitely that was not because of BT technology limitations but mainly because of poor implementation of the latter and slipshod design of analog audio circuits. The situation changes slowly but steadily. For example modern BT stereo headphones DR-BT50 from Sony can successfully compete with ordinary wired ones already. Their sound quality will be sufficient for vast majority of portable player owners for sure. In no doubt if manufacturers decide they could produce perfectly sounding BT headphones for demanding listeners as well. In that case the way digital audio transmits over BT protocol is a bottleneck in the whole audio transmission chain.

As the data channel used by A2DP is only 721 kbps wide some data reduction scheme is required. While BT specifications allow using of different audio codecs (mp2, mp3, wma, aac and even atrac) the only mandatory codec for all BT audio devices is subband codec (SBC). In fact many BT headphones (including above mentioned DR-BT50) already support mp3 codec but in real life it remains unused. Obviously it was made for the sake of compatibility but the same time portable devices could benefit from direct streaming of mp3 files through BT interface saving both audio quality and battery life. However in today's practice all high quality audio still transfers over A2DP by means of SBC codec.

SBC codec appeared in SoundExpert ratings more than two years ago (see 320+ kbit/s section). The bitrate used for testing (372 kbps) showed good potential of this compression scheme but it’s clear now that above setting doesn’t reflect real-life scenario of the codec use. Let’s look for example at these two major BT applications:

Toshiba Bluetooth stack for Windows (6.10) has three quality modes for SBC codec
  • High Quality – 328 kbps
  • Middle Quality – 229kbps
  • Low Quality – 201 kbps

IVT Corp. Bluesoleil (5.0.5) has two quality modes: High and Middle. It’s not mentioned what bitrates are used but the application has special “Status Window” available for any BT connection. Among other information it shows amount of bytes sent to and received from connected device. Simple calculations reveal the same bit rates for High and Middle quality modes – 328 kbps and 229 kbps. As Bluesoleil is capable of both sending and receiving A2DP audio data its status window helps to discover actual bitrates used by any connected BT device. For example cell phone Nokia 6500 classic also sends music to BT interface at 229 kbps and the bitrate can’t be altered in phone’s settings (may be some advanced music phones and players have such possibility, though).

Such unanimity in choosing SBC codec settings is not surprising. Bluetooth Special Interest Group (SIG) recommends using of those settings in A2DP specifications (A2DP_SPEC, Revision V12):

 

However, it is recommended for the encoder to support the following settings shown in Table 4.7

Table 4.7: Recommended sets of SBC parameters in the SRC device
SBC encoder settings* Middle Quality High Quality
Mono Joint Stereo Mono Joint Stereo
Sampling frequency (kHz)
44.1 48 44.1 48 44.1 48 44.1 48
Bitpool value
19 18 35 33 31 29 53 51
Resulting frame length (bytes)
46 44 83 79 70 66 119 115
Resulting bit rate (kb/s)
127 132 229 237 193 198 328 345
*Other settings: Block length = 16, Allocation method = Loudness, Subbands = 8

 

Taking all this into account it was decided to add SBC codec to SE testing engine with these new settings:


It’s worth to add that A2DP limits the available maximum bit rate to 320 kbps for mono and to 512 kbps for stereo modes allowing the use of many other codecs besides SBC. So audio manufacturers (of headphones especially) have wide choice of compression technologies capable of delivering high definition sound through the wireless digital channel at reasonable cost.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 4:29 AM in Audio quality

воскресенье, 25 Май 2008

AAC VBR@256 from iTunes

New codec in 256kbps section – AAC VBR from iTunes. Now both CBR and VBR modes of iTunes AAC codec @256kbps are under test. Definitely VBR mode consumes a bit more bits for encoding SE test files. Does this lead to any quality improvements? We’ll see. SE thanks Bill Justin for this codec suggestion.

 

        Slashdot   Slashdot It!
Posted by Serge Smirnoff at 12:25 AM in SoundExpert news

понедельник, 19 Май 2008

eAAC+ @48 kbps from Winamp 5.53

Low bitrate codec eAAC+ @48 kbps widely used for uploading music to mobile phones is added. Thank you Gert (aka Snoyt) for the proposal.

 

        Slashdot   Slashdot It!
Posted by Serge Smirnoff at 1:58 AM in SoundExpert news

пятница, 18 Январь 2008

HE AAC 64kbit/s codec from Winamp added

HE AAC codec from Winamp (aac he, cbr, 64 kbit/s) was added to SE testing system. Low bitrate sections of SE are a bit outdated at the moment, so starting with this codec we plan to fix the situation soon. SE thanks Giulio Franco for this codec proposal.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 2:37 AM in SoundExpert news

пятница, 9 Ноябрь 2007

New iTunes AAC added

Finally AAC encoder in iTunes was updated. The new one is included in iTunes 7.4 shipped with new Mac OS Leopard. Starting from version 7.5 iTunes for Windows also has this AAC coder. The bug of previous release is completely gone and resulting bit rates of encoded files are a little bit lower. That’s all to be said for sure in the beginning of testing it at SoundExpert.

The testing is started with CBR@256kbit/s setting which is used in iTunes music store. It looks like this bit rate will be the most popular soon. In a week or two first reliable results will show new encoder potential. If you want to get them faster, download and grade a test file from SoundExpert. Please, remember, the test file that will fall to your share will not necessarily belong to the newly added codec, it can be any of 95 codecs rotated in SE at the moment. Blind testing is blind testing. Thank you in advance for participation.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 12:18 PM in SoundExpert news

пятница, 19 Октябрь 2007

WMA 9.2, VBR:Q90 added

Wide support of WMA format along with MP3 by manufacturers of portable players and powerful marketing strategy of MS did the job – WMA is very popular.

Q90 setting was added to 192 kbit/s group although it renders SE test items at average 160 kbit/s bitrate. If there are some other contenders in this bitrate sub-group, new page "Coders 160 kbit/s" will be inserted.

In case of current activity of volunteer testers first reliable values for the rating will appear in a fortnight. SE thanks Stig Lykke Iversen for this codec proposal.

Try to grade one of the 94 codecs rotated in SE system at the moment (ZIP: ~3Mb)

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 12:03 PM in SoundExpert news

понедельник, 15 Октябрь 2007

SoundExpert system in simple words

New article explaining SoundExpert functionality in simple words appeared on Testing methodology page. Authors would be thankful for your feedback here.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 5:44 PM in SoundExpert news

понедельник, 27 Август 2007

Testing of iTunes AAC@256kbit/s finished

Finally the rating of iTunes AAC codec @256 kbit/s has got into 5% reliability limits and can be safely compared to other SoundExpert (SE) ratings of appropriate reliability.  This codec was added to the rating system shortly after EMI/Apple announcement of forthcoming sales of DRM-free and higher quality AAC@256 music files in iTunes Store. The testing ended up with 5.71 points of perceived audio quality for the codec. Not very impressive indeed. One of the reasons is the bug mentioned earlier. The other reason is that the coder was probably optimized for lower bitrates because its quality increases from 128 kbit/s more slowly in comparison with other codecs.  

But fortunately for music lovers (and unfortunately for SE) this rating might be pretty useless in the nearest future. There are persistent rumors out there that new Mac OS Leopard scheduled in October will have completely new AAC coder without that bug. And what is more there is no evidence that music in iTunes Store was coded exactly with the coder from iTunes and not the new one. When new AAC coder is available for public it will be added to SE for sure.

Meanwhile the latest iTunes Jukebox still uses AAC codec from previous versions, exactly the one tested here at SoundExpert. So its rating is still useful for Jukebox users. And still … Nero AAC coder is much better.

Thank you all participants of SE listening tests.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 9:49 PM in SoundExpert news

суббота, 26 Май 2007

Other Russian music stores besides Allofmp3

Making hard work of IFPI and RIAA preventing piracy in Russian Federation a little bit easier here is a list of other Russian on-line music stores besides far-famed Allofmp3 operated under the same licenses from collective rights management organizations as the latter. They and their customers could be successfully alienated exactly the same way – by means of broad educational campaigns in mass media and lobbying on a state level (questioning of Russia’s participation in G8 could be a point for example). All stores listed bellow are fully accessible outside Russia and operate in collusion with Visa, MasterCard and many other international payment services. Some important details of their illegal activities are also included.

GoMusic.ru - http://www.gomusic.ru
Catalog: artists 28 405, albums 35 815, (X72)*
Audio format: mp3 192 vbr
Price: $0.19 per song, $2-$3 per album

LegalSounds.com - http://www.legalsounds.com
Catalog: (X40)*
Audio format: mp3 128-320
Price: $0.09 per song, $1-$2 per album
    
MP3Sale.ru - http://www.mp3sale.ru
Catalog: artists 20365, albums 83397, (X66)*
Audio format: mp3 128-320
Price: $0.15 per song, $1-$3 per album

MP3fiesta.com - http://www.mp3fiesta.com
Catalog: artists 15 907, tracks 685 494, albums 62 316, storage 4.5Tb, (X53)*
Audio format: mp3 128-192 cbr/vbr
Price: $0.10 per song, $1 per album

MP3Sugar.com - http://www.mp3sugar.com
Catalog: artists 15 057, albums 49 955, tracks    588 357, storage 4.2Tb, (X62)*
Audio format: mp3 192 vbr
Price: $0.18 per song, 10% discont for an album

MusicMP3.RU - http://musicmp3.ru
Catalog: (X18)*
Audio format: mp3 192 vbr
Price: $0.12 per song

mp3stor - http://www.mp3stor.com
Catalog: artists 27 152, albums 76 396, tracks 902 641, storage    5.305Tb, (X78)*
Audio format: mp3 192-320 vbr/cbr
Price: $1-$3 per album

iomoio - http://www.iomoio.com
Catalog: (X160)*
Audio format: mp3 192 vbr/cbr
Price: $0.15 per song

MP3spy - http://mp3spy.ru/en
Catalog: (X200)*
Audio format: mp3 192
Price: $0.15 per song, substantial discont for an album

justmusicstore.com - http://www.justmusicstore.com
Catalog: artists 44 120, albums 159 890, tracks 1 214 560, storage 14.2Tb
Audio format: mp3 192
Price: $0.15 per song

MusicFond - http://musicfond.com (in russian only)
Catalog: (X86)*
Audio format: mp3 192 vbr
Price: free (one by one track only)

AudioStore - http://audiostore.ru
Catalog: (X15)* world and indie music mostly
Audio format: mp3 256
Price: $0.30 per song, $3 per album

CDonPC - http://eng.cdonpc.ru
Catalog: albums 53 011, artists 16 993, tracks 545 959, (X70)*
Audio format: mp3 192 vbr/cbr
Price: $0.15 per song ($0.13 within full album)

 


* Number of artists in catalog on the letter "X". It gives some additional comparative estimation of catalog size. Other figures (artists, albums, tracks, storage) are taken from site info.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 3:40 PM in Audio culture

вторник, 17 Апрель 2007

How good will iTunes Store AAC@256 files be?

EMI/Apple announcement of forthcoming sales of DRM-free and higher quality AAC 256kbit/s songs from iTunes music store is heavily discussed now. While almost everybody agree that this is a move to the right direction (final point probably means reasonably priced lossless downloads) there is no consensus about the reasons of choosing exactly 256 kbit/s. This bitrate seems too high for an average listener (which is quite happy with existing AAC 128kbit/s) and the same time it’s definitely inadequate for demanding listeners especially taking into account that Apple’s AAC encoder is not state-of-the-art one. Most prolate guesses explaining the choice are:

  1. Stimulation of higher-cappasity iPod sales as increased bitrate needs doubling of memory size for storing music
  2. Cleaver decision from the marketing standpoint – consumer gets twice as good sound quality for only 30% price increase. Now songs look even cheaper
  3. Stimulation of legal downloads as they are of better quality than most music in P2P networks (128-192kbit/s)
  4. Getting of additional advantage over Microsoft since many digital audio players supporting WMA are limited by 192kbit/s for this audio format
  5. Providing with some headroom for inserting watermarks which can help to track iTunes songs in P2P networks

Watermarking point seems to be a wild assumption though because …. Apple has much wiser marketing strategy than the RIAA does. It will take (let me guess) one week for A-students from any university campus to discover the fact and the principles of watermarking used and one week more for internet community to discuss possible consequences for file-sharers, music industry and Apple itself  and to develop methods for destroying watermarked information. After that the whole following year any person interested will throw mud at Steve Jobs and the Co. So watermarking technique is conceived only for research purposes (like birds tagging) which have to be clearly stated beforehand.

SoundExpert contributes its 2 cents to the discussion by adding AAC@256 from the latest iTunes to the rating system. Though new audio metrics used by SoundExpert is still experimental it’s the only testing methodology today capable of measuring perceptual audio quality margin. It helps not only to grade the quality of AAC@256 audio format but also to compare it with other formats such as mp3 or wma at different bitrates.

Blind listening tests conducted over the internet are essential part of the new metrics. These tests designed in a way that everyone, including you, can participate. Just download a test file, listen it and feedback your grade (all details inside test file).  Downloading a test file from SoundExpert you can get randomly one of 93 audio codecs rotated in the system at the moment including newly added AAC 256 kbit/s codec from iTunes.

First results for the rating are available on “Coders 256 kbit/s” page. Light-grey color of the bar indicates low reliability of the rating caused by insufficient number of returned grades. In order to be reliable each rating needs about 300 test files to be downloaded and graded. So, it will take some time (depending of volunteer testers’ activity) to get final results for iTunes AAC@256 codec.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 7:38 PM in SoundExpert news

воскресенье, 1 Апрель 2007

The 1st of April – Audiophiles Day

Audiophiles Day Yes, this is our holiday. It’s really funny that most of us having good education and common sense are easily assified by marketers and sales persons. Have you ever dreamed about those fascinating tube amplifiers with miraculous warm sound and hot air over device? (recommended for countries with cold climate). And how about those funny tiny glittering cones, advanced power cords and precious interconnection cables with very strict and serious arrows? You might think that wireless audio connections will put an end to this cable cheating. I’m not sure; I’m waiting for the first air conditioning system specially designed for listening/home theater rooms (... you know, some wireless audio digits could be slow because of dust particles causing so called “air-jitter” ...). It’s worth to mention the quality of modern over-compressed audio CDs and prices of iTunes songs with just tolerable 128kbit/s quality and restricted usability. And finally if you still want to have clear sound at home you have to buy Blue-Ray player and wait a little bit for all your favorite music reissued on new media by recording industry. Looks like joke, isn’t it? Definitely it’s our Day.

Why this happens. In layman’s terms this could be explained as follows. On the one side there are users of audio equipment (audio consumers) – emotional and even passionate music lovers having only their subjective impressions for grading sound quality and measuring their satisfaction. In practice it’s too hard to evaluate sound quality ignoring music used, appearance of audio equipment, its brand and price. Very often consumers are disunited, misinformed and manipulated by the other side – manufacturers. The latter are better organized and like any business they have clear goal and simple parameter for measuring their satisfaction – profit. Do you think now the manufacturers are the evil? In no way! Quite the contrary they operate almost perfectly using every market opportunity to be successful. The problem is that they don’t receive appropriate feedback from consuming side. Formally we can vote with our money for one or another manufacturer but we actually don’t know for whom. Our decisions are mostly emotional and driven by false concepts and ideas. In short: manufacturers know almost everything about sound quality, consumers – very little. Economists will tell you that such asymmetric information leads to marketing of low-quality goods as higher-quality ones. And we ourselves are responsible for the situation. We just love miracles and ready to be fooled by supplying side. Definitely it’s our Day.

What to do. Generally talking we (consumers) have to be better informed and organized in order to be able to influence the audio market which is driven by audio manufacturers today. Being more specific our purchasing decisions:

  • have to reflect our real needs
  • have to be cooperative

Obviously this is not possible without both existence of audio quality metrics and broad interchange of information and opinions between members of audio community. While reliable hearing-wise quality metrics is still under development (SoundExpert proposed its own one) effective communication environment already appeared. Forums, blogs, social networks and collaborative recommendation services are excellent places for consumer’s cooperation (and even for cooperation between both sides – consuming and producing). Of cause, due to democratic and free nature of discussions the results can also be misleading and even completely unexpected (favorite example from my own experience is on Winamp forum where I announced possibility of testing high bitrate coders at SoundExpert; few minutes later my post was suspected of SPAM and “changed slightly” by admins; since then I visit this wonderful thread from time to time as a self-training practice). Nevertheless freedom of expressing opinions, critical state of mind and self-irony are the virtues equally necessary for audiophiles and the market nowadays.

Have a nice Day! Our Day.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 4:17 AM in Audio culture

пятница, 23 Март 2007

ATRAC3plus 352 kbit/s added

ATRAC3plus codec from SonicStage 4.3 was added to 320+ group:

ATRAC3plus CBR@360.6 - ATRAC3plus 352 kbps (360.6 kbit/s FBR)
ENCODER: SonicStage 4.3.01.14050
- Import a CD
- Format: ATRAC
- Recording Quality: High
- Bit rate: 352 kbps
- 44100 Hz Stereo
DECODER: SonicStage 4.3.01.14050
- Save in WAV Format

Please don’t forget that each rating needs about 300 returned grades in order to be reliable. So initial values of any newly born rating are pretty random. This is indicated by color of bar – lighter ones are less representative, less reliable and need more grades to be returned. The system offers test files of such unreliable ratings a bit more often trying to maintain approximately equal reliability for all ratings.

SoundExpert thanks Osiris Walstra, Navneet Alang and others for this codec proposal. If you’d like some other codecs to be added to the rating system, please drop us a line.

You can help this and other ratings become reliable faster by downloading and grading a test file. It can be any of 92 codecs rotated in the system at the moment.

 

        Slashdot   Slashdot It!

Technorati Tags:

Posted by Serge Smirnoff at 12:41 PM in SoundExpert news