Ministry of Sound chases thousands of file-sharing Brits
Lawyers for UK dance music label Ministry of Sound have issued thousands of letters to British filesharers this month demanding compensation for alleged copyright theft.
The letters threaten recipients with immediate court action unless they pay immediate compensation in the region of £350.
Speculative invoicing?
London-based music IP lawyers Gallant Macmillan – a company that claims to be a 'leader in media and litigation' – mailed out 2,000 letters to Brits earlier this month claiming infringement of its client Ministry of Sound's copyright by downloading and sharing MoS albums online.
Another British legal firm ACS:Law has also sent out thousands of similarly threatening letters to Brits recently, requesting instant payment or the instigation of court action against the recipient of the letter.
The Guardian quotes Luke Bellamy, one of the alleged file sharers in question that received a letter from ACS: Law asking for an instant payment of £295 in return for downloading and sharing a tune by Cascada.
"Getting a letter like this is extremely worrying," Bellamy said. "I have never downloaded anything from this website and yet I am being chased for this money. My parents have been worried by this, and frankly I've got better things to do with my time than deal with this."
Michael Coyle, solicitor advocate and MD of Southampton-based law Lawdit is going so far as to offer to defend anyone that receives such letters from IP lawyers in court for free. Providing they genuinely didn't download and share the copyrighted material in question, that is!
Deborah Prince, head of legal affairs at Which? says: "Which? believes it is outrageous that ACS:Law is asking consumers to provide evidence to support the claims that it is making on their clients' behalf. It should have all the evidence it needs before making these allegations. If it doesn't, then it shouldn't be asking unrepresented consumers to provide that evidence."
Via The Guardian
 |  |
Read More ...
How face detection works
The first time I looked at the rear display of a camera that had face-detection software, it was an interesting experience. Point it at a person, and the software would superimpose a coloured square over that person's face. This would enable you to more easily frame the photo, ensure correct exposure for the face compared to the rest of the scene and make sure that the face was properly focused. So how did it manage it?
What's so special about a face that enables the camera to identify that this set of pixels is a face, but that set isn't? And in real-time too?
The camera doesn't have a chip with great processing power, either, so the algorithm must be extremely efficient. We should also remember that over the years, camera face-detection software has become pretty advanced. You can now expect the software in your point-and-shoot camera to work out not just the location of a face but also whether the person is smiling, and to take the photo automatically if so.
Back in 2001, Paul Viola and Michael Jones invented a new framework for detecting arbitrary objects and refined it for face detection. The algorithm is now known as the Viola-Jones framework. The first thing to realise is that face detection, whether by Viola-Jones or not, is not an exact science.
Just like we humans can be fooled by images that seem to contain a face when in reality they do not, so face-detection software can be hoodwinked. This phenomenon is known as pareidolia: the apparent recognition of something significant (usually a face or a human form) in something that doesn't have it naturally.
There are many examples of this, the most prominent being perhaps the Face on Mars – a photo taken in the Cydonia region of Mars that appeared to contain a human face in the rock – or the image of the Virgin Mary that an American lady found in a grilled cheese sandwich.
Face detection software can be fooled too, so we talk about the algorithm's rate of false positives (detecting a face when there is none) or false negatives (not detecting a face that's present).
A good guess
The Viola-Jones method has a very high accuracy rate – the researchers report a false negative rate of less than one per cent and a false positive rate of under 40 per cent, even when used with the simplest filter. (The full framework uses up to 32 filters, or 'classifiers'.) But we're getting ahead of ourselves.
The breakthrough for Viola and Jones came when they didn't try to analyse the image directly: instead, they started to analyse rectangular 'features' in the image. These features are known as 'Haar-like features', due to the similarity of the analysis of complex waveforms with Haar wavelets. These are simple square waveforms, and are named after Alfréd Haar, a Hungarian mathematician whow as working at the turn of the 20th century.
The first thing to do is strip colour from the image and work with a simple greyscale colour space. A simple way of doing this is to transform the image to the 'YCbCr' colour space, where the Y component is a measure of intensity or luminance in the original image (the other components can be ignored for now, although they could be used to distinguish flesh tones at a later detection stage).
To calculate the Y component (the luma) of a pixel, use the formula Y = 0.299R + 0.587G + 0.114B, where the values R, G, B are the red, green and blue components of a pixel's value.
Now that we have an image defined as a set of intensity values (from 0 to 255), we can use them to look for large-scale rectangular features in the image. We do this by summing the intensity values in various rectangular blocks.
Using these sums we can detect 'darker' blocks adjacent to 'lighter' blocks since the sum of the intensity pixels in a dark block will be less than the sum in a light block. These adjacent blocks are known as 'features'. (Note that, despite the fact we're using Viola-Jones to detect faces, the word 'feature' does not pertain to facial features in any way.)
Viola-Jones defines several different types of features (see Figure 1). The first feature is a light block next to a dark one, vertically. The second is the same but horizontally. The third is a light block sandwiched between two dark blocks (or vice versa). The fourth is a four-rectangle feature as shown.
FIGURE 1: Types of features used in Viola-Jones
Note that in a feature the rectangular blocks are the same size. The features can be at any scale or at any position in the image, so the features shown in Figure 1 are just examples at an arbitrary size. A particular feature can be positioned and scaled onto the original image.
The feature value is calculated as the sum of the pixel intensities in the light rectangle(s) minus the sum of the pixels in the dark rectangle(s). The value of the feature is then used in a filter to determine if that feature is 'present' in the original image.
It still sounds computationally expensive (that is, slow). To improve the speed of summing the intensities of the pixels in a given rectangle, we make use of a trick. For a given image we can calculate what's known as the integral image at every point.
This is merely the sum of the pixels of the rectangle from the upper-left corner to the given point, and it can be calculated for every point in the original image in a single pass across the image. The word 'integral' comes from the same meaning as 'integrating' – finding the area under a curve by adding together small rectangular areas.
Intense calculations
Once we have the integral image for every point, we can calculate the sum of intensities for any arbitrary rectangle by using the identities shown in Figure 2. We want to calculate the sum of the pixels for the rectangle abcd.
FIGURE 2: Calculating sums of intensities using integral images
We start off with the integral image for point a, and then subtract the integral images for points b and d. This unfortunately takes off too much (as shown by the double hatching) and so we add back in the integral image for point c. As you can see, it's a very quick calculation once we have generated all the integral images, presumably stored in an array.
We now have in place all the requisites to calculate the summed intensity values for a set of features. But what are we going to do with them? Compared with analysing the pixels directly, features provide a coarse, low-resolution view of the image. They're good at detecting edges between light and dark, bars, and other simple structures.
What we do is to implement a learning system. We give the face detection routine a set of 24 x 24 images of faces (and another set of 24 x 24 images of things that are not faces) and train the routine to recognise faces and discard non-faces. Viola and Jones used a database culled from the internet of about 4,000 faces and 10,000 non-faces to do the training. What's involved in the training?
Using 24 x 24 images, there are some 45,000 different ways to place one of the four types of feature onto the image. For example, for the first type of feature, you could have rectangles one pixel wide by two pixels deep, all the way up to 1 x 24, then 2 x 2 to 2 x 24, and so on. These different-sized features would be placed in various positions across the image to test every possible feature for every possible size at every possible position.
Note that the number of possible features, 45,000, is far greater than the number of pixels in a 24 x 24 image (a mere 576 pixels) and so you must reduce the number you use. Remember, you're calculating the difference between the pixel sum for the light and dark parts of the feature.
You could decide on a threshold for difference (which could be tweaked during training) whereby a feature is assumed to be detected or not. Using this, you would then apply every one of the 45,000 possible features to your training set.
What you'd find is that certain features are worthless at determining whether an image is a face or not – that is, there would be no correlation between the feature identifying a face and it not being one, and vice versa. These would be rejected.
However, some would have a high success rate at rejecting non-faces, and this is where the training comes in.
Face or not?
Viola and Jones then experimented with the remaining features to determine the best way of using them to classify an image as 'face' or 'non-face'. After experimentation they decided to use a training system called AdaBoost to build a classifier.
AdaBoost is an artificial intelligence (AI) technique similar to a neural network, devised to combine 'weak' features into a 'stronger' classifier. Each feature within a classifier is assigned a weighting (tweaked during training) that defines how 'accurate' that classifier is. Low weighting means a weak feature, high weighting a stronger one.
Add up the weightings of the features that test positive for a particular image and if that sum is above a threshold (again, tweakable during training) then that image is determined to be a face.
As it happens, during this training they found that there were two features that, when combined and properly tuned by AdaBoost into a single classifier, would pass though 100 per cent of the faces with a 40 per cent false positive rate (60 per cent of the non-faces would be rejected by this classifier).
Figure 3 shows this simple classifier in action. It uses two features to test the image: a horizontal feature that measures the difference between the darker eyes and the lighter cheekbones, and the three-rectangle feature that tests for the darker eyes against the lighter bridge of the nose.
FIGURE 3: The first classifier on a 24 x 24 image of the author's face showing the two features in use
Although they had been trying to implement a strong classifi er from a combination of 200 or so weak classifi ers, this early success prompted them to build a cascade of classifiers instead of a single large one (see Figure 4).
Each subwindow of the original image is tested against the first classifier. If it passes that classifier, it's tested against the second. If it passes that one, it's then tested against the third, and so on. If it fails at any stage of the testing, the subwindow is rejected as a possible face. If it passes through all the classifiers then the subwindow is classified as a face.
FIGURE 4: The Viola-Jones cascade of classifiers
The interesting thing is that the second and subsequent classifiers are not trained with the full training set. Instead they are trained on the images that pass the prior classifiers in the chain.
The second and subsequent classifiers are more complex and have more features than the first one, so they require more computational time. It seems remarkable then that there is such a simple classifier that will reject so many sub-windows without having to go through the calculations required for the more complex classifiers.
Remember that it's far more likely that a random image contains no face at all, and so being able to reject the majority of sub-windows with as small a set of calculations as possible is a good thing.
Final reckoning
The eventual cascade developed by Viola and Jones had 32 stages and used a total of 4,297 features (out of the original total of 45,000). The first classifier uses the two features described above, the next uses five further features and rejects 80 per cent of the non-faces. The next three use 20 features and subsequent ones even more. The whole training took several weeks.
Finally, when presented with an actual image, the face detector scans the complete image at multiple scales and with multiple locations per scale. The researchers found that scaling the image by a factor of 1.25 each time produced the most accurate results. They also discovered that they didn't need to test each individual location: they could skip a couple or so pixels each time and still get good results.
The overall result is that the Viola-Jones method produces accurate results very quickly, and certainly fast enough for the limited processing power of the average point-and-shoot camera to cope with.
| |
Read More ...
Review: Stello DA100 Signature DAC
Stello is a brand of April Music, a Korean manufacturer of distinctive audio electronics. The distinctiveness is felt not just in the looks, but also in the features: this is just about the only DAC we can think of with an I2S input. Matched – surprise, surprise – by a Stello CD transport, this is in fact an industry-standard interface, but is normally used within equipment (on circuit boards rather than between boxes). Its great advantage over the usual serial interfaces is that it carries clock signals separately from data, which should, in principle, make jitter less of a problem.
But then jitter has been shown to be a soluble problem, so it's arguably an interesting irrelevance.
Chips with everything
Other features are less remarkable but make up an attractive set, with all four of the usual digital interfaces on offer. Audio outputs come in both unbalanced and balanced flavours, while optional upsampling is offered to 96kHz or 192kHz.
Internal appointments are good too, the digital receiver, upsampling and DAC chips all being recent parts of high-specification, while analogue components are through-hole types and include dedicated balanced line drivers and a Class A mini-power amplifier.
The power supply is generous in capacity and well regulated (and runs surprisingly warm, even in standby) and all things considered, this is quite a lot of kit for the money. It's enclosed in a sturdy case and the front panel controls feel good to the touch (there's no remote available).
Purer sound
So we weren't able to try one of the connection options and only had one transport available with AES/EBU – that still leaves plenty to experiment with.
The good news is that the DA100 is not particularly fussy about which connection system one uses and, indeed, when we fed it from a laptop computer (first direct via USB and then interposing a USB-S/PDIF and jitter-reduction unit) and then compared that with a good-quality CD transport feeding the same datastream, we were hard-pressed to hear a difference.
We felt the CD gave a slightly purer sound, a little sweeter in the treble, but the difference was not the sort of thing one would get hugely excited about.
Little zing
The bad news... well, there doesn't really seem to be any. That said, this isn't the perfect audio component and it does have more than a trace of character, which means it won't be everyone's cup of tea.
It is fundamentally honest, we've no doubts about that, but it seems to add just a little 'zing' of its own to recordings. At first, that seems highly attractive and under any circumstances it's arguably very much preferable to the opposite, but it does mean that recordings that are already pretty lively can become a little manic and, of course, if it's combined with very lively components elsewhere in the chain the net result may seem a bit much.
Well judged
Frequency response is a bit flat, but we noted that distortion from the unit is not quite as low as we tend to expect these days.
It doesn't, however, do anything to disguise the excellent bass, which has a well-judged combination of extension, precision and tunefulness, nor the cleanly extended treble. Detail and imaging are good, though not quite the most astonishing we've ever heard.
Overall, this is an attractive and informative bit of hi-fi . We actually found it particularly successful with classical music, which is usually more restrained in tone and doesn't excite the DA100's temperament, but we would certainly recommend lovers of any kind of music to give a try.
Related Links
| |
Read More ...
Review: BitPerfect Reference 1
The BitPerfect Reference 1 is a comprehensively equipped and complex beast, but it can be set up to be appropriately commodity-like if the user prefers it. Company owner Karl Woodward, who has been involved with computers and pro audio for many years, opines that if his six-year-old daughter can figure out how to record, store and play media from disk, then he knows it will be plain sailing for any adult.
We tested the Reference 1 solely as a music server, but it also makes a very respectable video server having come about because Karl wanted a simple-to-operate, high-performance entertainment hub for his family at home.
If one is building a high-quality, PC-based music server, it does not involve too much additional effort to add video capabilities into the mix.
What's it good for?
The suite of software that comes with the Reference 1 presents users with considerable opportunities.
Do as we did and place it in a broadband-connected room with a music system and it will download music from the internet or let you listen to internet radio. It will also rip CDs, and retrieve metadata and cover art from the internet to allow you to store that music in its library or on any machine or device connected to the same network.
The primary software installed on the Reference 1 for organising and replaying music is Media Monkey. If the Reference 1 is located in a room where there is a television or projector, not only can that be used as the interface for controlling the server but users can rip their DVDs (including Blu-ray titles, where legal, the manufacturer insists) onto the unit – or network attached storage device – and watch them back through the Reference 1's HDMI connection, with the benefit of enhanced sound and picture quality.
It is worth noting that much of the software on the player is 'tweaked', so users need to be careful not to erase it: the installed version of the easy-option, Windows Media Player, for example, can, nonetheless, handle 24-bit/96kHz FLAC files without the additional plug-ins that the standard version needs.
Finally, one could always install MicroSoft Office and send emails while listening to music, provided your collection of rips is sufficiently uninspiring.
BitPerfect build
The BitPerfect Reference 1 is based on a high-specification Asus motherboard with an AMD dual core processor running at 3.0 GHz, along with fast DDR3 (Dual Data Rate) memory. All this makes the machine a speedy and responsive performer, even when running Windows Vista.
The OS is not the plain type you might have on your home PC, but has undergone many modifications and has numerous proprietary plug-ins and codecs (added by BitPerfect) to deliver the sound quality any audiophile listener will demand.
The Reference 1 also benefits from the highly regarded (in professional circles) E-mu 0404 PCI sound card, which offers 24-bit sound at up to 192kHz from its on-board crystal – or an external clock that BitPerfect will happily source and supply for you.
The DAC is an AKM AK4395 and all the external connections to the card – digital and analogue in/ out – are provided through flying breakout leads, which, unusually in audiophile circles, employ quarter-inch jacks alongside the more conventional RCA and Toslink sockets.
As the review system predominantly uses DIN sockets, a phone call to the ever-helpful Chord Company secured suitably terminated analogue interconnects.
The Reference 1 is housed in a modified all-aluminium Silverstone LC04 case, one of the most spacious and stylish HTPC (Home Theatre PC) enclosures on the market. The case integrated effortlessly with the host of Naim components on the system racks, except for its rather bright, blue power indicator.
Given that the BitPerfect is personalised-to-order, the company could remove or adapt the light if you find it intrusive.
Ripping duties fall to a LG drawer-loading Blu-ray/DVD-ROM drive, tested for offset before installation, while a two-terabyte Western Digital Caviar Intellidrive hard disk offers quiet, ecologically friendly storage.
The SATA (Serial Advanced Technology Attachment) cables to these devices are shielded locking types – built to MOD standards – and are superior to the usual low-rent types provided by most manufacturers, which are seldom great performers.
One or two further seemingly small points are worth mentioning, just to demonstrate the lengths to which this company will go to ensure its servers deliver the finest sound – and picture – quality. Its cables are irradiated – that's right, nuked – for enhanced performance. This, they tell us, eliminates any vestige of moisture in the molecular structure of the wire or joints so leaving the copper as pure as can be.
BitPerfect assured us, however, that handling these wires will not cause you to glow in the dark.
Finally, before installation every part of the main PCB is scrupulously hand-cleaned and inspected and the contacts treated with Kontak contact enhancer to promote long-term reliability.
The fabulously detailed sound of the Reference 1 mixes the scrupulous information retrieval of the best of American high-end devices with the energy, verve, and musical fluency of the best of British.
Listening to Duke Robillard's exquisite guitar-playing, it is hard to ignore his magical, subtle endeavours to extract ever richer tonality and expression from every instrument he plays. His delicacy of touch is sublime and strikingly communicative: the replay clearly makes obvious the elation he experiences when he adds new expressions to his instruments' vocabulary.
On another of his discs he works with singer, Susann Forrest on a cover of the Searchers' 1960's hit, Sweets for my Sweet. Here his guitar-playing is again gorgeously detailed with every aspect laid bare – tone, phrasing, leading edge attack, sustain, decay, reverberation and release – as is Susann Forrest's voice, making the performance vibrantly animated and truly joyous.
It's also worth noting the incredible dexterity of the Reference1 with low frequency information. It is outstanding with bass and drums and delivers a highly tangible recreation of the instruments, more often with quite startling dynamic realism.
The player's low end sounds rich and colourful, while being taut and controlled. There is no overhang or monotone boom evident here. Leading edges are crisp and pitch defi nition is flawless. That is just through the analogue outputs: for the best results, you need to hear its S/PDIF connection feeding the Naim DAC, which also has a meaningful, authoritative and accurate low end.
Bob Dylan's Modern Times album, perhaps best demonstrates the Reference 1's combination of vitality, dynamics and detailing with a subtle but thoroughly invigorating and inspiring rendition of Spirit On The Water. Dylan's voice is both explicitly detailed and emotionally persuasive. His band is also exquisitely portrayed and equally communicative and the two elements – voice and band – gel convincingly and effortlessly as one tight, cohesive unit.
The server was certainly not troubled by more complex mixes or dense orchestration. For instance, it made light work of Varèse Amériques, with its uncompromising instrumentation and dynamics.
Organiser
The Reference 1 is a first-class performer and capable of delivering uncompressed rips with true candour, balance and musical coherence.
Helping its portrayal to be even more engaging is the way it scavenges and then organises detail, even among instruments and voices, way back in the mix.
Perhaps, though, its most enticing characteristic is the way it extracts the passion from a performance: be that vocal or instrumental, the performer's emotions come across compellingly.
There is nothing more tedious than a component whose presentation gives the impression that the musicians would rather be down the pub than in the studio.
The enthusiastic Reference 1 completely avoids this charge and shows itself to be comfortably ahead of the competition in this vital respect.
Related Links
| |
Read More ...
In Depth: Upgrading to a Wireless N router: what to look for
Wireless routers are everywhere these days. Sign up with most ISPs and they'll even give you one for free – why would you need anything more? Well, let's see. We'll randomly pick one of the UK's biggest ISPs: Be. It provides an excellent service, but the free router is very basic. It's an 802.11g model, with no support for the newly ratified N standard, so performance will be hobbled. Anything with an 802.11n adaptor could be running at less than half its possible speed, and you'll probably have worse range, too.
We're not having a dig at Be, though – this is more or less the standard across all the free routers out there. ISP-provided models often do things like not allowing you access to important settings, or using a 10/100Mbps Ethernet port even if your computer is using a Gigabit Ethernet adaptor.
Performance is only part of the problem, though. Another is that the routers are using old technology, a particular issue when it comes to security. You should be encrypting your transmissions using the latest WPA2 standard, for instance, but even though it's been around for five years, most free routers support only the easily cracked WEP and outdated WPA standards.
Firewalls, intrusion detection and other security features are likely to be similarly outdated, and sticking with a freebie router means you'll probably miss out on a whole range of possible benefits, including web traffic prioritising, printer sharing, network storage, VPN support and port forwarding.
Choose to upgrade
Relying on free hardware carries a real cost, then, and if you'd like to have a faster, more secure and better-managed network, you'll almost certainly be better off with an upgrade. But be careful – there are plenty of issues to consider first.
Router manufacturers have been working hard to improve the range of their products in recent years, and if you have a large house or like to browse the web in the garden, this sounds like a good move. But there's a downside. If you live in an urban environment, many of your neighbours will also have wireless networks – and the greater their range, the more likely they are to cause interference, slowing you down.
We carried out a few scans in the suburbs of Leicester and found that most locations had at least 10 Wi-Fi networks within range, while several had 20 or more. The highest we recorded was 33. That much competition for bandwidth will guarantee performance problems.
You could use Inssider or a similar free tool to scan your neighbourhood for other networks, and change your network's channel to avoid some of them. But there are only three non-overlapping channels available for 2.4GHz Wi-Fi (1, 6 and 11), so if you have more than four networks within range there may still be clashes.
And that's before you consider interference from other devices that operate at the same frequency: microwave ovens, cordless phones, baby monitors and so on. A better idea may be to abandon the 2.4GHz band altogether and buy a dual-band 802.11n wireless device that operates at 5GHz as well.
Not only does this offer you more bandwidth, with eight non-overlapping channels rather than three, but there's also much less competition for it – so you should see immediate benefits in crowded urban environments.
The downside of the 5GHz band is that its higher frequency doesn't have the same range as 2.4GHz 802.11n devices. It normally has far better coverage than standard 802.11g hardware, though, so if you're moving up from a free 802.11g router then this shouldn't be an issue. And even if it doesn't work out for some reason, then because this is dual-band hardware, you'll be able to switch back to 2.4GHz any time you want.
There's plenty to like about this new technology, but be careful – not every dual-band router is the same. The Linksys WRT320N, for instance, is marketed as dual-band, and it really can work at 2.4GHz or 5GHz. But the important word here is 'or': the router has only one radio that switches from 2.4GHz to 5GHz as required.If you have a mix of technologies – both wireless G and N kit – and they're working simultaneously then this will inevitably slow you down.
A better option is to look for a router that's dual-radio as well as dual-band. The Linksys WRT610N is a perfect example. It includes 2.4GHz and 5GHz radios that can transmit and receive at full speed simultaneously, so there's no time wasted switching bands, and your adaptors always have the router's full attention.
Pick your hardware
The number of radios is one of the most important hardware features of a router, but there are others worth considering. Take a look at the antennas, for instance. If these are inside the case then the router will be a smaller, more compact package.
However, external antennas that use a standard connection (look for a common connector type, such as RP SMA) will be upgradeable, and that could help you squeeze more performance and range out of your setup. Browse a specialist store such as Wi-Fi Antennas to see what's available.
Every router comes with Ethernet ports that you can use to establish a wired connection to your network. In most cases, these are regular 10/100Mbps types, but many routers now include Gigabit Ethernet ports. If your PC supports Gigabit Ethernet then this is a definite plus, especially with high-performance routers (the best 5GHz models can reach over 100Mbps), where a 10/100Mbps connection would quickly become a bottleneck.
Connection to the outside world will be via a conventional phone cable if the router includes a broadband modem. In that case, there's not too much to say: as long as it supports ADSL2+ (all current routers do) then it'll get you online with the minimum of hassle.
If you have an ADSL2+ or cable modem already, though, or maybe want to connect both, then pay attention to extra WAN options. Billion routers are particularly interesting here. The BiPAC 7800N, for instance, features an ADSL2+ port and a separate WAN port to add another cable modem, fibre connection, extra ADSL2+ line or whatever you need.
The new BiPAC 7800GZ comes with a builtin SIM slot: just add the appropriate card and it'll directly support 3G connections too (which is much easier than trying to cope with USB modems and their many compatibility issues).
An increasing number of routers now include USB ports along with the usual LAN and WAN ports. Typically, these will be used as an easy way of adding network storage: plug in a USB drive and it'll be freely available to everyone on the network. Well, sort of.
One complication is how the router presents the drive to the rest of your network. What you ideally want is support for Windows Server Message Block (SMB), which provides a standard way for apps to locate, read and write files to a network resource. As long as this is included then you should be able to browse the network in Explorer, 'see' the router's USB drive, map this to a network drive on your PC and use it as normal.
Some routers offer rather less convenient methods. For example, the Asus RT-N13U forces you to access the drive through a built-in FTP or HTTP server, which probably won't be ideal for most people.
Don't assume that seeing 'FTP' in the spec means this will always be the case, though. The Linksys WRT610N, for instance, supports external hard drives as well as USB keys, and these can be used as FTP servers, media servers, SMB-based external drivers, or all three at the same time. Handy!
Not to be outdone, other manufacturers have added even more USB functionality. Some DrayTek routers, such as the Vigor 2710, provide a USB port that allows printers to be connected and shared across the network.
And D-Link's SharePort technology lets you plug just about any USB device into a router such as the DIR-685, then access it on any PC on your network as though it was connected locally. This requires extra software to be installed on your network PCs, and only one can use the device at a time, but it could still be useful.
Before you opt for a solution such as this, it's worth thinking about how it might affect overall performance. If it's a large network, and the router is already likely to be fully occupied most of the time, then do you really want it handling additional file transfers and print jobs?
We've also noticed that USB devices connected to a router are noticeably slower than when plugged into a PC – that's no real surprise when you think of all the work the router's processor is doing. So, while USB storage is useful for basic things (as long as you make sure it supports SMB), if you have heavy-duty requirements then a dedicated network storage device will still offer the best speeds.
Once you've decided on the basic hardware you need, it's time to consider the software-based features – and none are as important as security.
So it's fortunate that just about every 802.11n router comes with a good set of security features: firewall, intrusion detection, the latest WPA2 encryption, and WPA and WEP if you need them for older devices. Avoid the latter two if you can: WEP is easily cracked and of no real value, and using WPA may drastically cut performance.
While every router comes with some form of firewall, the implementation varies greatly. Some are very simple, but DrayTek routers in particular are as configurable as their software-based cousins.
The DrayTek Vigor 2910G, for instance, has built-in support for around 30 instant-messaging apps, eight P2P protocols and 16 web-streaming applications.It's easy to add more, and you're able to create custom filter sets to control what traffic is allowed and what will be blocked.
The most interesting security features are often the more unusual extras. One of our favourites is the ability to set up a guest zone (also known as 'guest access'). This allows the router to create a separate wireless network that provides access to your internet connection, but keeps users isolated from more valuable resources, such as your PCs. That's perfect if, say, you don't mind your kids' friends using the network to get their iPods online, but you'd rather they didn't poke around any further.
If you need to connect to the company network from home, Virtual Private Network (VPN) support will probably be useful. At a minimum, your router should offer VPN passthrough for IPSec, PPTP and L2TP: this means it'll recognise VPN traffic and allow it to pass through your firewall and other protective layers.
Some routers only support passthrough for a single session – a hassle if you need more than one connection. Ideally, you'll want support for multiple concurrent sessions, and this doesn't have to be expensive: D-Link's DIR-655, for instance, includes it for around £65.
If that's not sufficient, look for a router with VPN endpoint capability, which means it's able to initiate a VPN connection itself. A few routers, such as the Buffalo Nfiniti Wireless-N, come with a complete integrated VPN server. If the router supports remote access, and that's a feature you'll use, consider how it's protected.
Some D-Link routers now use a CAPTCHA-style 'type the letters you see in the box' graphic, making it much more difficult for hackers to use brute-force attacks on your system. Others may let you limit access to particular IP addresses, or customise the port number used.
A feature to investigate carefully is any talk of 'content filtering' or 'parental controls'. Sometimes these turn out to require a subscription. Occasionally they're competent free packages, such as NetGear's Live Parental Controls, but in many cases, they're near-useless keyword-based efforts where you manually enter a few suspect words and the router blocks any webpage that uses them.
If you need decent parental controls then download the router manual before you buy it, and take the time to find out what's included.
Controlling the priority of certain types of network traffic is an excellent, often overlooked feature of modern routers. You don't want your video stream to have major glitches because someone else in the household is downloading the biggest collection of MP3 files in P2P history, right? Get Quality of Service (QoS) configured correctly and the network may be able to take care of this on its own.
A lot of 802.11-certified equipment comes with some basic QoS functionality in the form of Wi-Fi Multimedia (WMM). If your router and network adaptors are WMM-certified, they have the feature turned on and your app supports WMM, you're in business.
The program adds QoS details to its packets that say, 'I'm important, me first', so the router prioritises them ahead of other traffic. There are problems with this scheme, though. Not enough apps support WMM to make it really useful, and while its automatic nature cuts network management hassles, it doesn't give you enough control.
For that you need QoS features in the router, so it's worth checking to see what's on offer. Billion routers such as the BiPAC 7800N support DSCP (Differentiated Services Code Point) marking, a scheme where sources tag their network packets with information about what they are. If your apps do this then you'll be able to do things like assign a guaranteed level of traffic to media streaming, and a maximum level to P2P apps.
You can apply these rules to particular ports, internal or external IP address ranges or even a set timeframe, maybe restricting P2P to 10 per cent traffic during the day but allowing it 90 per cent between 1am and 8am.
Linksys routers are generally very configurable, too. For instance, the WRT610N can prioritise traffic by MAC address as well as application or Ethernet port, so you can ensure that a particular network device always gets the same level of priority, whatever it's doing.
Alternatively, if tweaking all that sounds too much like hard work, you might prefer D-Link's solution. Many of its routers include a simple technology ('Wireless Intelligent Stream Handling') that will look for media streams and then automatically prioritise them.
Which QoS solution is best for you will depend on your setup and circumstances, but do take the time to explore what a particular router offers before you buy. Good QoS settings can make a great deal of difference to a busy network.
Speeding into action
Upgrading to an 802.11n router isn't a magic bullet. It's unlikely to have a huge impact on your wireless range (although it should improve performance, even at the outer edges of your network). It won't make 802.11g devices any faster, either. Also note that to get the most out of it you'll need 802.11n clients elsewhere (however, with prices starting at under £10, this needn't be expensive).
Still, once you're properly set up, there are all kinds of benefits on offer. Significantly better speeds; improved WPA2 security; the ability to avoid interference if you've got a 5GHz model; built-in automatic prioritisation of the most important web traffic; bonus extras, such as easy network storage or printer sharing; and a host of other features that come automatically with up-to-date firmware.
So, which router should you choose? Our preference would be a dual-band, dual-radio model, especially if you live in close proximity to other networks: escaping their interference will save you hassle.
The dual-band 802.11n router market is fairly immature, though, and not every manufacturer has yet got the new tech performing at its best.
If interference isn't a problem with your G network, don't feel you have to do dual-band right now. A single-band N router will deliver plenty of benefits at a low price, and you can take advantage of the greater range of dual-band products that will be on offer when you do decide to upgrade.
| |
Read More ...
Review: Funk FXR tonearm
Arthur Koubesserian is one of those audio engineers who genuinely thinks outside of the box. He has been doing so since he set up Pink Triangle in the late seventies and his Funk Firm continues the theme with its first tonearm. The Funk FXR takes a radical approach to the problem of resonance by placing a carbon fibre cross section in the middle of a thin, walled aluminium tube.
An approach which he claims makes an incredibly stiff, yet light-weight tube that's far superior to the beams found on other tonearms. But isn't that another Rega bearing, we hear you mutter.
Beam me up
Yes, at least that's what it is on the outside. On the inside Arthur has replaced the lateral bearings with ABEC 7 ball races from Switzerland.
But it's the armbeam that's the most significant – that CF cross-section runs the length of the tube and terminates in a Clearaudio-style slot to which you bolt an aluminium block with fixings for the cartridge and supplied SME fingerlift.
Alignment is a matter of establishing the correct overhang, which makes set up relatively straightforward.
The FXR is available in two forms. You can buy a complete arm which is based on a Rega RB301 and has a stainless steel shaft rather than the three-point fixing usually found. This version comes with a collar to allow VTA adjustment and costs £1,175.
We have the alternative form, which is a modified RB300/RB250, where the customer supplies an arm and Funk installs the new bearing and FXR arm for £900. This version retains the threaded base of the original and thus is less straightforward to adjust for height.
However, aftermarket collar systems are available, or you can use spacers.
Perceptions
The FXR is a well-executed modification and the parts that Funk adds are machined and finished to a high standard. The tube is anodised in red and the 'headshell' metalwork is anodised black.
The way that the SME finger-lift sits between the mounting block and the cartridge can make installation rather tricky, but once it's there it's a very nice lift to use. If you have two cartridges each can be bolted to a mounting block for easy changeover.
Perceived value is undermined a little by the Rega bearing section and while we've seen this on more expensive arms (Townshend's Excalibur, for instance), smarter implementations such as that by Audio Mods can be had for less.
Cosworth-syle upgrade
Funk's argument is that you are getting a Cosworth-style upgrade for your money and the sound quality does bare this out, most tellingly, when you compare the FXR with a standard RB300. The latter sounds overblown and loud with little image focus and quite severe smearing of detail by comparison.
But then again, the RB300 only cost £210 in its day, so you'd hope the revision improves matters, but what about something serious like an SME V?
The FXR also makes this stalwart sound colourful – the V adds a distinct bloom and muscularity to the sound, which once again smears detail, albeit in rather more subtle fashion. This was a shock.
The Series V is a superb piece of engineering and a highly regarded piece of kit, yet it was clear that the FXR delivered more of the music in a calmer and cleaner fashion. The Cosworth analogy would seem, therefore, to be surprisingly accurate.
Related Links
| |
Read More ...
Guide: How to diagnose faulty RAM
Memory is quick and easy to change, and because the result of buying new RAM is often a performance gain, it's one of the most common upgrades carried out on a PC. But things don't always go smoothly, and faulty RAM causes its fair share of heartache. Repeated 'Stop' errors, freezes and other crashes may well be caused by duff memory.
Even if you've never upgraded your RAM, if your PC has difficulties like these, memory may still be the problem. Substituting it with some new RAM may help, but if your memory isn't actually the culprit, the expense may not be worth it.
To make sure that the RAM is at fault, you need to run some tests. The most efficient is MemTest86 – grab it from www.memtest.org.
Go to the Memtest site, choose 'Free Download' and opt for the latest version in ISO image form. This will be downloaded as a ZIP archive, so browse to it and double-
click the file. Choose 'Extract All Files in Windows Explorer', then follow the wizard to extract the ISO file.
Use a free CD-burning program like CD Burner XP to burn the ISO image to CD. Launch the program and go to 
'File' > 'Write Disc from ISO File' before finally browsing to the ISO image and clicking 'Write Disc'.
See the walkthrough below for details on how to run the program.
Memtest86 is very thorough, and because it runs outside Windows you can be sure that you're checking RAM addresses that might normally be occupied by Windows components (and thus are likely to produce crashes).
Running continuous testing for a couple of hours should ensure that you've checked each address multiple times. The longer you run the test, the more accurate your results will be.
If you get errors reported and you have multiple RAM modules, remove all but one and run Memtest86 again. Swap your other modules in and retest to confirm which is causing the issue. This will save you from replacing more memory than necessary.
Once you've identified the faulty module(s), you can start looking for replacements. Your motherboard manual or PC documentation should tell you the specification of the RAM you need, but useful paperwork isn't always to hand.
Memory specialist Crucial provides a useful online tool to identify the type of RAM you need. You can either use its scanner tool (if you use Internet Explorer and you don't mind having your system probed or the possible privacy implications that this might have) or you can use Crucial's own Memory Advisor product.
This is a simple database in which you select the manufacturer, product line and model number of your off-the-
peg PC, laptop or motherboard. Once you've found out the specification for the memory modules needed, you can shop with Crucial if you like, or you can browse Kelkoo and others to find the best price.
Thoroughly test existing memory
1. Boot from CD
Once your Memtest86 ISO file has been burned to CD, restart your computer so that it boots from this disc. You may need to enable CD booting in the BIOS or select the Boot menu at startup for this to happen. You should boot directly into Memtest86 and start testing.
2. Run automatic tests
Memtest86 automatically runs a number oftests on your RAM. If all is well, the memory should pass all of these tests. You can run continuous tests to detect intermittent problems. If errors are reported, you may need to replace one or more of your memory modules.
3. Configure specific tests
You can configure the program by pressing [C] while it's up and running. This opens a menu that gives you the full choice of tests to run and enables you to set a memory address range to test. That option may be useful if you're getting intermittent problems.
| |
Read More ...
No comments:
Post a Comment