istartedsomething

Cue Scoble tears. This is one hell of a technology from Microsoft Research that I think will have a huge impact on how amateurs and hobbyists write music. The technology and software prototype is called MySong.

In a nutshell, the software records your singing (preferably in tune) through a microphone, and it systematically generates an instrumental accompaniment for your song. The quality is even comparable with a professional accompanist, not to mention the cost and time involved.

If you don’t believe me, have a look at this quick 5-minute demo with loads of practical examples.

The technology was developed by Ian Simon from University of Washington in collaboration with Dan Morris and Summit Basu from Microsoft Research. They’ve published all of their research and methods in a paper (PDF) of the same name to present at the CHI Conference 2008 in a couple of months in Italy.

Without getting into too much detail, the technology uses the Hidden Markov statistical model which has been ‘trained’ by preprocessing a database of nearly 300 musical lead sheets. The sheets come from a variety of sources, genres and popularity to give a broad set of melody and chord sequence combinations. When it comes to the end user, it uses this abundance of data to find chord sequences which work best for each segment of melody. Presumably that means more ‘training’ can be done to improve the output even more.

The paper also points out “there is not a single correct accompaniment for a particular melody; chord selection will vary among musicians and genres, and a single musician may recognize many appropriate chord sequences for a single melody.” As a result, the software also gives the end user some room for creative adjustments in the form of “jazz factor” and “happy factor”. Both of which have a reasonable effect on the resulting accompaniment as demonstrated in the video.

If that wasn’t enough already, the researchers sought reviews from 30 independent musicians asking them to rate the different accompaniments produced by three different systems from one vocal input. They compared the result of MySong with hand-crafted chords and as well as Band-In-A-Box, the “state-of-the-art” commercially available software which they recognize to be the only automatic accompaniment generation system.

Here is just one example of the twelve they’ve ranked. I’ll leave you to judge for yourself.

Like many cool technologies at Microsoft, this remains still a research project. Whilst there’s no written indication of a commercial product as a result of the technology, but I don’t see why they wouldn’t or even just release the software as is.

As for future work, they say “additional development will focus on improving and diversifying the audio generated by MySong; the system is already able to supply chords interactively to a pattern based arrangement tool, which results in compelling audio output. Several study participants indicated that MySong would be of significant value for learning music theory; we are thus excited about exploring educational applications of this technology.”

To my surprise, the TED conference organizers has swiftly uploaded the presentation recording from Microsoft’s announcement of their WorldWide Telescope technology showcased no more than 12 hours ago, when in contrast, videos are usually only made available online after a year or two of exclusivity. Remembering, attendees pay several thousands of dollars to go to this conference – one of, if not the best conference in the world, so to make an exception for this particular presentation is really something.

Ah, product feedback surveys, what will we ever do without you? You share your most intimate secrets and future product ideas with complete strangers all in the name of marketing research. Thankfully Microsoft surveys are no exception.

An anonymous tipster pointed out to me in a recent (private) Windows feedback survey sent out by Microsoft, asked how often users use the Start menu search functionality, was a screenshot highlighting the aforementioned feature but not of Windows Vista and certainly not anything we’ve ever seen before.

The most logical explanation would be of course, this is a mockup of Windows 7.

If that is indeed the case, which is impossible to prove I might add, then we’re witnessing a number of changes from the Start menu in Vista today and even the Start menu from the leaked build of Windows 7 Milestone 1.

Notably the Start pearl/button/orb rests in a curved corner that is darker to the rest of the taskbar which is now considerably more transparent and white. And quite possibly the taskbar is “double height” by default. Some applications in the start menu also have a cascade button on its right, leading me to believe its some sort of extended menu, for what purpose I don’t know.

There are also a few things in this screenshot which stand for the wrong reasons. Some of the applications have icons which doesn’t match. For example, Outlook has the Windows Mail icon and PowerPoint has a XAML file icon. Then on the right side, the link buttons which are expected to be “Pictures, Music, Games etc” have been replaced by Lorem Ipsum filler text.

In case you were wondering, the screenshot has since been removed from the web server where the survey was hosted. It has been replaced with a more familiar screenshot from Windows Vista. Obviously, too late.

It’s not very often you find a job advertisement which doesn’t tantalize your curiosity, so it’s only to fair to take the following with a grain of salt. This Microsoft job posting for a “user experience designer” was published on the 23rd of February 2008.

The focus of the job description refers to a startup project inside Microsoft called Tangram. What they aim to achieve with the technology is intriguing at the least.

The Tangram team provides an opportunity to work on the engineering challenges of an Internet scale services technology coupled with 3D interactive experiences that have finally become possible due to ultrahigh resolution screens and high performance processors. We are changing what people expect from the Internet by delivering entirely new ways to see, interact with and use data of deep interest to consumers and business people. This is a project with high visibility in Microsoft. We are assembling a team of world class engineers with very strong analytical and technical skills. We are looking for people who can define the problem, come up with a solution, and implement it. End to end. Ship it in 6 months.

Tangram is a startup product team with the highest levels of visibility.

We aim to ship our alpha service and tool this year, and ship a revenue producing V1 in 2009, within a major Microsoft go to market. Tangram has the opportunity to change what people expect from the Internet by delivering entirely new ways to see, interact with and use data of deep interest to consumers and business people.

We are looking for a Senior user experience designer who has a burning passion for quality and simplicity. This is an opportunity to exploit ultrahigh resolution displays, interactive 3D, photorealism and immersion, and reset what people expect from the Internet.

By no means a coincidence, a “Tangram” is actually an ancient Chinese puzzle game composed of 7 pieces of shaped objects called “tans” – including several triangles, a square and a parallelogram. The objective of the puzzle is to arrange the pieces, without overlap, to form an outline of familiar objects. It might sound limited, but the possibilities are in fact endless with the right creative mind.

To add more fuel to the fire, during the Microsoft “Mix n Mash” interview in December 2007, Bill Gates said the following:

Now, certain structured things are still hard to learn, like go to the Web and try and understand foreign aid. Because that’s kind of a numeric thing, the classification of (inaudible) is hard, the Web does not do that well today.

Now, some software breakthroughs in terms of letting you visualize data in rich ways and normalizing it as it comes from different sources, over the next three of four years that will happen.

In fact, most of my day today was in a visualization group where we’re talking about what breakthroughs are needed so that if you want to study a question like that, take economic flows, you want to understand this sub prime thing, you want to see the numbers, okay, what percentage of people bought what type of mortgage over what period of time, when does the interest rate go up, what does this mean, what portion of their income is it. If you really want to understand it more than just trivial text-type articles, there’s nothing out there today that lets you do that.

Coincidence? I think not. All we can do now is to keep an eye out for that alpha hopefully coming sometime this year. Otherwise I’m sure Scoble will let us know in advance, with a few tears for extra impact.

One of the two big Microsoft news out of the Game Developers Conference today of course is the extension of XNA Game Studio to the Zune. It means now with one development framework and toolset, game developers can make the relatively simple transition between three platforms to publish their games on – the PC, XBOX 360 and the Zune.

You might have seen pictures on Engadget and Joystiq, but nothing beats full-motion video. Dan Fernandez has uploaded a short snippet of the Microsoft keynote with Chris Satchell showing off a game demo on the Zune. Debunking some of the fears it wouldn’t play well, the controls looks pretty intuitive and the graphic are smooth as well.

In the small spectrum of software developers who writes compatible software with 64-bit versions of Windows, you’d expect Microsoft to be the last to drop the ball. Today, Microsoft’s Windows Live Messenger team did exactly just that and dropped compatibility for 64-bit versions of Windows XP and Windows Server 2003.

A reasonable person might be led to believe Windows Live Messenger 8.5 is somehow incompatible with those versions of Windows. Unfortunately, reason seems to be out of the equation.

As part of the automatic rollout of the final version of Messenger 8.5, Jura of the Windows Live Messenger team announced on the blog today,

A special note for those of you running Messenger 8.5 Beta on Windows XP 64-bit or Windows 2003 Server: the final version of Messenger 8.5 will not install or run on your OS. We don’t want you to get stuck out in the cold so you will not receive the mandatory upgrade to the final version. However, these operating systems won’t be supported by future Messenger versions.

It’s almost as if someone held a gun to their head to announce this because as a matter of fact, the final version of Messenger 8.5 does infact install and run on 64-bit XP and Server 2003 as proven by Rafael Rivera and Jonanthan Kay.

To be perfectly clear, the only hurdle prohibiting end-users from doing so is the custom installer application. All you need to do is avoid the custom installer and install the MSI file directly. No ‘hacking’ required.

I really don’t understand why they would lie about the software not installing or running on 64-bit XP and Server 2003. This is not even a matter of support. They could have outrightly said “we do not support these 64-bit operating systems”, but to say “it doesn’t work” is absolutely untrue.

They could have as well just said,

Thanks for beta-testing WLM 8.5 which installed and ran just fine on your 64-bit XP systems. Your efforts to improve the product, for the rest of us, is much appreciated.

Now it’s time to part ways – it’s not me, it’s you – and you should probably downgrade to the older and less featured but “stable” version of WLM. Trust us, the software doesn’t really work on your system. The cake 64-bit support was a lie.

Windows Live Messenger 8.5 remains compatible and supported on 64-bit versions of Windows Vista.