I was fortunate to be the first to get a full live demo on this sound recognition technology at the 2009 World Copyright Summit in Washington, DC, thanks to Hanna Pantle (BMI’s Asst. VP, Corporate & Media Relations) and Landmark Digital’s VP, David DeBusk, both wonderful folks – you should get to know them. Landmark Digital, owned by BMI, has put forth the BlueArrow digital audio recognition technology that came as a result of BMI acquiring Britain’s Shazam Entertainment Ltd. BMI’s acquisition included patents covering the BlueArrow technology, and Landmark is using this as a springboard to develop more audio recognition products and services. BlueArrow is currently in Beta testing, and from what I saw, is based on sound principles (pun intended) and something that could have long-range implications on our music delivery business models (more on that later, but first some info on how it all works)…
This particular type of real-time audio recognition technology is not via watermarking, yet can ‘hear’ beyond the sound effects and dialogue that obscures music identification in traditional fingerprint technology (the type used in MediaGuide—which only monitors radio anyway). Since music in TV, film and advertising is rarely used without additional audio elements, traditional fingerprinting technology is not really effective or useful to the score composer (or even for songs used in anything but otherwise silent film/TV montages). Within two seconds into the demo, I saw identifying info pop up on the screen, including an instance where a sample appeared to be used within the title being monitored—quite useful for the major labels and big publishers, considering the preponderance of ‘remixes’ using samples of classic recordings. I’m told the system identifies sound files with about 99% accuracy, and it certainly was quick to bring up instances where the music was being used, whether it was on radio, TV or in a commercial. The amount of music and the global genres covered was astounding—this is not just your basic rock, pop, jazz, country and classical coverage. It is evident that this has not been a simple feat, either—putting all the metadata into the system and correlating it all with the sound files has obviously been a monumental task. It’s an ongoing process, of course, and new music is being input all the time, expanding the reach and scope of what is being monitored and where.
It seems that they have radio covered on all fronts, including advertising, but still have a way to go in getting all the major TV, Cable & Satellite markets completely wired in (although what they already have set up is still better than the nothing we currently have elsewhere). They currently are monitoring (based on Neilson ratings) the top 25 cable and satellite networks, plus ABC, CBS, NBC & Fox in Los Angeles and Nashville. They are working on expanding that lineup, and are also moving forward with agreements to monitor discos & clubs worldwide. In addition, and despite the inherent problems and politics involved in monitoring non-U.S. movie theatres, concert halls, arenas, etc., and dealing with the numerous PROs and location owners, they are in the process of securing cooperation in monitoring music uses in all types of live venues around the world. Anyone who’s negotiated a contract knows how much time and effort and compromise it often takes to get everyone to sign on the dotted line, and when dealing with such a myriad of owners and interested parties, it can be quite a painstaking process. I don’t think it will take Yoda-like patience to see this all come to fruition, though. It appears Landmark is making good strides and is committed to moving things along as quickly as possible.
Blue Arrow provides data analysis and reports based on the clients’ specific needs and requests, and the system is sensitive to emerging trends. That’s good stuff. The system also requires that all the metadata and sound files be uploaded into their system by their folks (unlike tech that puts the process into the music owners’ hands)—which, depending on what side of the fence you’re on, could be considered a good thing or a bad thing. While great for major labels (and others who can afford to pay for an account and have the BlueArrow folks put their music into the system, then monitor and report back to them with the data), it currently seems to leave out the indie and boutique label artists and that solitary creature, the score composer who is not John Williams or Hans Zimmer.
From all indications, though, they seem to be truly looking at a way to make things work for the rest of us—those not independently wealthy and/or under the corporate umbrella. I think that with a little nudging on our part, we might be able to help move that along—at least with the folks over at BMI.
How so? Since BMI owns Landmark, it would be kinda neat (not to mention a PR boost) if they could figure out a way to provide the BlueArrow tech to all its BMI writers and publishers—if not as a freebie, at least for a nominal, yearly “value add” fee. It might be well worthwhile for BMI composers to take note and take action, writing/calling your BMI rep to ask about BlueArrow and how it can be made to work for the “little guys” with BMI. If enough start clamoring, perhaps it’s something that could come into the realm of possibility sooner rather than later. If ya don’t ask, ya don’t get, so start writing, calling and emailing. Who knows? Those registered with other PROs might also be able to benefit in the long run if it can first become an economically feasible option for BMI’s non A-list artists and composers.
So back to the tech…
They take a digital sample from an audio source (TV, radio, cell phone, Internet, cable broadcasts, satellite, HDTV, etc.). Regardless of whether it’s only a tiny bit from the middle of a song or virtually obscured by dialogue or sfx, the BlueArrow system knows what it is and has all the details. It monitors broadcasts 24/7/365 and compares the fingerprints against the metadata in their system.
Now here’s where it gets interesting and complicates the whole “re-titling” practice (for those who have been under a rock for the last 30 years, re-titling is where one composition by a composer is embodied in one sound file/master recording of the work and then given different titles by every library in which it resides. The ‘new’ (or alternate) title is registered with the library’s applicable performance rights organization [PRO] that corresponds with the composer’s PRO, but only after the library has also put its own publishing entity as owner or co-owner of the ‘title’—thereby allowing the library to participate [in whole or in part] in the ‘back-end’ publishing portion of performance royalties paid for every TV/cable airing and foreign theatrical performance of the program in which the composition/master’s use was reported to the PRO. This practice means there can be many ‘titles’ attached to one composition/one recording, none of them bearing the original title that was used when the composer originally registered the actual work with the U.S. Copyright Office)…
So… Every fingerprint match delivers a title with time, date and source of the audio. Since the system is matching the AUDIO source file, regardless of what TITLE is associated with it, that/those additional title(s) will come up as a match. Puts a whole new spin on the old library argument of, “But when we re-title your work, we’re doing you a favour, since you’ll know where your income is coming from based on the title.”
No more. The BlueArrow system is designed to trigger recognition by SOUND, so every title associated with that sound file is going to come up (it’s how sampled pieces within a sound file will come up in their monitoring system). That means a composer would have to check with every library in which he/she has that track to find out which one licensed the sound file and to whom. If the use being recognized, however, is NOT from any valid, licensed situation, but is from an unlicensed use, there are potentially dozens of sources from which that unlicensed use may have generated. With only one title with each library (unique works and unique recordings), a composer would know exactly which library to contact about any unlicensed use of the work.
So, contrary to the long-standing library mantra of “it makes things easier to track your income,” sound recognition technology actually complicates things, making it more time-consuming and HARDER to track the uses of any re-titled track. If a composer has retitled tracks out there with multiple libraries, this tech will now hamper the ability to monitor where music is being used and by whom. Each sound file will be popping up with multiple titles, multiple publishers and multiple master owners. The libraries will ultimately be falling all over each other trying to figure out whose title got what hit in the monitoring.
When the fingerprint match delivers the titles identifying the date, time and source of the audio, that data is analyzed and delivered to the clients. Isn’t it going to be interesting when that same data comes up for more than one client? Who is going to be able to claim the use, since both their titles come up as valid against the sound file? BlueArrow doesn’t know how this re-titling issue will ultimately be handled. Since the head of the music department at a major network with whom I recently met had no idea about re-titling until I informed her about the practice and how it works, it’s not surprising that Landmark’s VP of Business Development was also unaware of the practice. He said that frankly, he’s not sure what the implications will be. They’re currently Beta testing a handful of libraries (they/I am not at liberty to say which ones), but I can say with relative certainty that the libraries were not forthcoming about their practice of retitling tracks. If any of the libraries currently testing BlueArrow have the same track from the same composer with different titles (not an unlikely scenario, given the rampant retitling that goes on), it could prove to be a little sticky, and somebody’s gonna have some ‘splainin’ to do…
This complex pattern recognition matches the sound against millions of other patterns stored in the system and doesn’t CARE what title anyone has given it. It’s going on sound only and it may end up as a 52-card-pickup scenario when trying to determine just WHOSE placement/use is being picked up from that movie airing on HBO 2 in L.A. on Tuesday, July 28th 2009 at 3am Pacific time.
So BlueArrow does quickly and accurately monitor performances of a work, but what about unidentified music in a film or TV program that airs that may or may not be supported by cue sheets? How can the client find out about those and get paid by the PROs for such unlicensed and/or unreported and/or under-reported uses? It simply hasn’t been cost-effective to spend the man-hours needed to do real-time searches through boxes of DVDs of programs that have aired (films, TV shows & ads) and match them against shelves of CDs (music in a library or at a label), only to find one unlicensed or unreported composition use in one program. The BlueArrow tech can identify sound files fed into their system from such DVDs and CDs, and they are working on identifying previously unidentified material for their clients. They’re continuing to develop and adapt the tech so that (hopefully) fewer compositions and recordings fall through the cracks. Data is everything to today’s digital world, and with under-reported and unlicensed music use percentages varying from 15% to 70% (depending on which source one is to believe), the more info one has to support the use of music from one’s catalogue, the more money one can potentially collect—both from those who may have used a piece without permission or incorrectly reported a use, and/or from one’s PRO.
So here’s my editorial of the day… and I may digress here:
The fact that BlueArrow technology recognizes a sound file and then may end up matching it to more than one title in the system is not a failing of the technology, but rather, a failure in our business models that have allowed this as a likely scenario. We, as an industry, have allowed this to perpetuate, many citing that it’s just “good business.” Some of us have spoken agin’ it, others fear it.
The way I see re-titling in today’s marketplace? It makes NO good business sense. At the very least, it has always stood on shaky ethical grounds, and now it can ultimately be seen just as an antiquated method to monitor music in our technologically sophisticated market. It’s no longer the future or the present, but should be relegated to the PAST. We often hear the word ‘dinosaur’ bandied about when speaking of old technology and business models, and that’s where we should relegate the whole concept and practice of re-titling music.
Any unique track of a unique composition shouldn’t be out there with ten titles, and with ten different publishing entities noted as having a piece of the back-end pie. When the General Counsel of the U.S. Copyright Office says that at first glance the practice seems “unethical” at best, then we need to rethink how we’re doing business.
Music libraries don’t re-register those ‘new’ titles with the U.S. Copyright office and reference the original copyright filing number. I know. I checked. They can’t, because to do so would be official FRAUD, unless it’s a transfer of ownership (all or part) and they are actually acquiring a portion of the bundle of rights to the copyright itself. That’s not what they’re doing. Registrations with PROs are merely on titles, NOT actual copyrighted works, so the libraries file registrations of these new titles solely with the PROs, and not with the Copyright Office. This allows them to perpetuate their practice of re-titling without committing “official” Copyright fraud. They say it’s for bookkeeping purposes, but I’ve never met a reputable bookkeeper who also decided it was fair to skim off 50% of my in-perpetuity, back-end income from their efforts on my behalf. No; that’s why I pay them for their work—which is what sharing in the up-front licensing fees is supposed to be about.
Sound recognition technology and its continued and expanded use should put an end to all this bookkeeping banter and bring the focus back onto what is actually OWNED here… and it’s what we’ve always known and has always been broken down in music licenses: the composition and the recording of those compositions.
So back to the original subject… My thoughts on BlueArrow are twofold:
(1) Sound recognition technology that ‘hears’ past dialogue, sfx and noise, thereby allowing for the accurate identification of music uses, monitoring and payment for those music uses is a GOOD thing, and Landmark should be commended for it; AND
(2) When such tech is available to ALL music creators (not just big labels, publishers & libraries) in a cost-effective method for monitoring the use of their works… THAT is when I’ll be pulling out my pom-poms and yelling, “YEAH, TEAM!”
And a reminder (from my attorney): All statements above are my opinions and not intended as legal advice or counsel. No warranty or representation is made as to the accuracy of these statements. You should hire an attorney before entering into any agreement or contract. So there!
© 2009 by Gael MacGregor