Four takeaways from the Association of Independent Music Publishers' Fall 2013 panel discussion on "Digital Audio Recognition: Disruption and New Revenue Opportunities:"
- Metadata Tells Licensees Who to Pay - From copyright registration to cue sheets, quality metadata - the information about your assets - is the key to collecting cash.
For example, when multiple music publishers submit rights information to music services and such information does not agree (e.g., it adds up to over 100% song ownership), earnings are locked until the parties resolve the discrepancies and tell the music service who to pay. As another example, when a music service only knows who controls half of a song, the earnings for the unknown share(s) accrue in an escrow account until claimed.
Steven Corn, CEO & Co-Founder, BFM Digital noted during the panel discussion that since different music services require different sets of metadata, it is important to invest in high quality metadata and not to relegate metadata tasks to low skilled workers. - Choose Your Tools - Not all digital audio recognition technology is the same. For instance, Chris Woods, co-Founder and COO at TuneSat, explained during the panel that Shazam was designed for mobile users to recognize songs in bars, while Tunesat was created to identify production music in TV broadcasts. YouTube alone uses two types of digital audio recognition: Content ID, which identifies audio recordings (and songs that have been linked thereto by metadata), and Melody Match, which recognizes songs, regardless of the recording.
The basic categories of digital audio recognition technology include (a) watermarking technology, which is "active" because it requires content owners to encode unique IDs into masters in order to detect uses of such masters, and (b) fingerprinting technology, which is "passive" because it works by identifying the unique characteristics of a file, even if it is an old recording that was released before watermarking existed.
Bobby O'Reilly, CEO at ProTunes, pointed out during the panel that fingerprint technology can be used in conjunction with watermarking technology. For example, watermarking can be used to identify to which non-exclusive licensee a performance of a particular work should be attributed, while fingerprinting technology can be used to identify uses.
- Have a Strategy - Digital rights management in the era of the DMCA Safe Harbor provision is a lot of work for content owners. Jeff Price CEO and Founder of Audiam recommended during the panel discussion that you send all music services copies of your metadata so that the services will know about the copyrights and so that infringements may thusly be deemed to be willful. For his client Jason Mraz, Price developed a more specific technique to "scrape" the IDs of videos returned by a YouTube.com search of "Jason Mraz live" in order to find IDs for which to search in YouTube's rights management portal, which doesn't provide as many search results as YouTube.com.
Price 's tactics aside, nearly all monetized musical compositions on YouTube are identified based on metadata that links a song to a master, so if one links a song to a master that has been used thousands of times, one instantly matches such song to thousands of uses. Such power accorded to masters is why rerecords appeal to music publishers who do not wish to share revenue or content monetization decisions with a record company. On the other hand, YouTube does not allow record companies to monetize user-generated content without publisher approval, so publishers have the power to disincentivize recording rights holders from exploitation that a publisher disfavors. Further, other music rights holders, such as Activision, elect not to monetize YouTube exploitation of its music, in order to encourage unfettered use by users, which it deems promotional (i.e., which indirectly stimulates game sales) - There is No Substitute for the Human Touch (Yet) - DeDe Burns, Sr. Director of Strategic Services at ASCAP, recommends utilizing digital audio recognition technology on a supplementary, not stand-alone, basis.
To understand why this is so, consider that while technology may tell a performance rights organization (PRO) which repertoire was broadcast, the PRO must rely on cue sheets or other input from a human in order to ascertain (a) whether the use was, say, a featured, theme or background use and, (b) in the case of "retitled" works licensed on a non-exclusive basis, which publisher or production library to pay.
Also, Hunter Williams, Executive Director, Production Music Association raised the issue that in order to monetize YouTube uses that are less than 30 seconds, or copyrights embodied in film trailers (and other uses where dialogue, compression and post production - or even satellite broadcast - causes so-called "dirty audio," which defeats current digital audio recognition technology), manual matching is essential.
AIMP members can view or listen to the organization's "Digital Audio Recognition: Disruption and New Revenue Opportunities" panel discussion here.