Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now


PMC2018 Explores Getting Paid, Artificial Intelligence

By Steve Harvey. PMC2018 addressed a broad swath of issues related to production music, focusing not only on business issues, such as best practices for securing placements and, equally importantly, getting paid, but also how to maximize composition and production quality.

Hollywood, CA—“The more composers work with live musicians, the richer the texture of the music that can be licensed through a library,” advised Dominic Houston, head of the music team at Netflix, during his business keynote presentation at this year’s Production Music Conference (PMC2018) in late September.

The annual event, hosted by the Production Music Association, the leading advocacy group—with over 670 members—for the $1-billion-a-year global industry, is now in its fifth year. PMC2018 addressed a broad swath of issues related to production music, focusing not only on business issues, such as best practices for securing placements and, equally importantly, getting paid, but also how to maximize composition and production quality.

Houston, who related that he got into funk in the mid-1980s after spying Parliament’s Uncut Funk—The Bomb album art (George Clinton, in sheepskin chaps, shouldering a boombox and surfing on two dolphins) in his brother’s collection, dropped some significant numbers on the crowd. Netflix has hundreds of series in development, and there can be 150 pieces of music requiring sync licensing for each TV series or film, he said. Houston’s 25-member team must handle all that paperwork: “People won’t get paid on time if we don’t stay on top of it.”

Much of that music is original, of course, but production music does get used, especially on the marketing side, he explained. “I think there’s a consensus that the [production music] industry has really changed in the last five to 10 years; the quality of the music available has improved dramatically. The old clichés about library music are long gone.”

Netflix is challenged by scale, Houston continued, and leverages technology to smooth the process for all stakeholders: the music supervisor, the Netflix team and the library publisher. Aiding in that process, “We recently partnered with Trevanna Tracks; it’s an end-to-end project management tool,” he shared.

Related: PMA Conference Session: How Will AI Impact the Future of Production Music?, by Steve Harvey, Aug 30, 2018

Music library publishers employ a variety of tools to track broadcast use of their music, including fingerprinting and watermarking. Audio watermarking is metadata embedded in the audio file from its creation, spread throughout the audio spectrum, explained SourceAudio’s Hunter Williams on a panel entitled “The State of Audio Recognition and Monitoring Technology.” Fingerprinting analyzes the “DNA” of the audio file after the fact, creating a reference to that pattern, he added; it’s not embedded. Both methods have pros and cons, he said.

Current automatic content recognition (ACR) platforms include BMAT, Sound Mouse, Tunesat, proTunes, Digimarc, Audible Magic, ACRCloud and Sound Aware. Artificial intelligence is improving the accuracy of automatic content recognition, reported BMAT’s Iñigo Ugarteburu.

The ACR process still relies on broadcasters supplying cue sheets listing the music used, but challenges remain—for instance, detections can get matched to the wrong TV episode, and the use of stems, loops and beats can confuse things. Turner’s Shane Sanders noted that 16 hours a day of CNN’s programming is live. “The production staff doesn’t generate cue sheets for that on a daily basis; budgets are tight,” he said. Sound Mouse has been “an incredibly helpful solution,” he noted, generating dynamic cue sheets with video references for Turner and the performing rights organizations.

Reporting data varies significantly from one country to the next, said Paul Sims of proTunes. “There are 168 pieces of metadata that you need to collect,” he said.

BMAT handles 60 million tracks and monitors thousands of TV and radio stations worldwide for usage. Retitling—where multiple libraries offer the same track with different titles—is rampant and could lead to reporting errors. But, said Ugarteburu, “It can be analyzed and solved; there are ways of identifying which of those works are considered hybrid.”

A panel on AI-generated music went well beyond the preview in PSN’s October issue. Indeed, Mick Kiely of AI music platform Xhail broke some news at the event.

“We have a major corporate client who commissioned a branding track. We’ve created a platform so that now, when they create video—and they make maybe 250 internal educational videos every year—they can, in real time, create a new version of that brand identity. We’ve been able to extract the DNA that is their brand and can now produce unlimited music that is associated with that brand. That’s a new direction for us,” said Kiely.

Asked by moderator Andrew Gross where his platform was weakest currently, Drew Silverstein of Amper Music responded, “We’re really good at a cappella vocals and big choirs. One of the challenges is doing singer/songwriter-style vocals. That’s difficult for Amper to do convincingly—unless you want something to sound processed.”

Gregory D. Lapidus, who has been focused on intellectual property issues in the music industry for about 30 years, weighed in on the questions of copyright in AI-generated music. “AI has the potential to augment human ability,” he said. “I don’t think it’s anything to be afraid of.”

Copyright attaches when an original work of authorship is fixed in a tangible medium, he continued. In the case of AI-created music, the central issue is that of originality.

Want more stories like this? Subscribe to our newsletter and get it delivered right to your inbox.

AI software is created by someone and they program certain parameters, and some AI music software introduces randomness to give the impression of originality, he said. “But I don’t think AI is creating original works of authorship. I think that the people programming the software are the ones creating the original works of authorship,” said Lapidus.

AI means that the future of music will be very different, said Silverstein. But the value of human-created music is the artistic collaboration of people making art together, he stressed. “That will never go away. [Director Steven] Spielberg is not going to use Amper when he can work with [composer John] Williams.” But Amper does enable artists to be more creative, productive and successful, he said, comparing it to the pre- and post-Pro Tools worlds.

From the audience, composer Gil Talmi commented, “The words ‘moral compass’ come to mind. It’s not a question of, Should we be doing this or not, but Should we do it from a place of mindfulness? The technology can take over before we know what happens.”

Production Music Association •