i3 - January/February 2018 - 80

"Entertainment search provides a whole new
way for people to consume music, movies and
TV shows."-Brian Hamilton

SPONSORED

Music and Machine
"At a very high level," says Hamilton,
"we use a combination of supervised machine learning and editorial
training to better inform the algorithms." For example, a country music
specialist might listen to and classify
10,000 tracks. Then, says Hamilton,
"the machine learning expert and our
editorial expert will iterate together
to review a set of 'ground truth audio
tracks' and make adjustments to
the classifications of those reference
tracks. This enables the system to get
better and better over time as it learns
and gets more finely tuned by the
human experts."
The editors are training the
algorithms to understand what
combinations of sonic qualities like
harmony, timbre or rhythm in a
recording cause people to perceive
particular moods and genres. When
classifying Taylor Swift's catalog,
for instance, Hamilton says the
system "will identify that many early
recordings were consistently contemporary country-focused, while more
recent ones are more diverse, delving
into dance, R&B, and electronic-oriented pop styles."
The ability to accurately choose
from a broad but consistently defined
set of descriptors on a massive scale
gives machines a distinct advantage.
Writing about his "57 Channels" song
in 2014, Springsteen admitted to
having "no idea what we were aiming
for in this one outside of some vague
sense of 'hipness' and an attempt at
irony." For a more specific description,
ask an algorithm.
80
82

JANUARY/FEBRUARY
JANUARY/FEBRUARY 2018
2018

New Ways to Connect with
Content
All this data shapes the many ways
that consumers can discover and
interact with music, movies and TV
shows now and in the near future.
For example, given a voice search for
the aforementioned "romantic" song,
an entertainment platform or device
could tie the voice to an individual
user. Knowing that person's preferences and the listening context, it
could then help pull the perfect song.
Cross-vertical searches that blend
domains are also possible: "Play that
artist who was on the Jimmy Kimmel
show last night" would tap into both
music and television data. A system
could also supply a user with factual
answers to questions like "What's the
birthdate of that 'Spider-Man' actor?"
or "What was the top song in 1995?"
User interfaces which facilitate
navigation, search and discovery of
new music, movies and TV shows
are also highly visual. With music
now being played on 70-inch smart
TVs, high-resolution mobile screens
and computer monitors and bingewatching entire TV series in a single
sitting becoming the norm, highly
rich imagery has become as important
as descriptive data to the search and
discovery of new content. "We spend
a lot of time trying to make sure we
have what we call an 'image for everything', and usually multiple images
for everything to power various user
interfaces," says Hamilton. "We
optimize for mobile. We do HD, 4K,
3x1, 2x1, 15x9, bands, individual
artists and cover art."
All of those images relate back to

the broader set of metadata. And of
course, with the right data, the images
a user sees are highly personalized.
Giving the Boss More Efficiency
Hamilton points out that metadata's
benefits for discovery are only part
of the equation. "Content creation,
promotion and advertising are key
components that are powered by our
data as well."
The benefits of organized, wellcrafted metadata are also not limited
to music and video entertainment.
Such data can also help users engage
more deeply with sports, breaking
news and podcasts.
Discovery is still the data's most
exciting application. "Entertainment
search provides a whole new way for
people to consume music, movies and
TV shows," Hamilton says. "It's the
solution to the abundance of choice.
If we can get that right, then we'll
have a great increase in the demand
curve, and people's ability to consume
content will grow."
The abundance of content has
solved Springsteen's "nothing's on"
problem. Today, everything's on.
But if Bruce lacked a way to sort
through it all, he'd be justified in
recording another musical complaint.
Fortunately, algorithms fueled by
descriptive information about music,
movies and TV shows and user data
can put The Boss at ease. If he's still
up late staring at a screen, it will be
because something's on. And from all
possible options, that something will
be just the thing he really wants.
DAVILES

and machine.

II T
T II S
S II N
NN
NO
O VAT
VAT II O
ON
N



Table of Contents for the Digital Edition of i3 - January/February 2018

Contents
i3 - January/February 2018 - Cover1
i3 - January/February 2018 - Cover2
i3 - January/February 2018 - Contents
i3 - January/February 2018 - 2
i3 - January/February 2018 - 3
i3 - January/February 2018 - 4
i3 - January/February 2018 - 5
i3 - January/February 2018 - 6
i3 - January/February 2018 - 7
i3 - January/February 2018 - 8
i3 - January/February 2018 - 9
i3 - January/February 2018 - 10
i3 - January/February 2018 - 11
i3 - January/February 2018 - 12
i3 - January/February 2018 - 13
i3 - January/February 2018 - 14
i3 - January/February 2018 - 15
i3 - January/February 2018 - 16
i3 - January/February 2018 - 17
i3 - January/February 2018 - 18
i3 - January/February 2018 - 19
i3 - January/February 2018 - 20
i3 - January/February 2018 - 21
i3 - January/February 2018 - 22
i3 - January/February 2018 - 23
i3 - January/February 2018 - 24
i3 - January/February 2018 - 25
i3 - January/February 2018 - 26
i3 - January/February 2018 - 27
i3 - January/February 2018 - 28
i3 - January/February 2018 - 29
i3 - January/February 2018 - 30
i3 - January/February 2018 - 31
i3 - January/February 2018 - 32
i3 - January/February 2018 - 33
i3 - January/February 2018 - 34
i3 - January/February 2018 - 35
i3 - January/February 2018 - 36
i3 - January/February 2018 - 37
i3 - January/February 2018 - 38
i3 - January/February 2018 - 39
i3 - January/February 2018 - 40
i3 - January/February 2018 - 41
i3 - January/February 2018 - 42
i3 - January/February 2018 - 43
i3 - January/February 2018 - 44
i3 - January/February 2018 - 45
i3 - January/February 2018 - 46
i3 - January/February 2018 - 47
i3 - January/February 2018 - 48
i3 - January/February 2018 - 49
i3 - January/February 2018 - 50
i3 - January/February 2018 - 51
i3 - January/February 2018 - 52
i3 - January/February 2018 - 53
i3 - January/February 2018 - 54
i3 - January/February 2018 - 55
i3 - January/February 2018 - 56
i3 - January/February 2018 - 57
i3 - January/February 2018 - 58
i3 - January/February 2018 - 59
i3 - January/February 2018 - 60
i3 - January/February 2018 - 61
i3 - January/February 2018 - 62
i3 - January/February 2018 - 63
i3 - January/February 2018 - 64
i3 - January/February 2018 - 65
i3 - January/February 2018 - 66
i3 - January/February 2018 - 67
i3 - January/February 2018 - 68
i3 - January/February 2018 - 69
i3 - January/February 2018 - 70
i3 - January/February 2018 - 71
i3 - January/February 2018 - 72
i3 - January/February 2018 - 73
i3 - January/February 2018 - 74
i3 - January/February 2018 - 75
i3 - January/February 2018 - 76
i3 - January/February 2018 - 77
i3 - January/February 2018 - 78
i3 - January/February 2018 - 79
i3 - January/February 2018 - 80
i3 - January/February 2018 - 81
i3 - January/February 2018 - 82
i3 - January/February 2018 - 83
i3 - January/February 2018 - 84
i3 - January/February 2018 - 85
i3 - January/February 2018 - 86
i3 - January/February 2018 - 87
i3 - January/February 2018 - 88
i3 - January/February 2018 - 89
i3 - January/February 2018 - 90
i3 - January/February 2018 - 91
i3 - January/February 2018 - 92
i3 - January/February 2018 - 93
i3 - January/February 2018 - 94
i3 - January/February 2018 - 95
i3 - January/February 2018 - 96
i3 - January/February 2018 - 97
i3 - January/February 2018 - 98
i3 - January/February 2018 - 99
i3 - January/February 2018 - 100
i3 - January/February 2018 - 101
i3 - January/February 2018 - 102
i3 - January/February 2018 - 103
i3 - January/February 2018 - 104
i3 - January/February 2018 - 105
i3 - January/February 2018 - 106
i3 - January/February 2018 - 107
i3 - January/February 2018 - 108
i3 - January/February 2018 - Cover3
i3 - January/February 2018 - Cover4
http://www.nxtbook.com/nxtbooks/manifest/i3_20190708
http://www.nxtbook.com/nxtbooks/manifest/i3_20190506
http://www.nxtbook.com/nxtbooks/manifest/i3_20190304
http://www.nxtbook.com/nxtbooks/manifest/i3_20190102
http://www.nxtbook.com/nxtbooks/manifest/i3_20181112
http://www.nxtbook.com/nxtbooks/manifest/i3_20180910
http://www.nxtbook.com/nxtbooks/manifest/i3_20180708
http://www.nxtbook.com/nxtbooks/manifest/i3_20180506
http://www.nxtbook.com/nxtbooks/manifest/i3_20180304
http://www.nxtbook.com/nxtbooks/manifest/i3_20180102
http://www.nxtbook.com/nxtbooks/manifest/i3_20171112
http://www.nxtbook.com/nxtbooks/manifest/i3_20170910
http://www.nxtbook.com/nxtbooks/manifest/i3_20170708
http://www.nxtbook.com/nxtbooks/manifest/i3_20160102
http://www.nxtbook.com/nxtbooks/manifest/i3_20160304
http://www.nxtbook.com/nxtbooks/manifest/i3_20160506
http://www.nxtbook.com/nxtbooks/manifest/i3_20160708
http://www.nxtbook.com/nxtbooks/manifest/i3_20170506
http://www.nxtbook.com/nxtbooks/manifest/i3_20170304
http://www.nxtbook.com/nxtbooks/manifest/i3_20170102
http://www.nxtbook.com/nxtbooks/manifest/i3_20161112
http://www.nxtbook.com/nxtbooks/manifest/i3_20160910
http://www.nxtbookMEDIA.com