Transcriber: Hubert Stasik
Reviewer: Ines Dif I began studying piano at the age of four, and as a young child I
would hear music on the car radio while my mother was taking
me to school, and I would start to sing, the next note that I was sure was about
to play, and I asked my mother, do all people hear music like this? Predicting what the next note will
be before they've heard it? She assured me that I heard
music in a unique way. Little did I know that decades later, AI would be trying to pred
ict that next
note or the next series of chords to help a human composer as a virtual collaborator
in writing music. It’s similar. The reason why music can do
this is because music is math. And when you when I mean by that is
that music is highly quantifiable and measurable in time, pitch, uh, duration, volume, frequency, everything. And it's highly systematized. AI extrapolates patterns and other earlier
iterations of music to determine what might be the next logical series
of pitches, chords,
or lyrics. And in so doing, it offers a human
collaborator, a human composer, the possibility of that next note chord, or at least a series of lyrics in the same way that your smartphone
proposes possible words for you to use while you're
typing a text. Today I work at the intersection
of music, law, and technology. I'm often asked to give an opinion and
testify in court as to whether or not one song copied another in the in the Protection of Intellectual Property
and Music creativity in our Le
gal system. And yet, after a lifetime of study, I am still trying to find out the
answer to how a musical work, full blown as it is, came
out of nothing. Zero. How did a show like Hamilton come to be, and how did Beethoven come up with his
fifth Symphony when he was deaf and lived in total isolation? I wonder how Beethoven would have been
even a greater composer had he had AI and other technology to help him hear his
music and experience the vibrations, and instead of waiting for an orchestra
to
learn his great works, slowly and deliberately, over time, he could have had a virtual orchestration
of it in his studio and immediately hear his own music.
Here's an example. I still feel that that's a masterpiece
of four notes in originality. Still today. I feel that music, is being democratized by AI because it's not all about
just the geniuses. Average people can now write music,
access it, and develop some of their own creativity
through the use of AI to assist them and help them with skil
ls they
may not otherwise have. How is this possible and
why is it important? Because of all forms of communication. I truly believe that music is the
most universal of all languages. Regardless of whether or not one hears
the words and understands them. In a concert, I see people in concert
swaying to music, crying tears, dancing with joy because of the spirit
of the tones and the melodies and some of the other elements of
a concert when it's sung in a foreign language that
they don't understan
d. This is because music goes straight
past intellectualization into and deeply into our hearts, in my opinion. Music is also a great healer
and it always has been. But today, with the use of AI, music has become something of
a superpower as a healer. By this, I mean that music has reached
the depths of mentally ill, cognitively impaired, and at times
brain damaged patients in a way that only medication and
only therapy alone have not. This has been proven scientifically,
and recently, the Natio
nal Institutes of Health have
partnered with the great university, MIT and the Media Lab there to discover
that there are gamma frequencies when used in audio form, that improve or even
turn back the advancement of Alzheimer's. In addition, in a Boston area hospital, patients who are severely paralyzed due to
cerebral palsy are able to write their own music just by wearing a headset
and by looking at a screen. I've seen the expression of
pure joy on their faces. When imprisoned in their own bodi
es, they hear the music that they themselves
created. This is very, very hopeful. Music also reaches beyond with
eye reaches beyond healing, and helps average people with limited or
no musical education to create music and participate in the creative
and communication process. It allows people to overcome
barriers also. But in terms of it helping
people to create music on their own with very little education. We have heard now that there
is a program called a hyperscore, that young children who
don’t have a musical education and maybe don’t even yet read, can draw
with colors and figures on a screen, and AI can interpret that
into musical sounds, and encourage them that their creativity
means something and is being expressed, and some of their deeper feelings.
As well we can now. We are now learning that AI is overcoming
certain barriers that musicians have always had in many ways. By that I mean
that composers were always taught. Limit your song to the lowest and highest note that the
singer, the soprano,
can sing or only write music, that the bass can play, nothing too high,
nothing lower than the bass, etc. You’re limited by the instrument,
or the vocal capacity, or the talents or the skills of the
musician and the instrument. But, for many years, in fact,
for over a century, composers have wanted to work
around those obstacles, and they’ve done it in
non-technological ways. For example, there's an entire body
of piano music written only for the left hand, for concert pia
nists
who had their right hand amputated or disabled, sometimes in a war. But with AI, we've even overcome the
barriers of physical instruments or vocal techniques by enhancing
it through the use of AI, and enlarging our, our experience of sound
and experience in performing. AI has also enabled us as music has over
the century, to travel over distance,
time and even death. By that I mean that today we know of
recordings that are being created when the composer or band is no
longer either perform
ing, or perhaps the members
are no longer alive. There are many pieces of music, both in the classical repertoire
and the popular repertoire, that were never finished
because of the death of a composer who left his opera partly
written, or a song wasn't completed, or it was lost somewhere in
a studio or in a vault. Before, after, before which
the, it could be completed, the writer passed away or
the band disbanded. Recently, we've heard of a Beatles
song that has been created from the singing of
John Lennon, that needed to be cleaned up in a way
that it could be better heard, and eventually was recently heard
in a recent film about the about the Beatles and about some
of their experiences. Whereas this had been lost over time
because it was impossible to create, to clean up the audio until AI could learn
from it enough to improve it. Also, many films are using music, or vocal lines from artists who are
no longer with us by learning from, in this case, Edith Piaf, the great
French singe
r and actress. Earlier films and recordings trained AI, and recently there was a recent biopic
in which she was actually shown as if she were the one speaking and
singing when she actually wasn't. Still, with all of these expansions of
creativity and the abundance of content, there are risks involved that
need to be considered. For example, not every artist wants his
or her likeness or voice impersonated. There are estates involved
in legal disputes because an artist's voice was used
without per
mission, and the music industry is very, very
strong in its enforcement. And you may have heard about certain
takedowns recently due to deep fakes of some of perhaps your favorite artists. Also, not everybody wants to be implied
as being the composer of something, just because the computer and machine
learning could learn from it enough to write what might have been the
next Taylor Swift song, but she didn't write it.
So there are legal and ethical risks in doing everything that
possibly could b
e done by technology. Where are the boundaries between
inspiration and copying? This is a question I'm actually asked
to answer in the court of law, when copyright infringement and music
plagiarism has been suspected or accused. I was asked to be an expert witness by the Marvin Gaye estate, in representing
his song, Got to Give It Up, against Pharrell Williams song, Blurred Lines. In the recent past and in such, I was asked to dissect the music for the
jury and help them understand if I saw them
as one is copying the other, and if it met the legal standard
called substantial similarity, which in means that it has been
copied in a legal sense and consequences would take place. I was able to dissect the music
and find that, in fact, the music had been reverse engineered
in a way that, in fact, I felt that there was a strong
enough similarity of a large number of unique qualities from
Marvin Gaye’s song found in Blurred Lines to the level that I did believe that
it was substantially simil
ar. I was able to show the jury one
particular area that was very dramatically similar, in that the two
works had a very similar midsection. One was called a parlando
in Marvin Gaye's song, and the other was called was
a rap in the other song. But the way they began and ended were
identical in something called, word painting, where the identical words
up, down, round were set to the identical melodies to illustrate those
words symbolically as a high note for up, a low note for down, and
the exac
t same notes, etc. The jury awarded a resounding victory
for the Marvin Gaye estate, and the music industry has been
reeling about it ever since. Still, that aside, I truly believe
that the miracle of humanity is the desire to create and invent something
that was never there before. We are all born with an imagination,
curiosity, and a voice. The original instrument. I believe that we have a deep need to
be heard on some level, all of us. With AI, we have expanded the ability
to communicate and
to collaborate as never before. In the words
of the great Victor Hugo, “Music expresses that which
cannot be said and on which it is impossible
to be silent.” Thank you. (Crowd Applause)
Comments