
[Music]
Y
[Music]
how
despite how much we have changed
as a society, the voice is still today
main mode of communication between
people our voice is a tool
very versatile that allows us to transmit
a lot of information and not only through
meaning of the words we say
but also on how we say them the tone
of voice the intensity at the speed we
allow you to add elements of
expressiveness to the message that we are
transmitting the voice is also a
powerful musical instrument with which
express ourselves artistically the voice is
generates when the air comes out driven from
the lungs and go through our larynx
vibrates the vocal cords that in
reality are the folds that open and
they close at a speed that can go
from 60 to 1000 times per second in the
case of a soprano during phonation
this acoustic wave goes through then
our vocal treatment that depending on
its form and disposition
the position of the tongue the opening of
the mouth allows us to generate the
different vocal sounds this sound
we can capture it through a
microphone and convert it to a signal
electrical or digital what allows us
then store it transform it
visualize it and then reproduce it
through a loudspeaker converted again
in sound but what happens if we want
create a voice that has never left
a human mouth voices and artificial
they do not try to imitate human physiology
but what they try is to create
a signal as similar as possible to a
actual recording that is to say we are not
imitating the phonetic device but what
we do is try to reconstruct the wave
acoustic that this device is
producing
and the question we can ask ourselves is
it makes sense to express ourselves as humans to
through an artificial voice next to a
team of researchers we’ve been for years
working the case of the sung voice
trying to develop technologies that
allow us to generate an artificial voice
sung with all the possibilities of
musical expression I’m going to put a
example where from a score
we can create this song
[Music]
so that you
but you will not live
others
now you can imagine a composer
of opera that is composing his work in
your home and you can hear their ideas
musicals sung by a soprano or by
the voice of a virtual castrato or us
we can imagine reliving the voice of a
famous actor to generate dialogues of
a new movie or why not from a
new interactive video game or have
your own artificial voice that
answer for you to calls
telephone or generate an artificial voice
that can double content in others
languages to create artificial voices
the methods used are based on what
we call automatic learning these
methods
what they do is analyze a lot of quantity
of data duly noted in our
case our data are the recordings
of a person’s voice and the text that
accompanies these algorithms what they seek
is to identify patterns in
acoustic characteristics of these of
these signs
that is to say in these recordings according to whether
is a question the intonation
in a way ending the sentence going up
the intonation these are characteristics
acoustics that these methods can
They can also learn how
recordings and features
acoustic in these recordings come from
from a specific person we can generate
an artificial voice of a person
specifically and if we want to add
emotions what we need
will be recordings with different
emotions with what we can generate
then different artificial emotions
I’m going to give you an example of a
normal education in neutral mode
and now the same voice with a little more
angry
Saint Louis then
we can create an artificial voice of
anyone, the answer is
what if we have enough data is
say in our case if we have
enough recordings of a person
we will be able to create a voice
artificial
currently the methods that we are
using need around a
recording time at least until the
dozens of hours at the best of the
cases if we want to get a voice
artificial with maximum expressiveness
but researchers around
world are investigating systems with
those that from a few minutes of
recording of a person power would create
an artificial voice these new methods
what they do is add recordings of
hundreds of different people creating
a map of voices in great detail and
then with a few minutes of recording
a single person we can already
take advantage of this map of voices with
detail to create a new voice
artificial we have created
different artificial voices with our
team here you have the drawing of a
virtual singer maika with whom the
users can create new songs or
We also participate in a campaign
advertising where the fans of the ear of
van gogh
they could compose lyrics that was sung
on top of the group’s music but for
see the expressive possibilities of a
artificial voice the best thing is that you listen
an example of an experiment that
we made we took the voice of three
speakers who are not particularly good
singers and we made them sing a
song from scratch is to say a song
that they never recorded
When I went, you went
it’s my time
when I sing
do not
no no
[Music]
and
next
yes
do not
[Music]
Let’s see
or
saiz
climbing
friend
[Music]
to be detected
[Music]
yes
[Music]
as we have seen in this example
the artificial voices next to the
latest advances in intelligence
artificial puts us on the table
ethical debates about authenticity and
privacy in the society of the future the
doubt is if we want to overcome the
reluctance to synthesize something so intimate
and personal as the voice or really we
terrifies that possibility of not knowing
distinguish something real from something that
It is not real but it is also very
important to take advantage of the
creative possibilities of the new
technologies to create new ways
cultural that is really what
we have been doing since the beginning
of humanity and finally I
I leave with a question would you go to a
concert of a hologram by david bowie
presenting his new album created with
an artificial voice
see you there thank you very much
[Applause]
[Music]
Y
[Music]
how
despite how much we have changed
as a society, the voice is still today
main mode of communication between
people our voice is a tool
very versatile that allows us to transmit
a lot of information and not only through
meaning of the words we say
but also on how we say them the tone
of voice the intensity at the speed we
allow you to add elements of
expressiveness to the message that we are
transmitting the voice is also a
powerful musical instrument with which
express ourselves artistically the voice is
generates when the air comes out driven from
the lungs and go through our larynx
vibrates the vocal cords that in
reality are the folds that open and
they close at a speed that can go
from 60 to 1000 times per second in the
case of a soprano during phonation
this acoustic wave goes through then
our vocal treatment that depending on
its form and disposition
the position of the tongue the opening of
the mouth allows us to generate the
different vocal sounds this sound
we can capture it through a
microphone and convert it to a signal
electrical or digital what allows us
then store it transform it
visualize it and then reproduce it
through a loudspeaker converted again
in sound but what happens if we want
create a voice that has never left
a human mouth voices and artificial
they do not try to imitate human physiology
but what they try is to create
a signal as similar as possible to a
actual recording that is to say we are not
imitating the phonetic device but what
we do is try to reconstruct the wave
acoustic that this device is
producing
and the question we can ask ourselves is
it makes sense to express ourselves as humans to
through an artificial voice next to a
team of researchers we’ve been for years
working the case of the sung voice
trying to develop technologies that
allow us to generate an artificial voice
sung with all the possibilities of
musical expression I’m going to put a
example where from a score
we can create this song
[Music]
so that you
but you will not live
others
now you can imagine a composer
of opera that is composing his work in
your home and you can hear their ideas
musicals sung by a soprano or by
the voice of a virtual castrato or us
we can imagine reliving the voice of a
famous actor to generate dialogues of
a new movie or why not from a
new interactive video game or have
your own artificial voice that
answer for you to calls
telephone or generate an artificial voice
that can double content in others
languages to create artificial voices
the methods used are based on what
we call automatic learning these
methods
what they do is analyze a lot of quantity
of data duly noted in our
case our data are the recordings
of a person’s voice and the text that
accompanies these algorithms what they seek
is to identify patterns in
acoustic characteristics of these of
these signs
that is to say in these recordings according to whether
is a question the intonation
in a way ending the sentence going up
the intonation these are characteristics
acoustics that these methods can
They can also learn how
recordings and features
acoustic in these recordings come from
from a specific person we can generate
an artificial voice of a person
specifically and if we want to add
emotions what we need
will be recordings with different
emotions with what we can generate
then different artificial emotions
I’m going to give you an example of a
normal education in neutral mode
and now the same voice with a little more
angry
Saint Louis then
we can create an artificial voice of
anyone, the answer is
what if we have enough data is
say in our case if we have
enough recordings of a person
we will be able to create a voice
artificial
currently the methods that we are
using need around a
recording time at least until the
dozens of hours at the best of the
cases if we want to get a voice
artificial with maximum expressiveness
but researchers around
world are investigating systems with
those that from a few minutes of
recording of a person power would create
an artificial voice these new methods
what they do is add recordings of
hundreds of different people creating
a map of voices in great detail and
then with a few minutes of recording
a single person we can already
take advantage of this map of voices with
detail to create a new voice
artificial we have created
different artificial voices with our
team here you have the drawing of a
virtual singer maika with whom the
users can create new songs or
We also participate in a campaign
advertising where the fans of the ear of
van gogh
they could compose lyrics that was sung
on top of the group’s music but for
see the expressive possibilities of a
artificial voice the best thing is that you listen
an example of an experiment that
we made we took the voice of three
speakers who are not particularly good
singers and we made them sing a
song from scratch is to say a song
that they never recorded
When I went, you went
it’s my time
when I sing
do not
no no
[Music]
and
next
yes
do not
[Music]
Let’s see
or
saiz
climbing
friend
[Music]
to be detected
[Music]
yes
[Music]
as we have seen in this example
the artificial voices next to the
latest advances in intelligence
artificial puts us on the table
ethical debates about authenticity and
privacy in the society of the future the
doubt is if we want to overcome the
reluctance to synthesize something so intimate
and personal as the voice or really we
terrifies that possibility of not knowing
distinguish something real from something that
It is not real but it is also very
important to take advantage of the
creative possibilities of the new
technologies to create new ways
cultural that is really what
we have been doing since the beginning
of humanity and finally I
I leave with a question would you go to a
concert of a hologram by david bowie
presenting his new album created with
an artificial voice
see you there thank you very much
[Applause]
[Music]
Please follow and like us: