WEBVTT

00:00:01.460--> 00:00:08.674
My name is Henk van den Heuvel and I am a speech
technologist at the Radboud University of Nijmegen.

00:00:09.325--> 00:00:15.948
I have an incredible passion for
sound and for making it available.

00:00:15.973--> 00:00:22.850
And it's right in the 20th century that sound
could actually be recorded and analyzed

00:00:23.330--> 00:00:26.799
I started my career as a phonetician
but developed into a speech engineer

00:00:26.824--> 00:00:29.847
and that's a development which I
consider as extremely important.

00:00:30.250--> 00:00:34.923
For the Digital Humanities, the
field in which we are active,

00:00:34.948--> 00:00:39.270
it means that speech technology
has a lot to offer with

00:00:39.295--> 00:00:44.001
regard to opening up audio and
video and making it available.

00:00:44.040--> 00:00:50.612
As speech technologists we can make an important
contribution by digitizing analogue material

00:00:50.637--> 00:00:55.117
such as tape recordings,
turning them into digital data.

00:00:55.236--> 00:01:00.314
But as speech technologist the next
step is to make the data accessible,

00:01:00.339--> 00:01:03.982
because if you digitize it and
put it back on the shelve,

00:01:04.069--> 00:01:07.545
people have the possibility to request it, but
they don’t know what the content is about.

00:01:07.656--> 00:01:12.443
By applying speech technology methods
you can really make a difference.

00:01:12.483--> 00:01:16.109
I will give you some examples of
what speech technology has to offer.

00:01:16.165--> 00:01:20.257
Think of automatic speech recognition
(ASR), which makes it possible

00:01:20.282--> 00:01:24.572
to convert the spoken
audio into text,

00:01:24.715--> 00:01:27.373
so that it can be read.

00:01:28.531--> 00:01:35.365
You can transcribe the text manually, but this is
incredibly more time consuming then using ASR.

00:01:35.619--> 00:01:42.238
What hardly no one knows, is that speech
technology is also able to recognize speakers.

00:01:42.263--> 00:01:46.458
It is able to identify who
is speaking at what time.

00:01:46.483--> 00:01:52.919
This information can be added to the recording
enriching it, making it readable and more accessible.

00:01:54.030--> 00:01:57.100
What the speech technologist is
able to do, is on the one hand

00:01:57.125--> 00:01:59.975
make the material
available in digital form

00:02:00.030--> 00:02:05.595
and on the other hand, and this is what I
find most interesting, opening up the data,

00:02:05.627--> 00:02:11.276
And for me as speech technologist the material
that belongs to the field of Digital Humanities

00:02:11.379--> 00:02:17.830
is about people who want to voice their experiences,
who they are, and what they have experienced,

00:02:17.855--> 00:02:22.998
and what they consider important, what they
want to sing what they want to tell the world.

00:02:24.244--> 00:02:30.744
Speech technology offers opportunities
to further delve into this material.

00:02:31.045--> 00:02:36.410
It is not about casual talk, that you
use when talking on your mobile phone.

00:02:36.513--> 00:02:38.679
It is about material
that poses a challenge

00:02:38.743--> 00:02:40.940
Because on the background all
kinds of sounds can be heard

00:02:40.940--> 00:02:46.457
cars are driving by, doors are slammed,
several people are speaking at the same time

00:02:46.488--> 00:02:51.419
Speakers are often elderly people, for
instance when interviewing veterans,

00:02:51.712--> 00:02:54.377
and this offers a particular
challenge for speech technology.

00:02:54.402--> 00:03:00.515
We are able to surpass these obstacles and to offer
these people a way of opening up their stories

00:03:00.555--> 00:03:06.990
in a fast and efficient way and to make
sure that their voices are actually heard.