Embed code
Note: this content has been automatically generated.
current experiments
can get results so so basically this is like a case study
so i i basically it's talk very briefly about speaker discussion pens and what they are
uh how the random data and then also the weather like detection is and um what attacks
after the speaker traditional parameter speak about metrics and how you fuse those two together
i was like i'm going to church right now i'm so stupid he is because occasions
system is basically a new in the role a speech sample from some people
you extract some features you model speaker and then you when you
then create a model so you speakers uh in this in the system and then when a test somebody wants
to actually check that uh uh looking into the system of the cape may ask it by the system
he speak into the microphone and uh if you sample those
six weeks the features extracted from the sample and uh
it's compared to uh with the model of the speakers in the database and then
uh it the checks if they're actually the speaker is uh is is user is the the one that isn't that the based on that
if he's matching uh refuse actually matching that the person who's pretending to be
and have like a decision score or whether this is a a person is verified remote
so that the big o. is out all the experiments uh four speakers occasion uh something
like this like a like a plot that do so only give faces like uh
a face recognition but also for speak recognition similar you have on the left side you have uh
uh scores that belong to a zero of the process the call
that basically known jane uses the users that uh that
that that are not those who claim they yeah basically different user send them blue scores of those that uh
uh actually jane user so the point of the system the good ones to separate these two
scores as far as possible so they don't overlap that is great system that means
it's works really well with the glasses speakers were well they're overlapping
too much the the speaker recognition system is not very good
so um how do you do this type of expense and back to so uh as
a as already mentioned binder uh uh as well so you you need to database
and you need to have some kind of a protocol that as a subset and that the basic the
full training you train like a general how model of
what is speech speaker in general is speech
and then uh um you train uh also in intro people specifically
and then uh you also passed on the different speakers
and uh i decided the base unit software and hardware which is quite important and um
you need specialised software that can do these uh tasks
and uh uh extract features generic models compute metrics
and uh at all so you need to have a environment around this of the on
and the hardware that is actually possible uh to use like we store big databases
do you is the g. p. use something there's a lot of neural networks now that require
a lot of people searching for example um so how
do run experiments so uh well one way mm
it's popular way you do take some typical sure software like uh something special is it is quite probable
and you know there's a lot of libraries a lot of things that the return for you
but there's no infrastructure set up for easy to
run an easy to reproduce anything even baseline
so this is an example that you can you have to do is review young researcher
this is a typical software some people might recognise with and um you basically have this type of weather long
very long file like bay somebody doesn't like go and this is basically the state of the art thing
and then have to figure out understand all this and there's a lot of hard coded like bash script here
that have some ah things and some environmental variables in
the linux system that you don't know actually you
what that system is what the what does the required to run that by scrape
on i mean which basher and who knows it's a lot of things here
and uh you like copy folders of data uh from one folder
to another to run your feature extraction so if you have
for different features tractors you have to have four different corpus of your data which can better by it
you have to have four provides disk space and then there's a lot of things
so it's quite a quite difficult situation here at this to me so
it's very difficult to reproduce later so if you say you give descriptor somebody else's
guy say or you have a great model together to stock in a
round the same thing yeah you can that's great good luck to him and
then you never play his email again pretend you don't exist basically
and then so there's another way so as standard short we have this bob framework that we try to
have um a good back to so for a approachable to sharing the cotton the results and uh
uh papers uh uh and also we can generate figures for the
papers uh in a in a repeatable manner uh so
both framework what it does it uh it's actually was built
is a possibility my that is important to exhibit design
uh and uh 'cause some genetic to change for experiments that the useful uh like a threshold
to the face recognition experiment but as we have the same the same tool chain
applies for speak recognition the only thing you need to change is a small blocks and
sidetracked with everything else is actually the same that framework is exactly the same
when recognition well it's different at the base and the different feature extractor so that's about it everything else assume it's easy
so it is the complexity of accessing dating stalling analysts any running things
so that there are some problems there that that is possible to the user should still you know
you still have to have uh the uh uh the good stuff hardware to run this thing
there's some good friends are hard to run on their laptop so you need to maintain
this constantly you need to design experiment as well i mean but that's that's
you always get if you want to do except new experiment you have to design it and actually run it
and then you need to learn how to report unsure results it's as additional thing framework does not really provided
so therefore of the next step revolutions that as
people from that if you just so you'll
basically need for what is important is to understand in tactics is is possible
to the user would you still need to do when using b.
you still need if you have a new database you want to use you still have to ported into plot for which is the process you
have to go through like a great interface support for when stand whether they teach them how to do with it then you have to
to design the tool chain like actually like up your specific framework for
your experiment like what the building blocks what do they do like
which date uh how they connected with what algorithms and how they're get on the last which sometimes can be
uh and work effort to to have to put in the beginning and allocate experiment in
and around and then running is the easy part so the setting up a bit
can be her so he is the um the the actual practical experiment that
that uh i did for them one of the papers mm so i'll
are are basically start from the report so this is the public this is the
is the link that i have here the slides will be available can click late you don't need to that group that long uh uh at
so this is a public report that uh that is basically linked in the into the paper that was published
as a footnote and then person can go here see documentation that like above description what exactly this thing does
uh i have the grabs a the like the ticker so which is kind of like a particular
some uh uh expenses up like uh the the scholars unlike metrics
of uh and then you can click on links experiment itself
and so it's quite a different from them um
this the simplest experiment for was taken
faces is it much more blocks we can see there's a large list of
to uh to uh like uh the blocks that you need to go
through to complete this thing i will complicate feature structure model building
a lot of uh you know big an ally lighting block scorers scoring
the you know all those on the ledge database and also the
timing can notice it to tend to run on our farm out for
the training for example but it takes a couple thousand seconds
so which is what like a a so you can see like algae members uh basically you want to go get it looks like
so this is basically german or so then important but also be that the a
lot of things are hidden here in the sense is provided by people from
so they they so there is a simple i wrote was pretty simple right here
because they're already there's a lot of underlined call that is available
like bob actually use installed indeed so sort of demand
c. plus plus implementations that you just call functions here for your actually specific experiments which quickly uh_huh
okay so this is the experiment the uh for the different verification
uh there are like a different types experiences different databases that they that they use
because i want to see how are from different uh different types of things uh uh so yeah
so far look at so i can also like i'm just show you how you can for can create a new one so i can just uh
to be longer than the other one i'm not good run anything because it takes like a
a few hours to actually finish but you can change the the basis and the there's a selection of gifts
you 'cause you little protocol so that the basis you can try and different things of parameters much more
complex you can have a lot of you know uh uh the good the good thing about the is that when you select a lot
if if it doesn't match and vitamin or something it doesn't appear in the list because you have to look at least here
of a different uh things you can run like a like let's say feature extractor i
have i can have different uh it it it you know my this one that
mm kay so uh this is basically speaker together so you you go to some
just the way the um what it is so well act detection is um
it's uh related type of experiment but instead of a refines because you figure out if they're
trying to spoil the system basically the trying to pretend uh actively yes she died actually
page that an interesting uh so uh you trying to figure out
uh if somebody's actually trying to be pretend somebody he's not
so the typical way so this is like a point of attacking the system that that
most to because uh this type of attack like a one which call here
is when somebody record your your steely recording or download for me to be recording use of your voice
and played back to their education system in trying to pretend that you you uh she's you basically so
i think okay so they're basically that x. can fully occasion system so that access elaborate this reply back
because you you play back just it with all the changes a basic another speaker instead of
person speaking in microphone directly there's a speaker playing back the voice into
the microphone that's only difference really in that is that what that
so that tax here that overlap with the g. with the actual jane scores that means
system much on the stand that that is actually real person a and person little
uh authentic it'd be trying to be instigated by the system that means
you can spoof it that's actually one of the main reason why this type of speaker recognition systems are not used in actually practised because they're afraid
but people just that's moving them not very reliable nonsense so therefore that a lot of issues of
how to detect the study but actually train a different type of model different type of system
to separate the attacks from their real scores you learn some other parameters of the signal to figure this out
and then when you should try that then you have to joint these together to systems you have attack detection system and
occasions just me join them together one after another altogether in
parallel or something so so for back for example
so we we have basically yeah i go go from ripple public report because just
a huge which is so if if you have a this type of uh
uh also g. m. m. m. f. c. c. baseball it's a like a quite a standard
standard way to do it so this is another tool chain you can see here
and uh i so now so this is this is basically similar stuff
just a bit different a different type of system and modelling
so now i want to just show the uh example when you join these two systems together so the goal is here
you have this is before you have attacked detection in education system your attacks and not being recognised as attacks
and now you add a affected actions is that together with the education system and now
you have clear separated your good users the the nice guys from the all the
not like space nicely so you have a separation please oppression in the score so you your your system is it
very fine people correctly and this isn't attacks so now getting bit by what do you
need to do not have two different systems right is that the show you
two different put chains so they're a little bit of a problem here then when you create a big
uh we should join this thing you have to draw something like this
because they have to mainly do it with well click and click but you know
double droll these blocks the comment this is like a circuitry
of the big go liberate a due process or something
it's like the g. or something
so this is a bit about sixty plus blocks and the you better not make mistakes
when you do that
because you have to correct them later so you see right so so you do that but
you well the good part here i mean a a scientific experiment requires effort right there's
no way you can escape that i mean if you want to get your p. g.
or get a grant or whatever or paper probably just you have to put effort
so i mean the good thing about these type of approach is you do it once and then you can
basically just the forget about it and talk for years to come and
it's going to be there for you to show and impress everybody
goes maybe also i dunno depends on the girls and basically and and it's verifiable so
people can actually take that explain now that i did once and use it
change the parameters change a little bit block a little blocks and see if it actually
change something databases like there's like five different databases you can put proprietor the
basin around the same topic spread is a state of the art experiment
as to the the uh things you use you can you can for this uh also
told changes huge but you can add small blocks there if you want an improve
the system and the the different type of things so it's it's requests uh it's it's
quite some effort but it's it's stages you for ever so which is very nice
um and you have a public reported that cannot nobody can delete and to you know yep stops paying for the for the costing
let's call that never happens until i'm still alive just um okay
so i i hope it gives you some type of
idea of uh you know how to do this type of experiments like that actual expensive the published can be problems
so you design so that once and but then you you basically just you know you're done right after that you
do that you can repeat you can surround different data can we defy you can change shy compare results
like a likely the boards and all those things if you people interested is not that hard and you can see how actually down like the each other
if you click you have the court exactly like what is going on so
we can really clearly see what what was happening it's very nice
so this is there's was like uh this uh paper that is actually basically it's uh
this results will present in the papers another think is a good for the researchers
when you have a paper in the conference is limited by certain number page like four five six ten right and then after
that it's not paying i mean even journal you can have sixteen page button to pay a two hundred bucks per page
which is kind of sad uh when you have a lot of things to share so these
actually allows you to sneaking then he give much more information than the paper pass
the format allows and you you win because this report that just look uh you know i social so we can can
protect you can put another five papers inside of text and the people can read all the details they want
and the code and everything so it's like much more each information than just the ten pages paper
and you extend your basically page size unlimited early if you want if you're like the right yep that's all

Share this talk: 

Conference program

Sébastien Marcel, Senior Researcher, IDIAP, Director of the Swiss Center for Biometrics Research and Testing
24 March 2017 · 9:17 a.m.
Keynote - Reproducibility and Open Science @ EPFL
Pierre Vandergheynst, EPFL VP for Education
24 March 2017 · 9:20 a.m.
Q&A: Keynote - Reproducibility and Open Science @ EPFL
Pierre Vandergheynst, EPFL VP for Education
24 March 2017 · 9:54 a.m.
VISCERAL and Evaluation-as-a-Service
Henning Müller, prof. HES-SO Valais-Wallis (unité e-health)
24 March 2017 · 11:35 a.m.
Q&A - VISCERAL and Evaluation-as-a-Service
Henning Müller, prof. HES-SO Valais-Wallis (unité e-health)
24 March 2017 · 12:07 p.m.

Recommended talks

An introduction to TensorFlow
Mihaela Rosca, Google / London, England
26 Nov. 2016 · 2:01 p.m.