PHONEXIA

movedearΤεχνίτη Νοημοσύνη και Ρομποτική

17 Νοε 2013 (πριν από 3 χρόνια και 8 μήνες)

160 εμφανίσεις









PHONEXIA

CAN I HAVE IT IN WRI
TING?












MATERIÁLY PRO UČITELE




Our company was established in year 2006. We focus on speech technologies. We develop speech technologies such
as voice recognition, speech recognition what means speech
transcription and also key words spotting and language
identification, gender identification. The voice recognition and language and gender recognition is language
independent so we focus on US market, European market but as well as Czech market.

The spee
ch recognition is quite general term. We can speak about voice recognition or speech recognition in terms
of transcription. If students want to focus on this, this is lot of mathematics, statistics and studying of signals.

First is Czech obviously, the sec
ond is English. The English language version is trained on non
-
native speakers, which
is quite interesting for the world because most of the engines are trained on native speakers therefore it fails with
non
-
native speakers. The English version contains 50
.000 words but the Czech version contains one million of Czech
words. This is given by the language.

Each technology can be used separately or together with other engines. I would start with the voice recognition or,
let’s say, speaker recognition. You c
an use it in a task, let’s say, voice as a password, for example. You can call
somewhere to some call centre and the call centre can recognize yourself by your voice. The second easy
-
to
-
use
application is speech transcription. Everyone can imagine dictatio
n that the voice will be transcribed automatically to
text. But also it can be used for multimedia indexing. For example, Internet is full of video and audio files, so you can
use the technology to transcribe automatically video and audio content and there
fore it will be useful for search
robots, search engines to appear it in the list of search results.

I would mention also the project called Přednášky.com, you can find it on Internet. This is a project of Brno
University of Technology. They recorded the
lessons from Bachelor degree studies and every user can type any word
-

Furrier transformation, for example and system will find this word or this phrase in video content and give you the
research results.

The second example might be Superlectures.com, wh
ich is quite similar project. But this project focus on conference,
on organizator of conferences. So the conference can be recorded to video or audio files and these files can be
transcribed automatically so it would be searchable for any user. You can fi
nd it on Internet.



Naše firma byla založena v roce 2006. Zaměřujeme se na řečové technologie. Vyvíjíme řečové technologie
,

jako je
rozpoznávání hlasu, rozpoznávání řeči, což znamená přepis řeči a vyhledávání klíčových slov, identifikaci jazyka a
pohlaví. Rozpoznání hlasu a p
ohlaví nezávisí na jazyce, takže se soustřeďujeme na trhy v USA, v Evropě, ale taky v ČR.

Rozpoznání řeči je docela obecný pojem. Dá se mluvit o rozpoznání hlasu nebo řeči co do přepisu. Jestli se studenti
chtějí tomuto oboru věnovat, je to spousta matemat
iky, statistiky a studium signálů.

První je pochopitelně čeština a angličtina je druhá. Anglická verze se nacvičuje na nerodilých mluvčích, což je docela
zajímavé pro svět, protože většina vyhledavačů je uzpůsobena rodilým mluvčím, a proto u nerodilých sel
hává.
Anglická verze obsahuje 50.000 slov, ale česká verze obsahuje milión českých slov. To je dáno samotným jazykem.

Každá technologie se dá použít samostatně nebo v kombinaci s jinými vyhledavači. Začal bych s rozpoznáváním hlasu
nebo, řekněme, rozpoznáv
áním řeči. Dá se využít jako hlasové heslo, například. Můžete se dovolat do nějakého call
centra a oni vás tam podle hlasu poznají. Další uživatelsky snadná aplikace je přepis řeči. Každý si umí představit
diktát, kde se hlas automaticky přepíše do textu.
Také se to dá využít na indexaci multimédií. Internet je například
plný audio a video souborů, takže tu technologii můžete použít k automatickému přepisu jejich obsahu, což je
užitečné pro vyhledá
vací roboty a vyhledavače, které

jej začlení do výsledků hle
dání.

Také bych zmínil projekt s názvem Přednášky.com, který můžete najít na internetu. Je to projekt Technické
Univerzity v Brně. Zaznamenali přednášky bakalářského programu a každý uživatel může napsat jakékoli slovo


Furrierova transformace, například,

a ten systém toto slovo vyhledá v obsahu videa a předloží vám jej ve výsledcích
hledání.

Druhým příkladem by mohlo být Superlectures.com, což je docela podobný projekt. Ale tento projekt se zaměřuje na
konference a jejich organizátory. Takže konference s
e dá zaznamenat do audio či video souboru a tyto soubory se
dají automaticky přepsat, takže budou na internetu vyhledatelné pro každého uživatele. Dají se najít na internetu.





write
check

discuss

share

match

ans
wer

find
info


EXERCISE 1













Discuss and share your answers to the following questions:

1.

When you have English lessons listening to spoken English, does a transcript of the text help you while
listening? How so? Are there times when it

is difficult and times when is
it easy for you? Explain under what
circumstances English speech is difficult to understand (e.g. intonation, accents, native speakers, etc.)

2.

If you watch a film in English, do you usually watch it with the subtitles on? Do
you ever watch the news with
closed caption transcriptions? Can reading text while someone is talking interfere with your experience of
other communications such as gestures, or facial nuances?

3.

Have you ever used any automatic speech or voice recognition s
oftware? Have you ever used an Internet
site that has voice to text or text to voice transcription capability available? If so, describe your experience.












EXERCISE 2




Are these statements TRUE or FALSE?


The technology of the co
mpany is able to recognize if a speaker is a man or a woman.


They cannot sell their product abroad because it is only relevant for the Czech market.


If you want to study the field of speech recognition, you should expect a lot of math and statistics.


Searc
h engine software trained on non
-
native speakers are useless for native speakers.


The technology enables you to use your voice as a password.


If you are able to transcribe video and audio files, they can be included in search results of search engines.


Sup
erlectures.com serves mainly bachelor programme students.

EXERCISE 3




Listen and fill in the missing text.

1.

Our company was
established

in year

. We focus on
speech

technologies. We develop


technologies such as voice
recognition
, speech
recognition
what
(it)
means
(for)
speech
transcription

and
also key words
spotting
and language identification,
gender

identification.

2.

The


language version is trained on


speakers, which is quite interesting for the worl
d because
most of the engines are trained on
native speakers
therefore it
fails
with


speakers. The


version
contains
50.000 words but


3.

You can use it in a
task
, let’s say, voice as a
password
, for example.

4.

For example,
(the)


is full of video and audio
files
, so you can use the technology to transcribe


video and audio
content

and
therefore
it will be useful for search

,
search engines
to appear
it
in the
list of search

.

5.

I would
men
tion

also the project called Přednášky.com,
(that)
you can find
it

on Internet. This is a project of
Brno University of Technology. They recorded
the

lessons from
Bachelor

degree studies
and every user can

-

Furrier transformation,
for example and
(the)
system will


this word or this phrase in


content
and give you the


results.

6.

The second example might be

, which is quite
(a)
similar project. But this project focus
(es)


on


, on (the)

of conferences. So the conference can be

to


or


and
these files can be


automatically so it would be


for any us
er.



EXERCISE

4













Match the following English expressions with their Czech equivalent.

1. SEARCH ENGINE

A) ZALOŽIT


9. MENTION

I) ÚLOHA

2. CONTAIN

B) OBSAH

10. BACHELOR

J) OBSAHOVAT

3. CONTENT

C) ŘEČ

11. ESTABLISH

K) ROZPOZNÁNÍ

4. TRANSCRIBE

D) SELHAT

12. SPEECH

L) DOHLEDATELNÝ

5. RECOGNITION

E) HESLO

13. SPOT

M) POHLAVÍ

6. GENDER

F) SPATŘIT, VŠIMNOUT SI

14. TASK

N) ZMÍNIT

7. FILE

G) SOUBOR

15. PASSWORD

O) VYHLEDAVAČ

8. SEARCHABLE

H) BAKALÁŘ

16. FAIL

P) PŘEPSAT


PHRASE BOOK














Use t
he vocabulary list

to develop questions you could ask the Manager in a mock interview.

With a partner, take turns being the Manager. Role play asking and answering your questions.

CUSTOMIZATION

The programming capability to adapt technological systems to
user demands.

INTERFACE


The technology by which sound, text, or visuals are represented and transmitted across



platforms, devices, and systems of use.

MOBILE DICTATION

Software for mobile phones with speech recognition capability to convert sound into

text.

MULTI
-
MODAL


Audiovisual term which refers to different kinds of media.

TO MENTION


To speak about something. The speaker is considering what information should be given



within the context of the topic.

PASSWORD


A set of letters and or numbers
in combination that is chosen as a private pincode to open



access to a particular file or data within a technological application which operates.

SMART TVS


Intelligent systems that integrate data processing features such as speech recognition




soft
ware for automatic transcription of voiced text with other television viewing options.



EXERCISE 5












TRANSCRIPTION CONTEST

Watch the first 60 seconds of the video twice. While you watch, try to transcribe what you hear. Then compare
your

results with the video script. Who managed to write most of the text without mistakes? What strategy
proved to be the most efficient?

Discuss with your classmates

EXE
RCISE 6




GRAMMAR AND SYNTAX TEST

1.

The speaker in the video said: “
But
also

it can be used for multimedia indexing.


a.

The position of the word
also

in the sentence is wrong. Which of the following sentences are
correct? A) But it
also

can be used for multimedia indexing. B) But it can
also

be used for multimedia
indexing. C) But i
t can be
also

used for multimedia indexing?

2.

Look at the following sentences.
Which of them uses the word
also

i
n a correct place?

A)
I have a brother also.


B)
I have also a brother.


C)
I also have a brother
.

3.

Fill in the rules.

A) We put
also

between

t
he subject and the verb.

B) If there are two or more verbs in a sentence, we put
also

between the two.

4.

Insert
also

in
to

the following sentences.

A) I can be the best student in the class.

B) I lived in Prague.

C) My computer

has been st
olen.

D) She h
as a lot of time at the moment.

5.

Make
sentences
using the
content of the video with the following vocabulary words and phrases:
voice
recognition,
searchable, gender, transcribe, establish, speech, password, interface, multimodal
, and
customi
ze
.





WHILE YOU WATCH
EXERCISE 2


1.
TRUE, 2. FALSE, 3. TRUE, 4. FALSE, 5. TRUE, 6. TRUE, 7. FALSE

EXERCISE 3

1.

Our company was
established

in
(the)
year
2006
. We focus on
speech
technologies. We develop
speech

technologies such as v
oice
recognition
, speech
recognition

what
(it)
means
(is)
speech
transcription

and
also key words
spotting

and language identification,
gender

identification.

2.

The
English

language version is trained on
non
-
native

speakers, which is quite interesting for

the world
because most of the engines are trained on
native speakers

therefore it
fails

with
non
-
native speakers
.
The
English

version
contains

50.000 words but…

3.

You can use it in a
task
, let’s say, voice as a
password
, for example.

4.

For example, Inte
rnet is full of video and audio
files
, so you can use the technology to transcribe
automatically

video and audio
content

and
therefore
it will be useful for search
robots
,
search engines

to appear
it

in the list of search results.

5.

I would
mention

also t
he project called Přednášky.com,
(that)
you can find
it

on
(the)
Internet. This is a
project of Brno University of Technology. They recorded
the

lessons from
Bachelor
degree studies and
every user can
type any word

-

Furrier transformation, for example an
d
(the) system will
find
this word
or this phrase in
video

content and give you the
research

results.

6.

The second example might
be
Superlectures.com

which is quite
(a)
similar project. But this project
focus
(es)

on
conference(s)
,
on (t
he)
organization

of

confer
ences. So the conference can be
recorded

to


video

or
audi
o files

and

these
files

can be transcribed automatically so it would be
searchable

for any
user.


AFTER YOU WATCH

EXERCISE 4


1O,

2J,

3B,

4P,

5K,

6M,

7G,

8L,

9N,

10H,

11A,

12C,

13F,

1
4I,

15E,

16D








AFTER YOU WATCH

EXERCISE 6


1.


B

2.

C

3.

A) We put also
between

the subject and the verb.


B) If there are two or more verbs in a sentence, we put also between the
first

two.

4.

A) I can ALSO be the best student in the class.


B)
I ALSO lived in Prague.


C) My wallet has ALSO been stolen.


D) She ALSO has a lot of time at the moment.

5.


Phonexia works with the development and use of
voice recognition

technologies
.



The use of
searchable
keywords can be processed by text or spee
ch.



New voice recognition software is able to distinguish between
gender
s.



The possibility to
transcribe
lectures and conferences greatly disseminates the content to more than
just the group of people who attended.



New technologies
establish
new poss
ibilities for interactivity.



The voice recognition software can recognize
speech
for text or a command.



Access to the company’s database requires a
password.


It is possible to use speech recognition software inside of television sets for a new kind of

interface.


Today’s technologies are becoming integrated
and

multimodal
.


Phonexia offers its customers the possibility to
customize

their technology to the

users’ needs.