WORKSHOP ARRANGING COMPETITIONS IN SPEECH CAPTURING

1
WORKSHOPARRANGING COMPETITIONSIN SPEECH
CAPTURING
2
CHANGES IN PRAGUE

• Jury structure was changed
• jury president, coordinators, power force,
  assistants
• Regulations were restructured
• Competitions were controlled more effectively
• translations, dictations, valuations
• Result lists were produced faster
• automation added
• Transparency added by publishing the texts and
  dictations
• Reports of competitions published (together with
  interviews of champions)
• Analysis of organizing competition written (base
  for manual in the future)

3
MANAGEMENT IS COOPERATION
4
PROCESS MANAGEMENT

• Timetable
• - status reports
• File templates
• contents
• naming standards

5
TIMETABLE/PRAGUE
Translation
Dictation
PRAGUE
GER
FR
SP
CHE
CRO
CHINA
HUN
ITALY
FIN
Text choice
Text modification
Lang. check
21.1.
22.2.
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
June
July
6
TIMETABLE COMPARISION
PRAGUE
Text choice
Text modification
Lang. check
Translation
Dictation
21.1.
22.2.
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
June
July
Aug
Peking
Hyphnations Final texts
Original transl.
Jury check
Public comments
Dictation
Text choice
Text mod.
Lang. check
Retranslations
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
June
July
Aug
7
CONTROL OF COMPETITIONS
Translation Translation Hyphenation Original text Modified text with modification marks Hyphenetated text (Latin alphabet text) Competition text
Dictation Comparision to text Speed Original dictation Spreadsheet to calculate the speed Competition dictation
Correction Valuation Calculation of points Competition papers Spreadsheet to calculate the points
8
AVAILABLE TOOLS
Translation Guidance Document templates Example documents Practical guidance to calculate the syllables Spreadsheet to calculate the hyphenation
Dictation Guidance Audacity
Correction Rules Spreadsheet to calculate the points
9
SYLLABIFICATION

• Option 1 use a syllabification service
  http//www.wordscount.info/hw/syllable.jsp
• Option 2
• Syllabification macro
• Macro to calculate syllables
• About the macro
• depends on the used system (even Word-version)
• one predefined macro available in the package
• creating an own macro is, however, easy see
  document "Practical guidance how to count
  syllables easily"

10
SYLLABIFICATION EXAMPLE

• Divide the text into syllables e.g. by putting
  "/" after each syllable. (Not "-" because it is a
  common mark inside the text)Syllabification
  macro changes "-" marks to "/" automatically
• Calculate syllables by
• - selecting the area
• - by findreplace (find "/" and replace "/")
• Word gives the number of replacements
• At the end you may delete the "/"-marks by
  findreplace (find "/" and replace "/")
• Be prepared to syllabificate the text several
  times

11
DIVISION INTO MINUTES
Divide each minute into 15 s sections. There is
an Excel-file to help. Do the final work after
the original text has been accepted! Each section
(not each minute!) from a new sentence (can be
different than in the basic text)
12
RECORDING

• Audacity
• - download
• http//audacity.sourceforge.net/download/windows
• - version 1.2. or 1.3.
• - wav or mp3
• Good equipment
• Good environment
• Enough time all the text including the practice
  dictation the same day

13
MP3 file

• If mp3 is used, then download LAME MP3 separately
• - download http//lame.buanzo.com.ar/

14
Toolbox

• pause

play
stop
input volume
output level meter
input level meter
output volume
time elapsed
record
zoom in/out
cut/copy paste
15
Zooming the recording
16
BASICS

• File - New
• creates a new project
• project is a working area not final recording
• you may open a saved project and continue working
  with it
• File - Export
• creates the final recording (choose then the file
  type mp3 or wav)
• remark that with other tools you may create
  wmv-files (do not use)
• Edit - Preferences -Audio/IO
• choose Channels 1 (Mono)
• Edit Preference - Quality
• - choose Default Sample Rate 22050 Hz

17
SELECTION

• The actions effect on a selection.
• Select an area by pressing first pushing left
  button of the mouse and then moving the cursor to
  the right.
• From Edit you get a list of actions you may
  choose (e.g. delete/cut/paste)
• From Generate you may add Silence
• From Effect you get useful tools, like
• Normalize
• Change tempo

18
COMPARISION OF ORIGINAL DICTATIONS/PRAGUE
19
NORMALIZATION
- 5 (second example)
20
NORMALIZATION

• Select an area
• Choose Effect Normalize...

21
Changing tempo

• Select an area from the beginning to the end of
  the minute.
• Choose Effect Change Tempo...

length of selection is 58.01 seconds
change it to 60 seconds and press OK
22
EXCEL SHEET TO VERIFY THE SPEED

• This is a tool used by jury.

Running time from beginning of recording by
Audacity
23
EVALUATION FORM

• Used during the competitions
• manages the competitors
• manages the evaluation process
• calculates the points and transcription time
• creates the competition results
• gives some statistics (in the next version)
• Use of the form strongly recommended
• final results given fast
• increases the reliability of the results

24
EVALUATION FORM
Correctors evaluate
Jury evaluates
Pink fields can be filled
Calculates automatically results
Transcription time given by the correctors
If these are given in order they are in the
transcription, then it is easy to verify that the
amount of 4 and 1 point mistakes is OK
Mistakes minute by minute
Shows rejected minutes in red
25
DATABASE STRUCTURE
Competitor follow-up
Competitor forms-lang1
Competitor forms-lang1
Evaluation forms/lang1
Competitor forms/lang1
Button "Accept"
Evaluation form1/lang1
Competitor form1/lang1
Package given to the correctors Only the language
to be corrected
26
WORKSHOPARRANGING COMPETITIONSIS COOPERATION