wim de vilder – filter : verwijderen van ademhalingsruis uit spraaksignalen probleemoplossen en...
TRANSCRIPT
Wim De Vilder – filter : verwijderen van ademhalingsruis uit
spraaksignalen
Probleemoplossen en ontwerpen, deel 3
Problem Statement
Newscasters present the news with a very quick
tempo Between two sentences they require a large breath
Can be a distraction for the viewers
Tempo can be so fast that the viewers cannot
understand
Problem Statement
Wim De Vilder : an example
Problem Statement
Examples : Different pitch = Time-Scaling
Original Signal
Fast Version
Slow Version
Pitch Corrected
Problem Statement
The Wim De Vilder-filter and Time-Scaling ? Automatically (in real-time) know the difference between
speech and breath
Allow speech to pass
Slow down signal without distorting pitch
Digital signal processing = key to the problem Characteristics (features) extracted from the acoustics
Difference between speech and breath (classification)
Pitch extraction from audio signal
Problem Statement
The Wim De Vilder-filter and Time-Scaling ? Automatically (in real-time) know the difference between
speech and breath
Allow speech to pass
Slow down signal without distorting pitch
Digital signal processing = key to the problem Characteristics (features) extracted from the acoustics
Difference between speech and breath (classification)
Pitch extraction from audio signal
Problem Statement
The Wim De Vilder-filter and Time-Scaling ? Automatically (in real-time) know the difference between
speech and breath
Allow speech to pass
Slow down signal without distorting pitch
Digital signal processing = key to the problem Characteristics (features) extracted from the acoustics
Difference between speech and breath (classification)
Pitch extraction from audio signal
Problem Statement
The Wim De Vilder-filter and Time-Scaling ? Automatically (in real-time) know the difference between
speech and breath
Allow speech to pass
Slow down signal without distorting pitch
Digital signal processing = key to the problem Characteristics (features) extracted from the acoustics
Difference between speech and breath (classification)
Pitch extraction from audio signal
PlanningTeam 1 (Wim De Vilder Filter) Team 2 (Time Stretching)
30/9 Problem Statement : First Group Meetings
Voice activity detection : features Voice activity detection : features
07/10 Voice activity detection : classificatie Sample rate change / framing
Feature : Zero-crossing rate/periodiciteit Time Stretching : Overlap Add Synthesis (OLA)
14/10 Feature : spectrale energie Time Stretching : OLA
Feature : spectrale energieTime Stretching : Synchronous Overlap Add Synthesis (SOLA)
21/10 Schrijven tussentijds verslag SOLA : Time Domain Auto Correlation
Feature : LPC Pitch Synchronous Overlap Add Synthesis (PSOLA)
25/10 Deadline mid-term report
28/10 Feature : cepstrale energie Pitch Detection : Zero Rate Crossing
4/11 Feauture : tijdsinformatie Pitch Detection : Modified Zero Rate Crossing
Features : combinatie Pitch Detection : Auto-Correlation Techniques
12/11Bayesiaanse classificatie + Gaussian Mixture Models PSOLA : ImplementationBayesiaanse classificatie + Gaussian Mixture Models PSOLA : Implementation
18/11 Real-time implementatie in Simulink
Real-time implementatie in Simulink
25/11 Real-time implementatie in Simulink
Real-time implementatie in Simulink
27/11 Deadline infobrochure
02/12 Real-time implementatie in Simulink, preparation for demo
Real-time implementatie in Simulink, preparation for demo
9/12 preparation for report, presentation
16/12 Presentation
Praktisch
2 sessies per week (seeTijdstabel P&O3) Monday 13.50-18.00
Thursday 13.50-18.00
2 hours interaction per week
E-mail for questions/problems!