The wonderful journey to the VRP islands - PowerPoint PPT Presentation

1 / 20
About This Presentation
Title:

The wonderful journey to the VRP islands

Description:

The wonderful journey to the VRP islands. 2nd COST Workshop, Aachen ... Narrow band spectrum (0..22 kHz) plus --Singing Formant level --Proportional power Fo ... – PowerPoint PPT presentation

Number of Views:37
Avg rating:3.0/5.0
Slides: 21
Provided by: Peter249
Category:

less

Transcript and Presenter's Notes

Title: The wonderful journey to the VRP islands


1
The wonderful journey to the VRP islands
  • 2nd COST Workshop, AachenMay 12-14, 2008Peter
    Pabon,
  • Royal Conservatory, Institute for Sonology The
    Hague
  • Utrecht University,
  • Voice Quality Systems, Utrecht

2
Main focus Real time application of scientific
models. Musical / Clinical
Sonologys new toy, the 192 speaker Wave Field
Synthesis system.
Voice Profiler
3
1983
Analog To/SPLfilters switching AGC -gtPDP 11
1987
DSP TI32010 4k mem! To/JItter/SPL dual amp.
-gtPC802863 systems
1992
FP DSP M96000 80486 FGWIN Dual Amp. 4
systems
FP DSP M96000 80486 FGWIN Dual Amp.
Phonetgraphs - timeline
1995
I-DSP TI320C25 pentium FGW Laryngograph
Dual Amp. 25 systems
2002
Voice Profiler / Dual micr. Headset / Self
calibr. amp. 150 systems
VRP Interpetation scheme
2008
4
Current parameters in Voice Profiler version 4.2
5
Look at one perturbation parameter Jitter in
the VRP
jitter is an indicator for pathologic voice
quality.
6
(No Transcript)
7
Perturbation in the VRP Jitter Some results, an
overview
High jitter
Low jitter
8
Time to see the islands!
9
Jitter in the Phonetogram / VRP
  • Pathologic voices, more diversity, smaller,
    irregular shape, disconnected register areas.
  • The large VRP areas with high jitter are
    typically found with the normal voices
  • Pathologic voices regularly show VRPs with
    mainly low jitter values.
  • Extremely pathologic voices, cover very small
    areas, but still jitter follows the pattern that
    is also observed for the normal voices, change
    happens more faster with SPL.

10
Jitter in the Phonetogram / VRP
Increased jitter in the VRP is typically not a
clear indicator of pathologic voice quality!
11
Jitter in the Phonetogram / VRP
In many cases, the area with high jitter values
in the VRP is larger with normal voices then with
pathologies!
12
Jitter in the Phonetogram / VRPA possible
explanation
It seems that a sign of a healthy voice is the
freedom to vary more parameters independently,
this includes the regularity of its vibration
cycle.
13
Jitter in the Phonetogram / VRPAnother possible
explanation
Pathologic voices are more binary, have startup
problems and thus higher starting SPL and thus
express jitter over a smaller VRP area.
14
Policy in relation to customers that expect
pathology gives high jitter
We admit that Jitter and Shimmer were only
installed to beat any competition. Indication
that their direct relation to voice pathology was
problematic was clear from the beginning. Patterns
were highly reproducible and personal, and
thereby, may be still clinical relevant. Voice
or system? Extensive retesting with synth. Clear
SPL dependence also strong vibrato interference
convince that the extraction scheme is detecting
what it should detect.
15
Was it a good idea to put jitter in?
  • The Jitter parameter is very effective in
    clinical practice as turbulent noise measure for
    insufficient glottal closure in falsetto voice
    (mechanism II).
  • Its application as roughness/pathology indicator
    remains questionable, users are instructed to
    ignore the high jitter values in modal register
    (mechanism I).

16
Final conclusions
  • SPL and register are the main separators.
  • Only with material related to absolute SPL you
    can do a proper evaluation of the relevance op
    voice quality parameters.
  • Highest jitter values in the VRP are found with
    normal voices, low SPL, modal register, around
    the speech area.
  • The principle seems to be If the system is able
    to vibrate, its tendency is to vibrate periodic.

17
Some personal statements
  • One typical jitter value for one voice is
    nonsense.
  • Assessment schemes that are based on reduction
    are bound to fail, it is too early to aim for
    reduction.
  • Jitter shimmer are largely overvalued as
    clinical useful parameters!

18
Thank you !
Do not forget the guide!
19
Bad excuses
  • This is probably with only one typical voice
    disorder, they left out the .. or only took the
  • His jitter measure is not as good as mine, my
    jitter parameter would immediately show that.
  • This is just a commercial device using
    old-fashioned ideas .

20
Building of norms, variability and personal
characteristics.
Write a Comment
User Comments (0)
About PowerShow.com