Following the effects of voice training using computer phonetography

1
Following the effects of voice training using
computer phonetography
Voice Quality Systems

• Peter Pabon and Rob Stallinga, Royal
  Conservatory, Den Haag, the Netherlands
• Pevoc IV

2
Objectives
Since three years, the recording of a
computerized phonetogram is a standard procedure
for all singing students of the Royal
Conservatory

• (1) To support the singing teachers and students
  with objective information on their voice range
  and registers.
• (2) To do scientific research.

3
Material

• Singing department is the largest in the Royal
  Conservatory with ca. 181 students spread over
  14 years (80 female).
• A first recording is made at the start of the
  study,
• a second after 2 years of training, and
• a third is scheduled for the end of the study.
• Motivation problem ca. 25 of the students show
  up for a first recording.
• We now managed to get recordings of 76 students
  (42).
• From 9 students there is a second recording
  available (after 2 years).
• The group is very diverse, they differ in
  entrance level, training objectives, voice type.
• 90 of all recordings are female voices classical
  music style.

4
What to select, what is interesting to show?

• Typical reduction of the maximum levels reached
  in chest/modal register (part 1).
• Individual results / effects of voice training
  (part 2 of this talk).

5
A computer phonetograph with additional voice
quality parameters allows you to get a detailed
view on dynamic voice/register use.
An additional acoustical parameter called the
crest factor is recorded.It is a marker for
spectral slope The values of this parameter are
mapped using a green/yellow/red color scale. The
red area generally reveals the phonations in
chest/modal register.
6
Voice quality information in the phonetogram
remarks

• Registers and internal boundaries can be
  localised by two features
• The displayed voice quality parameter (crest
  factor).
• The trace that the cursor shows in the visual
  feedback while singing.
• Implicit / hidden but very strict practical rule
  
• The voice sticks to the vibrational mode
  (register) it is in, once started, the voice
  tends to keep the register chosen at voice onset,
  and can only be pushed to another vibrational
  mode in an abrupt transition. Singing very soft
  works fine to conseal such a transition.
• There is an overlap area between the two
  registers.
• Covering this area using two different register
  qualities can blurr the recorded quality in the
  transition zone.

7
Typical reduction of the maximum levels reached
in chest / modal register (of female, classical
trained voice) PART1
8
Loud phonations (above ca. 100 dB) in Modal/Chest
register are not done !
9
It can be assumed that most singers could have an
extra extension at the high end of their
chest/modal register like the sketched
10
Some singers can (still be persuaded to) bring
chest voice quality up to high Fo and push their
modal register up in Fo and reach extreme sound
levels.
Oops!
11
Equalization / Mixing

• Singing loud in pure chest/modal register is not
  necessarilly harmfull or wrong, but unwanted in
  classical music style.
• Aim equalization of voice quality over the
  musical range.
• Avoiding clear register transitions (smoothing).
• The term mixing is also used a specific
  colouring/articulation where the typical quality
  of the higher register is brought in at low pitch
  (in low register) and vice-versa.
• Mixing in the lower register has its price in the
  decrease of the maximum sound level that can be
  reached.

12
While still at relatively low Fo (where usually
chest voice is powerfull), a mixed quality is
searched and retained upward with Fo, leading to
a smooth quality change.
13
Additional remarks

• For the singer the equalization/mixing can be
• a natural feature (is there another way then?) or
• a still unknown (trainable) feature of the voice,
  or
• an acquired singing technique (you trained hard
  and clench to it), or
• a conscious choice (technique is mastered and
  applied on demand).
• The reduction in sound level due to equalization
  / mixing all comes from the focussing on the
  higher part of the range.
• Let us not forget a social aspect, for the female
  voice it is not considered nice to phonate loud
  in modal register.
• So, if it comes to the physiological extreme, the
  maximum levels in modal/chest register are often
  underestimated. The reason why is not retraceable
  from the recordings. We have to keep this in mind
  when evaluating the measurement results.

14
Part 2Individual results / effects of voice
training.
15
Front11/06/98Back 15/12/99 7726 (F)
Front15/12/99 Back 11/06/98
16
Front18/06/98Back 11/04/00 7744
(F)Front11/04/00Back 18/06/98
17
Front11/05/98Back 17/04/00 8029 (F)
Front17/04/00Back 11/05/98
18
Front05/11/98Back 08/05/01 8005
(M)Front08/05/01Back 05/11/98
19
Front05/11/99Back 17/04/01 8455
(F)Front17/04/01Back05/11/99
20
Front05/11/99Back 17/04/01 8455
(F)Front17/04/01Back05/11/99
21
Front05/04/00Back 20/03/01 8358 (F)
Front20/03/01Back 05/04/00
22
Front21/10/97Back 04/07/01 6930 (F) r.g.
Front04/07/01Back 21/10/97
23
Front30/09/97Back 30/06/98 7097 (F) r.g.
Front30/06/98Back 30/09/97
24
Front04/07/01Back30/09/97 7097 (F) (same
as previous)Front 04/07/01Back30/06/98
25
Conclusions

• Computer phonetography with additional voice
  quality parameters allows a very detailed view on
  the voice range and registers.
• Visualise the choices with mixing / equalization.
• Mixing in the lower register has its price in the
  decrease of the maximum sound level that can be
  reached, but in many cases the trade-off is not
  even noticed by singer/teacher or audiance.

26
Conclusions (2)

• The effects of voice training show in the range,
  shape and color/quality.
• The bottom end of the musical range and the
  boundary line between the registers seem to be
  the invariable elements.
• Growth is generally a sort of expansion at the
  high/loud end of the register (low or high
  register).
• Extensions at high SPL in the low register not
  seldom coincide with an upward shift of the
  threshold of phonation.
• Phonetogram body building not necessarilly means
  getting a better voice.

27
Following the effects of voice training using
computer phonetography
Voice Quality Systems

• Peter Pabon and Rob Stallinga, Royal
  Conservatory, Den Haag, the Netherlands
• Pevoc IV

28
(3) A dip with the start of mixing and
sometimes a knick on the Fo where mixing is
no longer effective.
29
The first dip with the start of mixing
In one phonation a smooth transition can be made
to a (generally softer) mixed voice quality and
back. There is never a sudden jump in quality.
So, the first dip should not be interpreted as a
register break, but as an articulatory
(colouring) effect.
30
knick on the Fo where mixing is no longer
effective
There is almost no quality change when climbing
in Fo and passing the register boundary. However,
the higher notes are reached only after a new
voice onset (register change) and only then the
extreme SPL levels are reached.
31
(4) Strong power increase with Fo on entering
full Falsetto/Head voice (second plateau).
32
Front10/11/99Back 21/02/01 7932 (F)
Front21/02/01 Back 10/11/99