A voice production element G J Verkerke
Here are presented the principles and the implementation of several prototypes that could repalce vocal folds and help laryngectomised patients to produce speech again.
Introduction Voice production is characterized by 3 main elements : an actuator, a sound generator and resonators.
Fig 1 : Schematization of voice production
After laryngectomy, the patient has no larynx anymore. His trachea is open at the basis of he neck so as to allow him for breathing. To keep on speeking, laryngectomized people can use their oespohagus to produce sounds. But these produced voices are not very natural, in particuar for women who have a higher pitch than men. Moreover, patients have to learn a new phonation mode, that is nor the case of the new prototypes that will be presented here
Fig 2 : Laryngectomized man
1) Review of methods envisaged to repalce vocal folds In 1929, Riesz presented an artifical larynx that consisted in a flexible tube pluged to the trachea at it opening on the neck. The fondamental frequency was set up manually.
Fig 3 : The western electric artifical larynx of Riesz.
Then, an other solution to replace vocal folds was found by pitting a buzzer on the neck. The speaker could change his mouth and tongue configuration to amplifie this vibration and produce different vowels or sounds.
Fig 4 : Speech vibrators and buzers.
2) Problematics The goal is to replace the vocal folds by a little system. There are some requirements to solve this problem: The geometry of the system sould be a shunt valve. As for the produced sound, the fundamental frequencies shoud be of about 110 Hz for the males and of about 210 Hz for the females, with a sound pressure level of 65 dB to 80 dB at 30 cm. The frequency-variations shoud be of +18% and -6%. The energy source should cover a range of 0.4 to1.5 kPa and 180 to 320 ml/s . As for the power spectrum, it should have a decay of 10-12 dB/octave To resolve this conditions, we have envisaged 3 principle solutions. 3) Design of a flexible lip One first solution was the design of a flexible lip, constituted by 6 parts . It looks like a sort of tail fixed to the vocal tract. When the air flow increases in the vocal tract, it should generate autooscillations of this flexible lip.
Fig 5 : Principle of the flexible lip.
Fig 6 : Representation of the flexible lip in the calculation domain
Fig 7 : Simulation of the autooscillations of the flexible lip
Fig 8 : 2½D modelisation of the flexible lip in a 3D flow
Fig 9 : Results of the numeric simulation
This principle was then implemented and validated. The first prototype had a 5 mm diameter. His behaviour was observed with a high speed camera.
Fig 10 : Prototype of thye flexible lip
Fig 11 : Representation of the experimental setup for the validation of the prototype
To validate the prototype, we simulated and the subglottal pressure with a lung model in the experimental setup.
Fig 12 : Lung model
At low flow, the “in-vitro” observation are very close to the numeric modelisation.
Fig 13 : comparision of the simulated and measured acoustic impedances
Fig 14 : comparision of the simulated and measured Fo vs mean flow
2 prototypes were developped for clinical tests, one for male and the other for female . Subjects had to train before controlling the intonation with the air flow . (Fo increases indeed with the air flow). We observed that the flexible lip was quite sensitive for moist. This voice producing prosthesis can quality of life substantially.
4) Development of a membrane prosthesis
Fig 15 : Membrane principle
A second colution consists in a membrane with added masses on it. The frequency f0, the pressure P, the sound pressure level SPL are dependant. The dimension analysis of this variables gives : f ⋅l , V
SPL ,
h P = F , 2 ρ ⋅V l
w , l
ρ ⋅V 2 ⋅ l k
,
m , ρ ⋅ l3
ρ ⋅V ⋅ h µ
So fo increases with the air flow.
Fig 15 : Prototype of the membrane prosthesis
Q vs. P , for h0l3
SPL vs. Q, for h0l3 h0l3I
0,200 0,160
h0l3III
0,120
Linéaire (h0l3I) Linéaire (h0l3II)
0,080
Linéaire (h0l3III)
SPL (dB)
Q (l/s)
h0l3II
0,040 0,000 0,00
0,40
0,80
1,20
1,60
h0l3I
80 70 60 50 40 30 20 10 0
h0l3II h0l3III Linéaire (h0l3I) Linéaire (h0l3II) Linéaire (h0l3III)
0,000
2,00
0,050
0,100
0,150
0,200
Q (l/s)
P (kPa)
Frequency spectrum for h0l3I
F0 vs. P, for h0l3
0 0
100
200
300
400
500
600
700
800
900
1000
-10
F0 (Hz)
dB (relative)
-20 -30 -40
140
h0l3I
120
h0l3II
100
h0l3III
80
Linéaire (h0l3I)
60
Linéaire (h0l3II)
40
Linéaire (h0l3III)
20
-50
0
-60
0,00
0,50
-70
1,00
1,50
2,00
P (kPa)
Hz
Fig 16 : Characteristics of the prototype
This prototype works quite well. However, it is difficult to produce low pitches. 5) Flexible airway reduction A third solution to replace vocal folds consists in reducing the airway with a flexible mass.
Fig 17 : Principle of flexible airway reduction
For that, there are three possible versions.
principle Fig 18 : Three possible versions of flexible airway reduction
Fig 19 : Principle of the flexible airway reduction
When the air pression is not very important, the balloon keeps it form. When the airflow increases, the balloon is disformed and lets more air flow. When the pression is two high, the ballon is disformed in the other direction and obstructs the airway.
Fig 20 : experimental validation of a prototype
The elasticity of the balloon is tested with two different materials : latex balloons and polyurethane balloons. We have compared the viscosity of water filling and then soap filling. The balloons have been built with several housing diameters and pre-stressed with several weights We observed that ð F0 decreases when an increase of viscosity. More pre-stress leads to increase ð F0, SPL and pressure. ð F0 decreases when the elasticity of the balloon material get higher. Lastly, a larger diameter lowers ð Flow . Conclusions Numerical modelling supports the design process. We have designed and implemented 3 solutions to replace vocal folds for laryngectomized people. The first one - i.e the flexible lip principle - has shown a high sensitivity for moist The second one - the membrane principle - is feasible. The thrird one - the obstruction principle – is very promising .
