Voiced Phonemes

Our synthetic phonemes rely upon parameters obtained from sampled utterances. The equipment available to us was restricted to an 11 kHz sampling rate, so our modelling was performed accordingly. Our sounds are presented here in the popular .au format (mu-law companded). Unfortunately, some audio software and hardware in use is restricted to sampling rates other than 11 kHz. (An 8 kHz rate is popular.) Our sounds will vary in pitch with the playback rate but will remain intelligible. Our sampled utterances and synthetic sounds may be converted to 8 kHz with suitable downsampling.


Voiced Sounds (vowels and pseudo-vowels)

ARPABET Example AR Order Original Sound Synthetic Sound
AA Bob 27 aa.au aa_synth.au
AE bat 30 ae.au ae_synth.au
AH bought 30 ah.au ah_synth.au
AO boat 29 ao.au ao_synth.au
AW down 30 aw.au aw_synth.au
AX about 30 ax.au ax_synth.au
AXR butter 30 axr.au axr_synth.au
AY buy 30 ay.au ay_synth.au
EH bet 30 eh.au eh_synth.au
EL battle 26 el.au el_synth.au
EM bottom 26 em.au em_synth.au
EN button 30 en.au en_synth.au
ER bird 30 er.au er_synth.au
EY bait 20 ey.au ey_synth.au
IH bit 22 ih.au ih_synth.au
IX roses 30 ix.au ix_synth.au
IY beat 24 iy.au iy_synth.au
L let 30 l.au l_synth.au
M met 30 m.au m_synth.au
N net 25 n.au n_synth.au
NX sing 23 nx.au nx_synth.au
OW book 30 ow.au ow_synth.au
OY boy 30 oy.au oy_synth.au
UH book 30 uh.au uh_synth.au
UW boot 28 uw.au uw_synth.au
W wit 29 w.au w_synth.au
Y you 30 y.au y_synth.au
Z zoo 30 z.au z_synth.au
R rent 28 r.au r_synth.au


Timothy D. Dorney and Robert H. Sparr
Electrical and Computer Engineering Department, Rice University
April 1996