Automatic Evaluation of Voice Quality Using Text-Based Laryngograph Measurements and Prosodic Analysis

<div>Perceptual breathiness (<svg height="8.68572pt" id="M114" style="vertical-align:-0.0498209pt" version="1.1" viewbox="-0.0498162 -8.6359 7.94191 8.68572" width="7.94191pt" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><g transform="matrix(.013,0,0,-0.013,0,0)"><path d="M578 512C578 619 494 650 387 650H140L134 622C219 615 223 609 210 542L127 119C112 40 104 34 23 28L17 0H235C317 0 387 7 444 33C515 65 564 122 564 195C564 289 495 335 412 352V354C505 373 578 422 578 512ZM486 510C486 422 423 367 314 367H257L294 565C299 591 305 602 315 608C324 614 339 617 362 617C421 617 486 595 486 510ZM466 200C466 100 393 35 296 35C222 35 198 51 212 127L250 333H303C388 333 466 303 466 200Z" id="g113-67"></path><glyph.data ascent="3473" descent="-2876" horiz-adv-x="601" vert-adv-y="601"></glyph.data></g></svg>) evaluation by 19 listeners, the SVR regression values (<svg height="11.9316pt" id="M115" style="vertical-align:-3.2957pt" version="1.1" viewbox="-0.0498162 -8.6359 23.3547 11.9316" width="23.3547pt" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><g transform="matrix(.013,0,0,-0.013,0,0)"><path d="M578 512C578 619 494 650 387 650H140L134 622C219 615 223 609 210 542L127 119C112 40 104 34 23 28L17 0H235C317 0 387 7 444 33C515 65 564 122 564 195C564 289 495 335 412 352V354C505 373 578 422 578 512ZM486 510C486 422 423 367 314 367H257L294 565C299 591 305 602 315 608C324 614 339 617 362 617C421 617 486 595 486 510ZM466 200C466 100 393 35 296 35C222 35 198 51 212 127L250 333H303C388 333 466 303 466 200Z" id="g113-67"></path><glyph.data ascent="3473" descent="-2876" horiz-adv-x="601" vert-adv-y="601"></glyph.data></g><g transform="matrix(.0091,0,0,-0.0091,7.748,3.132)"><path d="M156 406V710C117 695 54 682 9 675V647C73 642 75 635 75 573V25C131 -2 180 -12 223 -12C359 -12 478 92 478 240C478 357 387 451 278 451C265 451 251 447 236 440L156 406ZM156 373C175 384 208 392 235 392C315 392 387 328 387 214C387 97 332 30 251 31C197 31 170 66 162 84C158 93 156 104 156 118V373Z" id="g58-96"></path><glyph.data ascent="1007" descent="-369" horiz-adv-x="517" vert-adv-y="517"></glyph.data></g><g transform="matrix(.0091,0,0,-0.0091,12.535,3.132)"><path d="M384 109C349 76 310 59 268 59C197 59 123 113 120 245C240 248 367 259 384 262C403 266 407 275 407 296C407 376 338 451 254 451H253C201 451 148 425 105 378C65 334 39 272 39 203C39 88 112 -12 235 -12C266 -12 339 5 402 84L384 109ZM230 411C285 411 319 365 319 313C319 298 313 293 294 293C236 291 180 290 125 290C141 370 185 411 230 411Z" id="g58-99"></path><glyph.data ascent="1007" descent="-369" horiz-adv-x="434" vert-adv-y="434"></glyph.data></g><g transform="matrix(.0091,0,0,-0.0091,16.457,3.132)"><path d="M327 325C324 350 312 408 304 433C283 442 256 451 213 451C120 451 59 391 59 320C59 243 125 210 188 181C237 158 266 136 266 93C266 52 231 24 195 24C134 24 93 81 74 148L42 143C42 106 51 37 59 23C78 8 125 -12 177 -12C257 -12 345 35 345 127C345 198 292 234 216 264C171 283 131 305 131 348C131 387 156 416 195 416C243 416 277 378 297 317L327 325Z" id="g58-113"></path><glyph.data ascent="1007" descent="-369" horiz-adv-x="376" vert-adv-y="376"></glyph.data></g><g transform="matrix(.0091,0,0,-0.0091,19.833,3.132)"><path d="M303 37L293 66C280 59 256 49 231 49S173 63 173 143V396H281C294 404 298 428 287 439H173V575L158 577L92 509V439H46L17 408L22 396H92V107C92 27 128 -12 192 -12C203 -12 218 -9 237 2L303 37Z" id="g58-114"></path><glyph.data ascent="1007" descent="-369" horiz-adv-x="312" vert-adv-y="312"></glyph.data></g></svg>), and their best-fit line.</div>

Computational and Mathematical Methods in Medicine

fig6

Figure 6

Figure 6: Automatic Evaluation of Voice Quality Using Text-Based Laryngograph Measurements and Prosodic Analysis 