Robot reads ads: Data of the listening test and acoustic analysis

<em><strong>Likability scores on a 7-point Likert scale</strong></em> 1 = not likable at all … 7 = very likable <em><strong>Format of the audio file</strong></em> F1/F2/M1/M2_00/01/02/03_original/calm/energetic F1 and F2 - female voices M1 and M2 - male voices <em><strong>Texts in Estonian for synthesis</strong></em> 00 - <em>Olen kõnerobot (nimi) ja õpin reklaame lugema</em>. [I am speech robot (name) and I am learning to read advertisements]. 01 - <em>Hullud päevad kolmapäevast pühapäevani Vesiku kaubakeskuses.</em> [Crazy days Wednesday to Sunday at Vesiku shopping center.] 02 - <em>Sinu tegemiste õnnestumised saavad alguse heast ideest. Laenumarket </em>–<em> kõik tarbimislaenud ühest kohast. </em>[The success of your endeavors starts from a good idea. Loan market – all consumer loans from one source.] 03 - <em>Tule Diili ja vaheta vana uue vastu! </em>[Come to Deal and swap the old one for a new one!] <em><strong>Transferred styles</strong></em> calm energetic <strong>List of eGeMAPS parameter abbreviations</strong> <em><strong>A</strong></em><strong>1,2,3</strong> <em>= </em>difference of log amplitude of first, second and third harmonic to <em>f</em>0 amplitude <em><strong>A</strong></em><strong>F1,F2,F3 </strong>= difference of log amplitude of first, second and third formant to <em>f</em>0 amplitude <strong>alpha ratio</strong> = ratio of the summed energy from 50–1000 Hz and 1–5 kHz <em><strong>B</strong></em><strong>F1,F2,F3</strong> = bandwidth of first, second and third formant <em><strong>f</strong></em><strong>0</strong> <em>= </em>logarithmic fundamental frequency on a semitone frequency scale, starting at 27.5 Hz (semitone 0) <em><strong>F</strong></em><strong>1,2,3</strong> = centre frequency of first, second and third formant <strong>Hammarberg index </strong>= ratio of the strongest energy peak in the 0–2 kHz region to the strongest peak in the 2–5 kHz region <strong>harmonic difference </strong><em><strong>H</strong></em><strong>1–</strong><em><strong>H</strong></em><strong>2</strong> = ratio of energy of the first <em>f</em>0 harmonic (<em>H</em>1) to the energy of the second fo harmonic (<em>H</em>2) <strong>harmonic difference </strong><em><strong>H</strong></em><strong>1–</strong><em><strong>A</strong></em><strong>3</strong> = ratio of energy of the first <em>f</em>0 harmonic (<em>H</em>1) to the energy of the highest harmonic in the third formant range (<em>A</em>3) <strong>HNR </strong>= harmonics-to-noise ratio <strong>jitter </strong>= deviations in individual consecutive <em>f</em>0 period lengths <strong>LEq</strong> = equivalent sound level, computed by converting the average of the per-frame RMS energies to a logarithmic (dB) scale <strong>MFCC1,2,3,4 </strong>= first, second, third and fourth Mel-frequency cepstral coefficient <strong>loudness </strong>= estimate of perceived signal intensity from an auditory spectrum <strong>pctl </strong>= percentile <strong>pctlrg </strong>= range of the 20th to 80th percentile <strong>SDnorm</strong> = standard deviation normalised by the arithmetic mean (coefficient of variation) <strong>shimmer </strong>= difference of the peak amplitudes of consecutive <em>f</em>0 periods <strong>spectral flux</strong> = difference of the spectra of two consecutive frames <strong>spectral slope 0–500 Hz or 500–1500 Hz </strong>= linear regression slope of the logarithmic power spectrum for 0–500 Hz or 500–1500 Hz region <strong>VR</strong> = voiced regions <strong>UVR</strong> = unvoiced regions Eyben, F., Scherer, K., Schuller, B., Sundberg, J., Andre, E., Busso, C., Devillers, L., Epps, J., 310 Laukka, P., Narayanan, S., and Truong, K. (2016). The Geneva minimalistic acoustic parameter set 311 (GeMAPS) for voice research and affective computing. IEEE T. Affect. Comput 7 (2), 190-202. doi: 312 10.1109/TAFFC.2015.2457417 <br> <br>