16 Mar “Human all too human”: Grass­roots Radio at CHI2020

Grass­roots Radio keeps on achiev­ing impor­tant results, this time in the aca­d­e­m­ic domain with a paper accept­ed at the CHI Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems 2020. The paper is titled ““Human, All Too Human”: NOAA Weath­er Radio and the Emo­tion­al Impact of Syn­thet­ic Voic­es”, authored by Kris­ten Scott (Madeira Inter­ac­tive Tech­nolo­gies Insti­tute, M‑ITI), Simone Ash­by (Inter­ac­tive Tech­nolo­gies Institute/​LARSyS), and Julian Han­na (Inter­ac­tive Tech­nolo­gies Institute/​LARSyS).

 The paper presents a piece of research dis­cussing the inte­gra­tion of text-to-speech tech­nolo­gies into an open tech­nol­o­gy stack for low-pow­er FM com­mu­ni­ty radio sta­tions as it is hap­pen­ing in Grass­roots Radio.

The ACM Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems (CHI) series of aca­d­e­m­ic con­fer­ences is gen­er­al­ly con­sid­ered the most pres­ti­gious in the field of human-com­put­er inter­ac­tion. CHI has been held annu­al­ly since 1982 and attracts thou­sands of inter­na­tion­al atten­dees.

Here is the full abstract of the paper:

The inte­gra­tion of text-to-speech into an open tech­nol­o­gy stack for low-pow­er FM com­mu­ni­ty radio sta­tions is an oppor­tu­ni­ty to auto­mate labo­ri­ous process­es and increase acces­si­bil­i­ty to infor­ma­tion in remote com­mu­ni­ties. How­ev­er, there are open ques­tions as to the per­ceived con­trast of syn­thet­ic voic­es with the local and inti­mate for­mat of com­mu­ni­ty radio. This paper presents an explorato­ry focus group on the top­ic, fol­lowed by a the­mat­ic analy­sis of pub­lic com­ments on YouTube videos of the syn­thet­ic voic­es used for broad­cast­ing by Nation­al Ocean­ic and Atmos­pher­ic Admin­is­tra­tion (NOAA) Weath­er Radio. We find that despite observed reser­va­tions about the suit­abil­i­ty of TTS for radio, there is sig­nif­i­cant evi­dence of anthro­po­mor­phism, nos­tal­gia and emo­tion­al con­nec­tion in rela­tion to these voic­es. Addi­tion­al­ly, intro­duc­tion of a more “human sound­ing” syn­thet­ic voice elicit­ed sig­nif­i­cant neg­a­tive feed­back. We iden­ti­fy pro­nun­ci­a­tion, speed, suit­abil­i­ty to con­tent and acknowl­edg­ment of lim­i­ta­tions as more rel­e­vant fac­tors in lis­ten­ers’ stat­ed sense of con­nec­tion.