Presenter Sten Ternström
Presentation Title Synthesizing Singing - what’s the buzz?
Additional Author David Howard
Link to multimedia presentation Click here. [this will open in a new browser window, close that to return to this page.]
Link to abstract provided before conference Click here. [this will open in a new browser window, close that to return to this page.]
Link to speaker notes, if provided N/A
Question/Answer session The below has the question in text form, with a link to the audio form. The answer is typically in audio form only. The audio should open in a new window, close that to return to this page.



1 Don Miller – Those high frequencies, that’s all harmonic? 8 Khz and so forth? Audio link.
1 Audio link.
 
2 Martin Rothenberg – I’ve always wondered about that, why, without vibrato, for example, it just doesn’t sound natural. When you talk about the spectrum slope, it brings to my mind some of the things that I saw when I worked in the Stockholm laboratory, when I was there in 73/74. I worked on the synthesis by rule system with Granström and Carlson. Being a consonant person, I could see it couldn’t make good consonants because you couldn’t have the spectrum slope transitions […listen to audio for the rest of the comments]. Audio link.
   
3 Ingo Titze – Last year we wrote a small little research note in the Journal of Singing about having discovered a second singer’s formant, which is basically just the singer’s formant multiplied by 3. Just like in the normal tube you have 500 hz, 1500 hz, if you look at the epilarynx tube and look at the second predicted formant from that, it would end up 3 times the 25, 100 or 3000. It ended up being a bump around 7500 or 8000, I wonder if that contributes a little bit to the buzziness? That would then also explain why it would be there in natural voices. Have you discovered this little extra bump around 8000 or 9000? Audio link.
3 Audio link.