S702 D. Kewley-Port, W. Su 2/12/02
The Bradlow et al. (1999) article describes some continua used to test the effect of lengthening formant transitions on the perception of /da/ versus /ga/. Given the claim that SLI children have difficulties in processing short transitions, and that they not only find it easier to process 'slowed down' tranistions but that training with slowed speech improves reading, this lab focusses on synthesis of stop transitions.
The goal is to synthesize four stimuli with KLSYN. They are the endpoint stimuli of two /da-ga/ continua described in Bradlow et al., 1999. Two stimuli will have 40 ms transitions, and the other two 80ms transitions. Follow the steps below. You may work with a partner (identify in report), but turn in independent reports following guidelines for lab reports.
1) Extract information about the parameters for synthesis from the Bradlow materials. Sketch a spectrogram of at least one stimulus. Indicate clearly the major synthesis pararmeters (formants, durations, F0) for the stimuli to guide your synthesis strategies. State any information that was missing and what criteria you used to select those parameters. For resources you have my print out of a /da/, and the Klatt parameter book can be checked out from Kim.
2) Synthesize /da/ and /ga/ endpoints of the continua, one with 40 ms transitions and the other with 80 ms transitions. Print out spectrograms and parameter files for the four stimuli.
3) Listen critically to the stimuli. Have a naive listener identify to a few repetitions of the stimuli, writing down the responses from an open set of all CVs. From these comments and your own evaluations of the stimuli, comment on how appropriate they are to determine children's processing of CV transitions. Are the any changes you would recommend to the stimuli that would make them more appropriate?
4) When you are satisfied with your stimuli, upload the wavefile /ga/ with 40 ms transitions in on course class files, under the folder name >"contest3".