An experimental offline investigation of the performance of connected digits recognition was performed on children in the age range four to eight years. Poor performance using adult models was improved significantly by adaptation and vocal tract length normalisation but not to the same level as training on children. Age dependent models were tried with limited advantage. A combined adult and child training corpus maintained the performance for the separately trained categories. Linear frequency compression for vocal tract length normalization was attempted but estimation of the warping factor was sensitive to non-speech segments and background noise. Phoneme-based word modeling outperformed the whole word models, even though the vocabulary only consisted of digits. 1.

Abstract not found

Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. 1. 1st LEV. HEG, ALL CAPS, CENTERED Normal body text is 10 pt...

Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. Abstract is 9 points. Text is filled and justified. t = 1500 s, f = 0 o t = 1000 s, f = 90 o t = 500 s, f = 180 o t = 0.0 s, f = 270 o Bandwidth (Hz) 100 200 300 400 -10 -5 0 5 10 15 0 1. 1st LEV. HDG, ALL CAPS, Figure 1: Tables and figures may be eithe...

This is a layout specification and template definition for the paper of Interspeech 2004 Korea Conference, which will be held in Jeju Island, Korea, October 4–8, 2004. This template has been generated from WASPAA’99 templates and aims at producing conference proceedings in electronic form. The format is essentially the one used for the IEEE ICASSP conferences. 1.