Acoustic variability and automatic recognition of children's speech
Speech Communication
| Hi-index | 0.00 |
Two sets of speech recordings were made from children talkers ranging in age from 5 to 18 years. One set was recorded via telephone channels (TEL) and the other using high-fidelity recording equipment (MIC). Special considerations and techniques required for the recording of speech from children are discussed. Also presented are (1) a description of the recording environment including ambient noise levels, (2) the vocabularies used for both sets of recordings, and (3) the distribution of talkers by age and sex. A total of 478 and 436 young talkers (age 5-18 years' inclusive) were employed for the TEL and MIC recordings, respectively. Finally, speech was also recorded from 59 deaf talkers (age 6-14 years' inclusive).