Skip to main content

Table 6 Word accuracies of voice-tag recognition with batch mode and sequential voice-tag creations in cross-language experiments.

From: Voice-to-Phoneme Conversion Algorithms for Voice-Tag Applications in Embedded Platforms

Sources Word Acc. (%) on Distance
  Target language  
  Voice-tag creations  
  Sequential   Batch
en-US (baseline) 95.2 91.23 0
en-GB 91.37 87.13 0.61
de-DE 90.50 86.99 1.46
fr-FR 89.91 85.09 1.81
pt-BR 82.75 74.42 1.82
zh-CN-Mand 92.11 84.94 1.85
ja-JP 78.07 67.69 1.90
es-LatAm 89.62 83.33 1.92