Skip to main content

Table 3 WERs and PERs of ASR systems trained on TTS data and LibriSpeech data

From: Developing a unit selection voice given audio without corresponding text

Training data for ASR

Test data to ASR

WER (%)

PER (%)

Olive

Olive

2.54

1.02

LibriSpeech

Olive

3.74

1.57

Lecture

Lecture

5.91

5.19

LibriSpeech

Lecture

21.93

10.20