Abstract | ||
---|---|---|
In this paper, we report an improvement to our proposed easy-to-understand public-address system. This system broadcasts synthesized speech, which is parsed from speech-recognized text. In order to improve the performance our system, we implemented a Hidden Markov Model (HMM) based speech synthesis system dependent on the input speaker's speech. As a result, the performance of the speech synthesizer did not depend upon the number of learning sentences to generate the model, and a constant subjective quality value was measured using the MUSHRA evaluation method. We found that speech synthesized by learning 50 sentences exhibited sufficient performance. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/GCCE.2018.8574710 | IEEE Global Conference on Consumer Electronics |
Keywords | Field | DocType |
speech recognizer,HMM-based speech synthesizer,broadcasting system,PA system | Broadcasting,Speech synthesis,MUSHRA,Broadcasting system,Computer science,Speech recognition,Parsing,Hidden Markov model | Conference |
ISSN | Citations | PageRank |
2378-8143 | 0 | 0.34 |
References | Authors | |
0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Hirokazu Akadomari | 1 | 0 | 0.68 |
Kosuke Ishikawa | 2 | 0 | 0.68 |
Yosuke Kobayashi | 3 | 18 | 10.26 |
Kengo Ohta | 4 | 0 | 2.70 |
Jay Kishigami | 5 | 1 | 1.70 |