The usage of multiple Hidden Markov Models (HMMs) to prepare a Czech acoustic unit inventory and speech synthesis based on this inventory are presented in this paper ( Triphone HMMs are trained on the basis of the speech corpus spoken by a single speaker.
6SHHFK 6\QWKHVLV 8VLQJ +00 %DVHG $FRXVWLF 8QLW ,QYHQWRU\

-LQG
&]HFK 5HSXEOLF
DWRXVH#NN\ ]FX F]
$%675$&7
7KH SUHSDUH XVDJH
EDVHG RQ WKLV LQYHQWRU\ DUH SUHVHQWHG LQ WKLV SDSHUD &]HFK RI PXOWLSOH DFRXVWLF +LGGHQ XQLW LQYHQWRU\ 0DUNRY DQG 0RGHOV VSHHFK +00V WR
V\QWKHVLV
+00V DUH WUDLQHG RQ WKH EDVLV RI WKH VSHHFK FRUSXV VSRNHQ E\ 7ULSKRQH
D DXWRPDWLFDOO\ FOXVWHUHG GRZQ XVLQJ ELQDU\ GHFLVLRQ WUHHV 7KHVLQJOH VSHDNHU 7KH VWDWHV RI WULSKRQH +00V DUH
FOXVWHUHG VSHHFK VWDWHV DUH WKHQ XVHG WR DXWRPDWLFDOO\ VHJPHQW WKH
DFRXVWLF FRUSXV
HQDEOH XQLW LQYHQWRU\ DQG WR FUHDWH FRQVWUXFWHG D VSHHFK VHJPHQW GDWDEDVH 7KH
SRVVLEOH PRUH SUHFLVH FRQWH[W PRGHOLQJ LQ WKLV WKDQ ZD\ ZDV LV DVVXPHG SUHYLRXVO\WR
GHVLJQHG RQ WKH EDVLV RI WKH VSHHFK VHJPHQW GDWDEDVH 6HYHUDO&RQFDWHQDWLRQ EDVHG VSHHFK V\QWKHVL]HU FDQ EH
VSHHFK V\QWKHVLV WHFKQLTXHV WKH HQG D &]HFK WH[W WR VSHHFK 776 V\VWHP LV SUHVHQWHG DUH GLVFXVVHG IRU WKLV SXUSRVH ,Q
,1752'8&7,21
7UDGLWLRQDOO\ IRUPDQW V\QWKHVLV ZDV XVHG LQ WKH FRQWH[WRI VLPSOH 776 V\QWKHVLV > @ 7KLV V\QWKHVLV ZDV EDVHG RQ D
PDQXDOO\ GHULYHG UXOHV ZDV XVHG WR GHWHUPLQH SDUDPHWHUVPRGHO RI VSHHFK SURGXFWLRQ DQG D VHW RI
RI EDQGZLGWKV WKH V\QWKHVL]HUV VXFK
WKLV UXOH PHWKRG DQG
LQWHOOLJLELOLW\ :KLOH WKHVH RI VSHHFK RWKHU DUWLFXODWRU\ DV IRUPDQW SDUDPHWHUV IUHTXHQFLHV +HQFH DQG
V\VWHPV V\QWKHVLV FDQ DFKLHYH LV FDOOHG D KLJK V\QWKHVLV GHJUHH E\
LV VSHHFK JHQHUDWLRQ LQ D VHW RI UXOHV YHU\ GLIILFXOW WKHLU WR QDWXUDOQHVV DFFXUDWHO\ LV GHVFULEH W\SLFDOO\ WKH ORZ SURFHVV VLQFH RI
RILW
1RZDGD\V FRQFDWHQDWLYH VSHHFK V\QWKHVL]HUV DUH ZLGHO\XVHG VSHHFK XQLWV RU VHJPHQWV LQWR D GHVLUHG XWWHUDQFH 7KH6SHHFK V\QWKHVLV LV SHUIRUPHG E\ FRQFDWHQDWLQJ
LQKHUHQW SUREOHP RI FRQFDWHQDWLRQ EDVHG V\QWKHVL]HUV WKDW XQLW WKH\
+RZHYHU WKHUH DUH VHYHUDO UHDVRQV IRU WKH PDVVLYH XVDJHERXQGDULHV VXIIHU IURP L H GLVFRQWLQXLWLHV LV
LQ WKH FRQFDWHQDWLRQ HQFRXQWHUHG SRLQWV DW WKH
RI FRQFDWHQDWLYH V\QWKHVL]HU > @ ,W KDV EHHQ VKRZQ WKDWSURGXFWLRQ TXDOLW\ LV PXFK HDVLHU DQG GRHV QRW UHTXLUH WKH GHWDLOHGRI D FRQFDWHQDWLYH V\QWKHVL]HU RI DFFHSWDEOH
NQRZOHGJH SURGXFLQJ D RI UXOH VSHHFK EDVHG SURGXFWLRQ V\QWKHVL]HU SURFHVVHV RI D VLPLODU QHHGHG TXDOLW\ IRU
. 7KLV ZRUN ZDV VXSSRUWHG E\ WKH *$ý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
N+] 3LWFK DV\QFKURQRXV VSHHFK DQDO\VLV ZDV WKHQ
Adaptation of pitch and spectrum for HMM-based speech synthesis using MLLR...based TTS system in which each speech synthesis unit is modeled by HMM [...
Corpus-based approaches to speech synthesis have been advocated to overcome the limitations of concatenative synthesis from a xed acoustic unit inventory. ...
USING HMM-BASED SPEECH RECOGNITION/SYNTHESIS ...cient vectors for each speech unit is modeled by...
Speech Technologies Acoustic Modeling of Speaking ...synthetic speech using HMM-based speech synthesis....speech parameters of a speech unit such as the ...
automatically, using HMM-based approach [12, 13]....: Join Cost for Unit Selection Speech Synthesis...Kavai, H., Tsuzaki, M.: Acoustic Measures vs....
Speech Synthesis Using H... 5页 免费 An HMM... HMM-BASED SMOOTHING FOR CONCATENATIVE SPEECH ...acoustic quality of the Whistler Text-to-Speech ...
An HMM-based speech synt... 暂无评价 4页 免费 HMM-Based Speech Synthes... 185页 免费 Speech Synthesis Using H... 5页 免费 HMM-BASED SMOOTHING ...
RULE-BASED VISUAL SPEECH SYNTHESIS Jonas Beskow ...speech, especially under degraded acoustic conditions...modalities to be written using the same notation....
called HMM-based generation synthesis (or simply...All of these techniques depend on a acoustic ...Ho?man, “Speech synthesis using stochastic Markov...
HMM-Based Speech Synthes... 185页 免费 Speech Synthesis Using H... 5...(for which the acoustic evaluation is based on an Euclidean distance measure...
热门文档