From 73d4adcc9e0f24c2bd5a3b395669f3d69dc3a6bd Mon Sep 17 00:00:00 2001 From: CjangCjengh <101577701+CjangCjengh@users.noreply.github.com> Date: Sun, 14 Aug 2022 10:32:43 +0800 Subject: [PATCH] Revert "delete symbol.py" This reverts commit e3e355de808371f01501b622802afd99656414b0. --- .vs/VSWorkspaceState.json | 2 +- .vs/slnx.sqlite | Bin 90112 -> 90112 bytes ...211e1f28-c38f-4713-a77b-09725dc4fdd8.vsidx | Bin 0 -> 33102 bytes .vs/vits/v17/.suo | Bin 53760 -> 45568 bytes data_utils.py | 2 +- text/__init__.py | 10 ++--- text/cleaners.py | 39 +++++++++++++++++- text/symbols.py | 29 +++++++++++++ train.py | 6 +-- train_ms.py | 11 +++-- 10 files changed, 83 insertions(+), 16 deletions(-) create mode 100644 .vs/vits/FileContentIndex/211e1f28-c38f-4713-a77b-09725dc4fdd8.vsidx create mode 100644 text/symbols.py diff --git a/.vs/VSWorkspaceState.json b/.vs/VSWorkspaceState.json index e3b564f..5917527 100644 --- a/.vs/VSWorkspaceState.json +++ b/.vs/VSWorkspaceState.json @@ -3,6 +3,6 @@ "", "\\text" ], - "SelectedNode": "\\train.py", + "SelectedNode": "\\text\\cleaners.py", "PreviewInSolutionExplorer": false } \ No newline at end of file diff --git a/.vs/slnx.sqlite b/.vs/slnx.sqlite index d7477be03fb8555211f0c6a2cb6cff324bbbcb6a..c5f21710b0a70e82a60c992c50481ebfd2996be6 100644 GIT binary patch delta 2050 zcmZuyZ){Ul9KCOCZ~yGw`|Db|w(G{)b+}Dv>3}I9#Er42Y?dg06ohqM_gYHY_eT3V zH((SJA$%}Icw%Bg4DtcN5CSwYd>~;GqcKQyi6PM#qCtcYfEWW&NR0ROtzAp#O>W-1 z=bm%#@0|N?_f8r1P8mM%aElK&_Hv8Q51j;h48a|^44;uxWIunbQ0G0!>-WX6K{?*I zS$VtZx%@6-@v>jbXhzyD#N?Lc-WFLHm$z(*rG%&`WU|50iIxS`lh#Kw2|=AI=2EE= zCPveO%C6+8v`J!_^cnQ1w{EpCoJ(lj?1rq6StrP{a{Ogu;bfDmhI>)D+}r@23m1hJ zE%Jssmv?n8U(!{$(`>PFU4_3Fg4L!&0n056Y4`&+!?$o2K2+v(zvRADd%Sj6ZM?Rd zf52bk&(u0Bw=7?(t>Ck6u_#}yajxp+%y=TVGnbVbyD(~FZ*gfnG91kc!R=W|G@5?i z`OkmBt!rQ(vh^@j&P0=9B%VwO9ZN7$N0c9SdrYNiN_W54qcQCBF>-JwJeHKRx-!12 zzj=d(H*{d+XZ%PcDJJDegt_S`O>?(n=Uw?lmbUf}>($PFcEDla417kQ?(|{Qr0z|S zbE9l$){?T*s{t`7Be3!DGN7Cp@L-LsTpMUAJw6!l=Vv6%7&RjkR&7lsvvRg1>GNRJ z$=1>mip$Rkrm;bbEww%S47A?O_JqthCnt+GmV`Yn#zN-2>XDd@OPRE4atU=iSu~8L zgJ*K2}BsDfg1u(VkT>K9Uyan8}o$_QC0mx&4GvgX<`3NZalNZ64}W`wjL%Gsb? z!L+GNGk&wLXZ7oS!SqN;Qt1ghOa2Xp-ODupT)={3)6>#OE+te};-s1oLxd^=lyl*7 zCevYilP2V-WrvFi%^a@ss@z)lOjNMZVN~asa7+@}87kN0F%=!fBT+dTdD?Eci=a-2 zk)82^V`Yww(@YB{l4a8nLUy&5n3PUSVzCw%X5}egtc969xD8r7F)0-t4kkw@#NpCT zJ$h{yQCVi5CT9d@zF}95W|O*HG9j{KTB2Eq5e4{%92|o4aEeTm1l%V_;0An8o{;3voY%I91BASs_3jIH7o&` zbswu}#jA2Q(&DnfPbdJGgPZUK9>HI57jDCka6@qo`IVJJ4fa=I6|?6B2*LBvN*}9@ zm41|(U4anOw*?j|SZc)v9mhG^m5_8g&qfUDt=f z;s$JEKj4QFBA!Sv8beI@GQ>do!;c?)M1N?=5==}?0KXIY2u4tY$@t^ByK89)f8N}j z=Y5{@zVGwiv-g5+?*-dC0b%W-x?W-J*=@&x&LOw~=ivl7N%kpod3WR6ig1%1t=l5c z$k>K@XSTA$x&S+07bD4BB0-#!L7Rkn-R!`+K>m1rO@+|K&NtM;sugYFwzkG_*Xs86 z)hoO5HyWHSp)>!_3UE~&MXK|vT=^CJ0W&ZRm*6a0wEijoiOSuTBbDowJIbtby3*tP z4-Px8Ixjje=7X-+o$TvpeH&^!F$xJXjz-fH@mPGj7LB^(E0af@e}9YW+A(So9C#!( z9vzBhHFZZem9!Vo)>RnwaFku%6OapsY<+)Yz+zV~#mK{zcj`t~EaJQS8;FI&7Gw_z z4tBKP+h(0q37mmstd8nNf#7YyD8O$Sc&4{+@TERa;hg=|@7-+C0?imT^AsZ!)028~ zR81#G7ZvjSO&A5u9`w=Z%As;7R074yAQl<$K541@f*93vfvjfa(rS95I2~W31;kSs zjh~Mf0d{&IfGskc9jGsS-Wdo@FGxBtYT$j2=m{;MXN_#3*kAyomHe)W+^$?UHDWws z>UFi2Cy5=|N1$FmuL(JD&d?KCUQ|IiP{UcsF>TH$nH@=G#!6AYkM96y%?~hI26?@f z;5<|!Smm*-GPLocRlY@5TNy2z%4OnOv8~0`7%k`enXvXoPS0p#TGGg>MhUaZWIFJR zeLYV<*QbsR7wTp`gTBK2B7^=eYkoDrjpONKso`8gEA7NbEg_C*WdmSm28*72H&|V7 z33)2{a&e)BTd`P|_U(hYe z9PEN5Il$Bxd=1y>EWJPQiq+((j#gbFism_arJHiOL!k{LuNQMoV2D#%8dg?;2S>42Y=WAG)>;1t{; z*T^g!CLfTuXt(kiq#;gTWop{jaE^XXKcSLxiM~(Xr2FYO`IFWYS-DO7X%Fq;?|!$^ zLfleFD0RivSuM2)<-XY4*`+=4ReM diff --git a/.vs/vits/FileContentIndex/211e1f28-c38f-4713-a77b-09725dc4fdd8.vsidx b/.vs/vits/FileContentIndex/211e1f28-c38f-4713-a77b-09725dc4fdd8.vsidx new file mode 100644 index 0000000000000000000000000000000000000000..ee3c84edbf8b11cecb08195e90833f7aded28a78 GIT binary patch literal 33102 zcmb`P54@FCmG=)NnJM{S@(=3O2*rHv^Z!qZa79HRK)hbAfcOl9jFA!&Aex7ZW@KiJ z`6@Dwp_xODnVA_mX2vi|W@bo;$PkeknmMLMNai5#`u)yt&c5foM_>B9@BQ4vckQ+J z+I#K2*IxTM_nyn}(?-0%QmGtsmHlI1d(=?rW4vd-S^@QK_!`vKc2Zm0(6)j2r)`*h z4{RIO({}!_ww{3|4IDUl;BkWo4nA(+i9G{*1|K(g@ZjTm1`qBT+%s@+&oI+@ZEc=? zb$7IPlV-dVPHmF*MD_DQAhb^E^O<16s}(Dx(Xk9|M! z{nYm}-_L!&@BsV?Ry&H1*TWPjf#l{LI6IvcS(m zKa2b<_Ory#Qa{W5Ecdg*&pk{x5Bxmz^T^L*KTrHT_4CZnb3ZTq!n4Kji@+~Jzli)I z_KUi2=gE;g2=<3m}wLSQ4~aR z5G6sB22mD7c@Pyr>|w{O<2VfBD2U@APJ%cM;w*^sATEN$!+zSZNf;ziki_`U@m82kVQcj2U!wiX^>?>mIql8 zt4C63N!Y~cPEDZB7EW*e`P}|yDWfX-`97ahPrD2qXQ65G`7<&kL z+Zi*BqcD!cI0@r4jI%J#!?*|&4+p~b%{oq^Fp0w?36nHTvM|ZRqzF?F$HjKlhE1a| zjl(nv(=<%8FwMiX2s00d$@biw%Azof!z>B2G|aLv%fqY)a}P(&BEnqGqcD%dJPGqO z%(F1h!@LL!&kmx+i%lpjqOgd=A_bKRkzAGj^ZSW(@eAGn^SolEY9;dFXF

W4#YG+$MO>$~8dboBF z-xk3ZXtp*MXBJ5oD;5prpNC5c8)jQ#i*74t3ujAXleg&=N#fzk!ob$G8C$h|rAd+{ zNuDG{l6tt%Ebm%ku=uqlx5_k4(kw~yBrTH6!*yp_+0uqZw{3`-W@(aTNtP#Bk>nmO zL(AEgR4n#wyR73pP4X zpT&#K%#xa=ElW&xJUm>~$QPCo>}Xmv*}Uy!+S#+yW&7>n8pi>!5Vs(stYRSOD)|S~S^E}IotnlnX z%$^tQLbJSUsltNVHo;7@BF~B<_quc6!($pAJ?+M`#B905qTY7N`nCr(dtC7F5Qpa{ zyALgGTc)uyVS8vJ+tY^Kb{-z_aQ)kDX-VDkkewLYVe`u_OH0`v9sqHN*)3%$-?EgQ zC)Dg zczCqLQL)2d;cfwKA#0Ik>tTUqv-R-ciS4%ywk@GNUjw15#u*|0c+>N}maZq@( z+TB^i9v-E6z2WBMC4^(dtHLgWBJuE`%`}aBmB|!`j498OT9JBq3}@28t;^JaBgg}6 zN7N1>9@2S`a3?d|aX=ZA7Iqd`c!Xzg;Kt?_;`nl*Z7FTB>;d1F!zOG~GN;Wg8`!!w zW2?3|YD3$G+85rgwGFhlYW6;D*uWlp57yCsW=H$zj`lGf?c+MyJ3HFDI@%|8v`^|d zr=w#+wWAC=%CMu1I?A}COghT6qs%(WyrV3J4=;z8!SFIXt>c_C#+^H28A zE{ic`Fs2O0l+lw9ebn6d4h<3_iQJA34~bIY?w zj_53d&NA#Qqs}t!ER)VM?JTp-GVd&lb8J3k&{c+AWz<#1U1ic$rd?&$Rpwn~F`?>D zs0I_N;e={5p&C!9CibqhyJNz+atR|b9=1Z5DGK~x5D86;(pmO)kqc^MRC=#`;g zhCvyIWf+xVT!u**re&CwVP1ws8F^*omr+ngVHrhb6qiv_Mrj#kWt5juQN~^w`(+%I zaahJt8OLRulyO?dSsCYLT$G7dCVrU&WfGQ2R3>qmBxRD8NmeF#nG|K}m8oB*L79eS z8kK2Wrb(HmWtx>~UZzEvd1dC8Sx{zSnMGw5mswI~X_;kZmX}#k=3bfmWge7ySmsfg z$7P(kSx&7E`Jr8DfGf0%Lc^owWAF=_5+r_T6H z`9EcE?qzdjS2-$Qne6lJ=lc9j#@@7xr%(ON^b2n3MAfgWC4~hGT9~SXbAJiWp9wZ(jendP>JVHED zJX$RXj~RLp)O)C4SO~{?1bQZ1Eg%g3-P!-6~Hs zqWu(csyJPoVZ=ILr1C74XNz;hd8)r$)wsPe<&O6gaLkBY0s$HgbaC&j14HR4($##t|J zlzx-AS==H%Z$$g8D!(MYEWRSXD!yh!`)w+37k7wniEkUxZkNh^Y*VqX?-%=udy0FB zdyDF20C zSLI8MnBV8b9`W6RSpy zccsdo7r!8WQT&qlWh2h*bt*45!u|$viS$cVzE$OA;_c!c;+^7h@heUECqQWki2FRo*4`F(uB?`^CQEp5k6c`>ym8_m}=a@nEsPc&O37 zD+9zf>5ml$ii5=C#S_F6#df28S0a@YF%t{%Byos1RO}Fk8|}L?LgkU-XmN}Ju27w;197Vi=774H-87atH;h!2Vni4Tj9 zh%3ca;-lhf@o}SlSDsY)DRGUsR$OPa@5%<1H;S9Y&EgjEd82(-wyOM+%CD&Ws`#4p z+f?4J@>?puE$$R|iGBLk`tNJRI`1j&CGIWu6ZaMO7Y`HdJLsdRp93UPgwu#4z z1I0l`*q@;CiDJ7Ls6J6SQ~4y7hloSP4%MG7jugkJew;X7`Yx5b#fjo%af(=qQ^jfG zbn!xQh7t3bsq!pwwm3(eC(aiy7q1W(hzrH5jOcff%GZe3ir0yY#p}fz#2dvW;!^Pz z@m6t}c)NIq5#ufw@0R`^@m}#h@qY0EafSGx_>lOp5&f+|zT%$Z zUgF+jKO_3xU*!YEgT?;hq2l4<03+<$R6bT5C=OEn31Yh#sXh@iu@FxZhloSP4so~< z+SxK3OzZV)$$o5amV^#8odFNj;km&BLF zSHxGv*Tgr(ZQ^!uhY|h1t@2KBm)K`dokOv&xTm<6xVP9(+*jORJkW@7`iqB4KR`T6 z`eRidC=L>jSN(}9w~K)osXkM=5Kj_^h(pB=akvqFjZk@{I9ePdjuXd=onn{RElw0C z8_{p6@>Fq}I9!>SD)DM@k$8=Gt$3ZdSiD}m zLA+61A}$qg5pNZjiMNY)7~$`7mG2Vo7Vi=774H-87atH;h!2Vni4Tj9h%1d4?@^Uk zi;s&>sQxLH*NAJyb>ez)gSb)LByJYBh|h~Jh+D;%#FxcajPUO@mERDziQB~;;#=a| zM%e9Axz7h{<-SHff2zE<%KgNB#r?$t#e>EE;-TW<;sEg|u}wTy94HPFj~7o6PZZn5 zK#au1i1`#MpCk?uhl(BIaPf3;gg8OI5x_yj5H# z-Y(uD-YG5@?=qsFdsMzxyidGed_Y_wJ}5pUK5RrkD^*@4J}Ryj9~YkxpA?@G*NAJy zb>ey>`roMXCULX4MSNa-LEI|7B)%-ZBEBlVCcYtVGh)0Q;@i^i6nBYz_R{?+_7(RO z_Y(IO`-%G+(f@%eA1wA44;2p=2Z%?BZQ`-wKyi=}{hpxmiDJ9-k;;jfNq>^cL&TwC zhdA7ben+T0QXDOg5yy$+#ZIwH>=q}Alf@}wDNYrqiPOak#Tnv7;!JUtINJz+=cznj zyj;9OTp%tKuM)2o7m3%1*Ba69VwJBKZxC-3mxxQnTf|$%W#aAP9patha`7(lZt))R zUhzKhe(?cuh4`TOkP&`AqVh^{mH4Q*T6|o5LVQwuN?aqZHKO13DsK=sikrmE;ui6F z@da_K_>%at_=@p(zzA;ihCL{pS@M?r}F+PA1EFy z_7@Ko4;KfBM~Q9XvEo2+ka)a!f_S3XZiGLP%88hXg?N%UL>wx1h{MIx#S!92akMx_ z94C$!JH;-sTbw9P7N;2D?^KniiPOak#Tnv7;!JUtI9r?}&NHIl%T>NYTp%tKuM)2o z7m3#xVSk-?y?CRzRJ>KZUA$Ag%ZTT&yH&ndykA@)J|sROt`b)pVgIK0Yc!oI22>(Z`JWlKs zyT!?3DNYkF6fY8IiF3sH;uYdT@oFR1f04@9ii^b?#3kY_;xh3Lak+T6c&~WBxI%nL zd_-I&t~N5>RDM!iBmG)&o%9=2-Y9MoHyg2zzgGD-;%~*@i7$%37ylssm-t8Vb@5L| z#KW87U&VLC_w7@w|A6=*@x$T)M$GRZl|Q2LVd4?eAFc8+Dt}BIEP7&SM1Qf$xp=ZT zO!cR#e1A$1$cg62X{{!)d;*VAT@8W-mkBL7K ze=7dWi246Q<)>ACM&)P4=fwY1{jXL2jrd#fcjAji^!o?#kK&(H|7Y<{>Hn(oe~a&k zm43B#exJC9_yO^Q;)le2j2L%6l@AaP5)TnSVnn+mR6bHXT0BPeA2Z@y4pR99v0Zv! z>Zc=oC(tPkf^&7lCy9+C*hNkm7~!PpQWtz;&k{t zVdd-g>eZiX_ujpn;+29iC?S1NNMoegOKq?;g(fHV*i`#Tg8`#S@^;IllH)1Y&a!!d4vZa8c&fb=>f zd_a(&YN|ydK2KXY)n1iBknnNi3O*}Yi6MQK7LdLMiQc<*Kd-lQ<3@E0*3yDH_UtVF zU)nW|wi`Vk3$M+yrT) zy=v=s2_&xH5s)UKb`hkTAx(wE6MrtG8PFX9$@TYr=(uWbv=gD@EJs7)&ioP*&Kp{TY_4~##Bh08)Nq} z^vqg(Lc6wBUxYNWG5*}RTzh&?p9fgm%8jt-59!|^t%CGvNOK`w2Z`Q)6VeFieh+Ca zB<}gS7!99!s|>SOCOp z)up!~)y9BSvu&IuYZ))CF3t7z(WXGc-;(fy`L&sML$?L}xs`MlwVoR-sM`V^``ZDD zz0iYJ*JrE#XuGS?)?YNNpex02g25B9tn660?H_uNXgM8Zf74RiD0;9Eqm4%@9pBg=r^@M1fd!@Ybz24l7i>Fp9L?ZZ!7%0Cd}7aWSMJknm?%y<&~4i$9_1b&@;L?hSiY{tSuxmOt@FyZ0Ah8|&d(&A!;+ozLTbG-aSFT83!lAN zy96F=fQ~`E6Vhbp7B()ETTW-Y4DHw-x0@2fyXmN1igr!UYg{HS@n+OMVz1ivEQG`v zZFx5jYV&BUU0?5oGoOWCIL4-WfB0MBS~7YN5-rw1ng?5kGRIv8T{m>x+S<%9n%j8# zybesKb8E7_9bVCvJLo*%+}>{VuV&l0R`h`J=X&9|Z=>fGSZDn8 zcWoC|LSpUV zqu35(xU}?>x&{6;xOSY~xU}71`U-7%-EpUR8aYOZTGx(q<0<1gqQo^|+%x@i#W&Y| zaXaJ2oq*cQ(DQmoM?#|QaE3y!e`y0`3Ad(_})oF#kNfm)sqwtKCyzuJ#`p=*^V z-EQqdFT3njVK0m@?#DaOah;vdoCgzkJ8F5|F(15&US5E13Z&;C&4)Ax66eOV>PCAI zwLIrfLE^aEAaPtbOSa=Va_x4Y9ZvvzruR#sWA=BZ4_miF^lC6_e}vlGAh8{7H$lf;;N1R#TAsllH|~d%9@aR=Xk#2sXRk6;Y93p|4}Ow}t7ao;9E$2DMXr>D!IJ0Jbgf?k~u z-3aI&Z0wm&1w6mcpqA%1fkcZjke0&YU}VCB?bXYJPSn1HTK5pmUbr7~QA_^@K>C5b zYIy^HSKn*>x>1Y2Rqu7@ZGevVk;_Qi>%B~Y#8bg`ufVp|uQX`Eh+u3m{-~oZb-ev3 z(NjwFl)J@t^p9~$+t1lxwG(<<<2-7yf!fXRk0)wAB<3ONXgd$OAKR;TWApSdO1Ns= zeYQ*L?YIk!U9LFi>te%=%P8hN=<`yHu^zo}e;N0@D_A=jwFje?kuwF-Oz2!KZQ08_ z)N)48LHZ%u^@Vf*q=}F?5B>^gAlBt{#LMaSQoKOd>Ne$QJX;)t_Y8eG3bxd-XT}xh zkzh18k7w)c=0c(^e-Y$PC|3>t3q$2Bcy$)qxmDvBZ=lw#DSJuKZUaVJ3~4*0X^>d^ zQ%Lkag|r^EKX2@XJ=21or=v z-4xdjW3<|jrZL=_?#395zg}ZtG}j;F?_Fo^Z2My+IbYsKoG;JEJoHB$$Dj`XCq}Qc z&(lMT2))pPx+aU}cRT3C<*+Jj@xNd6(mAi(j258XbbHmJq!ZG7=q5mF+7Eig+S#bx zfOhjkP6uVJ-6k zXUY6Wi8l$`F~1Bze{9D)WjSouLdSN@{JcNdj`v5aS<;qX(Ux8{+oI=MJM4!$GvhHD z^A-K0R|#~?SIn3zVByAP7NCWjIc=E$cZ*?70gb zus=%dkI$3C(2j4&uY^~vo-EzmoPX?P1#BsesY{GC>Yjma9>z@|(Lc7M&uqt9wxcd- zTo=}kfzF)(=fV2=jM$Ddx)3_foX;EFH$I_o&sU(9yTIoX?k{f$w&T-3+bu;qK8>7) ztbCfidSzwWz6#w&v}{jr%}sU z-Vceh>}A_HMy;2+1=pSyTzihjTl`}f_i?nl782Y2JtY2ktX16qGiv+WtMW}qt0CPD zX*Q%!Lpl%A-`*~T?RLzg)ws+Xze3N<7V{zT$zwC51h&+qQR3QD;`u!XwafKnunn-(?*Fi>0Vt1j)5?9?U@PMAUu?5OB&&LI*<;Ok-K>8DExo@Y` z+l_@Z7qyEZ^@CLF51zW4&W(E*+TDb9UxdWixC0V*nrH9~)bgxyrw>8x5cKDAA}yFV zIFH90*X3|n++eQ?-_7EG8|$^s&X;Xy$2E8hQn$S-jNSE+Zi9|z;a65BC%vYi`^7F^wX>lQ;Hao_l}2U@%c-Bx?mK8C}4<~z`hg^v5di1;mZ zzk`k*G_8;umtOG>xE>a?S*C+`4HMQ{hH&$(DC$eC7G#du^Rp5sO1NN+_* z?;x#$R6ycMISLZz%k!}(YTtnFQAj)iuS43m-fkVFquwo^|BPDtSA|4dH^$@8aZO)` z^dI)B&{O87YoMD69X;Sv8Z!v{V-yD%x7Cg4X3jjs7~(nS%;^hb?R<P94<9>`q?L4&OS}|g|F3e#=PIJ z7mh|>IEEV+{|RO7XP{iQpTmM{brK}*G*_2<*$ExnaTnN*qjAs2pqBmNe^l(1*UrTl z=Rnu$op6fUCHAVl=bR2{K6KoD&X>D?DRkT!u07*^T)iFli0y8IZi>BXA5!e!xJPd0 z-07`o$6k03vKK}yclu`3o{0XILmCZRH+`{8_HytuVzhWjNfHrlIp zQh0AIfsS|2bVwJV_HsyvLSlrm-OcrOiy_TLts9LiPK!yXr3GV>Jv$3NXAgkwb7QubcUyu!vDR91>3o#~2P>3>{AtcRzuSvBq{^MJ?CrJz^@EN(&3Swu7Vqjob33fFd%|<$x-p#h&9)fV zwL25za;Fc0#MM|1i9Rz9xEjnL{ApvWaSy87E~?v3g~S!F*;=XM#&GSp9~|QnSkM=@ zGt=w+xmixE_dEg;pYYZ~;?w3_NH^K5_S?gOkeEf;j$T~?9Z!Ip`JT`%hAn6AMx%e6 z$86LxBF=-v9P3u&K%pNF7xcl>(n(Jp({e!9R*qTw>PZ zKe_IeKwTiUdi(EcTZPJ5@3*eb!WgdIyFV^8j?1*^LV)va-aPbN3yT)3(FI+zMJpY8 z+3i2MYt_NuqI0LhZR2<>(|EYymdTwD*RFZ#Fs=&=XWMEk++`ofaw(_8P zgVUNxt$F6?Z+CQV9{+c#d1ssjMr$?SR-QJmBy3yFvej{J)nD_t&MQdUvAQeY&A(RT zHtSq}=%tkh%}Lc+I0?3=AwiwybdAzXc*+m8zIQtp8_hGv7-zM(#hcqTj_V?^mBb$s zz2ht}Msr*>);8~(t8FErzuonZC*{Yk7uYtRDCnAZaCgT*yWLrP=K&<9^)-#URx8=8 z!)R_LyXskQcO=+0Ut{oKCwf^3sny${Ru)IM>Z+9vowBj>xi+I#jiGB52W5ShO>gI* zuiaKhL*kXS8q)f@gF6GfH@Z;BO6#heKgs Rzu+4fEVLUPl(}NtwOzLiZ8y+;vF=N^AE@oNPLxp?Ya^6N zUSRVB7iI794+4tB@JFJ-5))$#)gjCuO!#AnG4Ugs=noSTXyOPm&*}P-Lcm}|y~)Wv z_c{0Lob%js#dnqBPl^K$#!WgOU<`)&%_l>ROi~Ojyu3WUd(nwY%*%0H$^r>Rg6)O0 zqD!+FfjMk&QE$!Lg@f}PB$Nql6RK~ZP%8k&7f*)x4L5M&5(H=g=^E6Rq7-Mb1=(JW zy^z%xVJ{!jgF3AEefic zRx-bXkNgXrzH+O7P0*lp{u6$S7{NZ2 zev|=}2+ANz6onR&JeSqsM0s5)0iy`KfU-Mh(jHlt-E}l*8D)bqfNL@wWluuf5fq4V z%W~j6BFonbxaMdUC;>7{Gqm<2@R|t{&V;Z5)#E7Cr06<<5>5=2x6gYP2YvViO4>9; z3>i;eEfJVxzzY$wGRFcj!x*!`tf3jE6i$_hrsv^gpast8d%-01{_~8fG8u!}B`ty# zcQ432+~2{&j-pYfgLF}wmMcV?F~`mpLA*F9(0eS+fwd|eY$mc`%ZEMoNAh=~My!ZI zz}E$)BUX6W_Yu@5n;~WPOJ(cejKTp|%S>>fY=EtYbXl90>2SUUMGg|vLwrRWOj*|6 zDYOua2DPjV=ieGwSCB!&%v$p=o9d)rFa0p{#mD;L3jM)r8~cYM;b>(0$eL|Cdzb>= zwv|dVJJeAwhFfrOdYK1et8qtq1|=x+E%GOJp z0jeDIHaL!Po#1%T0KfGVLVLw3!BOcLeKb5Y5*dkfWDH_h4XNoA9QIftI%R^C-T&dV z$a2ThjCkKtTyQhRnB+N68pI$Ezk6GGGhX&)#FP-P-lz7H1Bf2EumUcR>7e~$2u=mo zfLPBfktHyfG=kV!0B~+Ge1FyrFQ2W1*f}Q%6;+Tr@7LW;`^Uo*}CePS4V5>tAqQ0UOIC8oGo5kZ~5fC;aBuy zJ2Lu~#SPzV={>a?ak##19i&DAnC0a9;a!SkD$t}`S>=jURlEB8>=CcqZFjp|ZhLqj z9I^Wx;i~>YXU(A3HOOYHm4G?L$DWo^^LBBANPJ z$nH>|qZ91ki0gp2KTIL*^Mw2DZjaAp?{oW{cCW`3sd0NegFctwS^yh)=hUiF))i#6BFyd4lBw%q?mTu>5Ie-iOC8Gcz)| zWat!ivLC^Ss1nXgQ>AX?y~v=`r_;X(tKTOXEWW_R;niP^j*f<7i>M4~Bw_{vo1?^2 zVLb>T_wtCZ#Jaz?EkzoLn1)z^n2v}*_#t{Kz60+Ds97yYBniq=)065@vllHZ4PKcW}Whd6*Z zi0DU9|6z^`;44SGjH4bVU&-7Gr?eG01}StZN@ao)VRLvQYlIO+rA*Kx&$2^@f0=@o zMrKKr*qg{WXdkF-M$iYp!NtZ;wc4+1u@Zi$st^Li#rbx1)P{;h2ikUpbATZUy1bK4BN((a4W1x!V*yc zVg}+sC-RfHu(8HFieda?Gm$$BG5)cg*yw;;5mrSK)&!TqZ)FH{L~a60i5PB1TK!ot z_LCbCUS&bhF{4aD-h{l6z*b1Gu~LK&EQ13+EOw_={_5h7%*AOnJ7?{?lzZ0irq*%g zqQc!{TXb7{wXGid2BM~|LuoP)BW&({GI%vt$2D=SoR(X~QBNQ+RZ?67a-E18t^;Wk z*UUM_CCJk_VRV0%J9HlDMcRZ6n+Q5XOJ#!hp;dWqgen-gwDItFRW!>47N2MozE;Dv zjE}hHV$aJ}-66`V&@fhbC;V!0LDlcrv-LgpkZbc!omq1kCmkM4>f*+yz_=V7E!}D;a5`LLIzIKBfo7O5 z%y3K}C$=Pgx@bdmm2JUKUXNM(eoyD%KTf{2ZrVWi>o-4-Tj8LC&Tj{8iBKS|Hv%jP zMzjruv<_8Q&4DW|(VE2>IHUJ#8TdW4SX_{k6TVo(HS(Ia<10<59VhlSa5gRjpIP|K z;OxBoTGW1f?;QAeAPk0kgM4rbP^Ws*mtm!iYlH0vvwRm{>VX7s!0QGw1EE1Ul5CY$0^aDwQw_XydlM02?~E6=hYBol*RvW%4h53p%1V zfiqeKK{4rM2rLvjVoLoPMT6K0l1g$x*yan>3H-Db0htPF(6~gQJYWcdBw}|!sLG1G zVlSWigowz?6$GlIC<0PLh=R4)u0xQMx4|9%a0m(z%bvm3UIZKq%u%k#N*7`SVk2S; zf`)KM7d*oHz-d&2Dp{?dkQ0Nz5qTB4bH%VWxm-qJ&dXDSC0p$IJXy?OvpG-1*&!&? z}`^DXjqR1Q+^q zaX5J)u^}S|iw?Iiq*xJ#9V&#dwbBY~REyzuNRfiF9Lj>m`|`07nGT)PJ6S1=OfMbx zlJ!%0$=EWs__Y{uK+561H{Sl_@Yc9vAAeRa&%Y95uaaLs-Fo4|+G8m=AU`~PLMVKe zMIF`mIFY-K74uN|g{mEN2P}Bn>A(-5+pdHiX_0vbbSLTL^w1$?p<6Pd3pDZ$$%gL7 zjt8j~pR0K%t;4lhF!+{a11t-#WPduVLDI5FaloT_`CMihrR*t5d zgpn35o7HaZ+`vaM+5d&(nt1oPl6SoC@bhbiR&9#hs7uv({mG}QsQ=U8JD4*qWTwfi z&rHwI*J*6FhD?pI&TiM#8|@7myWW&(H=506quwCqJclY}46asZL;s>+a4q-u>@I6p z9zG&+hZZO(Ad*?p!Q-8RGq%j1i$~7HY|L|`8SjzkFRCdJ(`j2)oATb$KQ=oy?#|B7 zPPlYVEPYDZh4qAM-`@1uo~|8Kk3Ie4_7|paTdVHK&fovs@w5MIOIoqB3;8V}^RBck z{^=~_o6B}wk8m74M>6nxQ7jB)TVQmK#4~^PdXanfTo{^@=gw7vD>oD&P^yo?ouBF% z$~~`uExqyZp*0eQtrZlSmJ}y}rFa+$N^&swJPNB+ktpH}zAvkS)4k`wbs*8#o2tOs z-Ym=H0~?vnag_cg8ug8)OtY;{V=(G7G{%O;OigB%E?r|X=uC}`jrAEBhAdyM6ZEB4 zF@+-NE>B_9%!1mgkh`o6Bq8+%tGSh_j2OJk1g5ZO)9wa76h*Fe66bXJY~0&0D6{<> z#_GTI$1R;Kf_|I)=&$^?JJBBnmJ z=&JrBndjvYo0;h8@4qb$&<~PVA1zY?wE6!i-mG+m~ct>t4RZ{Qa}1R zz)5okbuKoRXM2WTbFs&E*>}Jh)v0eTm4hL(#U1aO@fA|rg^!rp@_#|-&{R?03cF6M zhBLh~keMQx1^*DFE}AL}efw{Jai4wZgQvR_E^TGadt`sUaX#_(FH=W;_J`l>w?N6e Yv^hfFE>JwH-d_(8XKH7sFgGvz7p=)7UH||9 diff --git a/data_utils.py b/data_utils.py index 21064b4..4855699 100644 --- a/data_utils.py +++ b/data_utils.py @@ -215,7 +215,7 @@ def get_audio(self, filename): self.sampling_rate, self.hop_length, self.win_length, center=False) spec = torch.squeeze(spec, 0) - torch.save(spec, spec_filename, _use_new_zipfile_serialization=False) + torch.save(spec, spec_filename) return spec, audio_norm def get_text(self, text): diff --git a/text/__init__.py b/text/__init__.py index d89aa75..48ae82f 100644 --- a/text/__init__.py +++ b/text/__init__.py @@ -1,13 +1,11 @@ """ from https://github.com/keithito/tacotron """ from text import cleaners +from text.symbols import symbols -def initialize(symbols): - # Mappings from symbol to numeric ID and vice versa: - global _symbol_to_id - global _id_to_symbol - _symbol_to_id = {s: i for i, s in enumerate(symbols)} - _id_to_symbol = {i: s for i, s in enumerate(symbols)} +# Mappings from symbol to numeric ID and vice versa: +_symbol_to_id = {s: i for i, s in enumerate(symbols)} +_id_to_symbol = {i: s for i, s in enumerate(symbols)} def text_to_sequence(text, cleaner_names): diff --git a/text/cleaners.py b/text/cleaners.py index 690585c..00db1c7 100644 --- a/text/cleaners.py +++ b/text/cleaners.py @@ -295,7 +295,44 @@ def japanese_cleaners(text): def japanese_cleaners2(text): - return japanese_cleaners(text).replace('ts','ʦ') + '''Pipeline for notating accent in Japanese text.''' + '''Reference https://r9y9.github.io/ttslearn/latest/notebooks/ch10_Recipe-Tacotron.html''' + sentences = re.split(_japanese_marks, text) + marks = re.findall(_japanese_marks, text) + text = '' + for i, sentence in enumerate(sentences): + if re.match(_japanese_characters, sentence): + if text!='': + text+=' ' + labels = pyopenjtalk.extract_fullcontext(sentence) + for n, label in enumerate(labels): + phoneme = re.search(r'\-([^\+]*)\+', label).group(1) + if phoneme not in ['sil','pau']: + text += phoneme.replace('ch','ʧ').replace('sh','ʃ').replace('cl','Q').replace('ts','ʦ') + else: + continue + n_moras = int(re.search(r'/F:(\d+)_', label).group(1)) + a1 = int(re.search(r"/A:(\-?[0-9]+)\+", label).group(1)) + a2 = int(re.search(r"\+(\d+)\+", label).group(1)) + a3 = int(re.search(r"\+(\d+)/", label).group(1)) + if re.search(r'\-([^\+]*)\+', labels[n + 1]).group(1) in ['sil','pau']: + a2_next=-1 + else: + a2_next = int(re.search(r"\+(\d+)\+", labels[n + 1]).group(1)) + # Accent phrase boundary + if a3 == 1 and a2_next == 1: + text += ' ' + # Falling + elif a1 == 0 and a2_next == a2 + 1 and a2 != n_moras: + text += '↓' + # Rising + elif a2 == 1 and a2_next == 2: + text += '↑' + if i