| 1 |
#!/usr/bin/perl -w |
| 2 |
|
| 3 |
# mbrola.pl |
| 4 |
# |
| 5 |
# 08/04/2006 07:54:54 PM CEST Dobrica Pavlinusic <dpavlin@rot13.org> |
| 6 |
|
| 7 |
use strict; |
| 8 |
use Data::Dump qw/dump/; |
| 9 |
use Term::ReadLine; |
| 10 |
|
| 11 |
my $term = new Term::ReadLine 'Simple Perl calc'; |
| 12 |
my $prompt = "Prièaj: "; |
| 13 |
my $OUT = $term->OUT || \*STDOUT; |
| 14 |
|
| 15 |
|
| 16 |
my $letters = { |
| 17 |
'a' => [['a'], |
| 18 |
[[170,]]], |
| 19 |
'A' => [['a'], |
| 20 |
[[170,]]], |
| 21 |
'b' => [['b', 'e', 'b', 'e',], |
| 22 |
[[90,], [15,], [1,], [150,]]], |
| 23 |
'B' => [['b', 'e', 'b', 'e',], |
| 24 |
[[90,], [15,], [1,], [150,]]], |
| 25 |
'c' => [['ts', 'a'], |
| 26 |
[[30,], [100,]]], |
| 27 |
'C' => [['ts', 'a'], |
| 28 |
[[30,], [100,]]], |
| 29 |
'è' => [['tS', 'a'], |
| 30 |
[[30,], [100,]]], |
| 31 |
'È' => [['tS', 'a'], |
| 32 |
[[30,], [100,]]], |
| 33 |
'æ' => [['tS\'', 'a'], |
| 34 |
[[30,], [100,]]], |
| 35 |
'Æ' => [['tS\'', 'a'], |
| 36 |
[[30,], [100,]]], |
| 37 |
'd' => [['d', 'e', 'd', 'e'], |
| 38 |
[[70,], [15,], [60,], [1, 90, 100]]], |
| 39 |
'D' => [['d', 'e', 'd', 'e'], |
| 40 |
[[70,], [15,], [60,], [1, 90, 100]]], |
| 41 |
'ð'=> [['dZ\'', 'a'], |
| 42 |
[[100,], [110,]]], |
| 43 |
'Ð'=> [['dZ\'', 'a'], |
| 44 |
[[100,], [110,]]], |
| 45 |
'e' => [['e', 'x', 'e'], |
| 46 |
[[150, 10, 100], [1,], [50, 50, 90]]], |
| 47 |
'E' => [['e', 'x', 'e'], |
| 48 |
[[150, 10, 100], [1,], [50, 50, 90]]], |
| 49 |
'f' => [['f', 'e'], |
| 50 |
[[100, 10, 100], [100,]]], |
| 51 |
'F' => [['f', 'e'], |
| 52 |
[[100, 10, 100], [100,]]], |
| 53 |
'g' => [['g', 'a'], |
| 54 |
[[70,], [150,]]], |
| 55 |
'G' => [['g', 'a'], |
| 56 |
[[70,], [150,]]], |
| 57 |
#'h' => [['x', 'r', 'a'], |
| 58 |
# [[120,], [1,], [170,]]], |
| 59 |
'h' => [['x', 'a', 'x', 'a'], |
| 60 |
[[150,], [170,], [1,], [170,]]], |
| 61 |
'H' => [['x', 'a', 'x', 'a'], |
| 62 |
[[150,], [170,], [1,], [170,]]], |
| 63 |
'i' => [['i'], |
| 64 |
[[150,]]], |
| 65 |
'I' => [['i'], |
| 66 |
[[150,]]], |
| 67 |
'j' => [['j', 'a', 'r'], |
| 68 |
[[60,], [10,], [50,]]], |
| 69 |
'J' => [['j', 'a', 'r'], |
| 70 |
[[60,], [10,], [50,]]], |
| 71 |
'k' => [['k', 'a'], |
| 72 |
[[70,], [150,]]], |
| 73 |
'K' => [['k', 'a'], |
| 74 |
[[70,], [150,]]], |
| 75 |
#'l' => [['l', 'e'], |
| 76 |
# [[100, 10, 100], [100,]]], |
| 77 |
'l' => [['e', 'l'], |
| 78 |
[[100,], [100,10, 100]]], |
| 79 |
'L' => [['e', 'l'], |
| 80 |
[[100,], [100,10, 100]]], |
| 81 |
#'m' => [['m', 'e'], |
| 82 |
# [[90,], [50,]]], |
| 83 |
'm' => [['e', 'm'], |
| 84 |
[[50,], [90,]]], |
| 85 |
'M' => [['e', 'm'], |
| 86 |
[[50,], [90,]]], |
| 87 |
#'n' => [['n', 'e'], |
| 88 |
# [[90,], [50,]]], |
| 89 |
'n' => [['e', 'n'], |
| 90 |
[[50,], [90,]]], |
| 91 |
'N' => [['e', 'n'], |
| 92 |
[[50,], [90,]]], |
| 93 |
'o' => [['o', 'x'], |
| 94 |
[[71,], [70,]]], |
| 95 |
'O' => [['o', 'x'], |
| 96 |
[[71,], [70,]]], |
| 97 |
#'p' => [['p', 'a'], |
| 98 |
# [[70,], [150,]]], |
| 99 |
'p' => [['p', 'e'], |
| 100 |
[[90,], [120,]]], |
| 101 |
'P' => [['p', 'e'], |
| 102 |
[[90,], [120,]]], |
| 103 |
#'r' => [['r', 'x'], |
| 104 |
# [[90,], [110,]]], |
| 105 |
'r' => [['e', 'r', 'x'], |
| 106 |
[[150,10,100], [90,], [110,]]], |
| 107 |
'R' => [['e', 'r', 'x'], |
| 108 |
[[150,10,100], [90,], [110,]]], |
| 109 |
's' => [['s', 'a'], |
| 110 |
[[90,], [110,]]], |
| 111 |
'S' => [['s', 'a'], |
| 112 |
[[90,], [110,]]], |
| 113 |
'¹' => [['S', 'a'], |
| 114 |
[[70,], [110,]]], |
| 115 |
'©' => [['S', 'a'], |
| 116 |
[[70,], [110,]]], |
| 117 |
#'t' => [['t', 'a'], |
| 118 |
# [[90,], [120,]]], |
| 119 |
't' => [['t', 'e'], |
| 120 |
[[90,], [150,]]], |
| 121 |
'T' => [['t', 'e'], |
| 122 |
[[90,], [150,]]], |
| 123 |
'u' => [['u'], |
| 124 |
[[170,]]], |
| 125 |
'U' => [['u'], |
| 126 |
[[170,]]], |
| 127 |
'v' => [['v', 'a'], |
| 128 |
[[100,], [120,]]], |
| 129 |
'V' => [['v', 'a'], |
| 130 |
[[100,], [120,]]], |
| 131 |
'z' => [['z', 'a'], |
| 132 |
[[100,], [110,]]], |
| 133 |
'Z' => [['z', 'a'], |
| 134 |
[[100,], [110,]]], |
| 135 |
'¾' => [['Z', 'a'], |
| 136 |
[[100,], [110,]]], |
| 137 |
'®' => [['Z', 'a'], |
| 138 |
[[100,], [110,]]], |
| 139 |
'x' => [['i', 'k', 's'], |
| 140 |
[[60,], [60,], [50,]]], |
| 141 |
'X' => [['i', 'k', 's'], |
| 142 |
[[60,], [60,], [50,]]], |
| 143 |
'y' => [['i', 'p', 's', 'i', 'l', 'o', 'n', 'e'], |
| 144 |
[[60, 10, 90, 70, 110], |
| 145 |
[70, 10, 100], |
| 146 |
[50,], [40,], [60,], [70,], [70,], [10,]]], |
| 147 |
'Y' => [['i', 'p', 's', 'i', 'l', 'o', 'n', 'e'], |
| 148 |
[[60, 10, 90, 70, 110], |
| 149 |
[70, 10, 100], |
| 150 |
[50,], [40,], [60,], [70,], [70,], [10,]]], |
| 151 |
'q' => [['k', 'u', 'x'], |
| 152 |
[[70,], [110,], [1,]]], |
| 153 |
'Q' => [['k', 'u', 'x'], |
| 154 |
[[70,], [110,], [1,]]], |
| 155 |
'w' => [['d', 'u', 'p', 'l', 'o', '_', 'v', 'e'], |
| 156 |
[[54, 50, 100], [50, 50, 110], [85, 10, 100], [35,], |
| 157 |
[54, 90, 95], [10,], [100, 90, 110], [120, 20, 95]]], |
| 158 |
'W' => [['d', 'u', 'p', 'l', 'o', '_', 'v', 'e'], |
| 159 |
[[54, 50, 100], [50, 50, 110], [85, 10, 100], [35,], |
| 160 |
[54, 90, 95], [10,], [100, 90, 110], [120, 20, 95]]]} |
| 161 |
; |
| 162 |
|
| 163 |
my $phonemes = { |
| 164 |
'a' => 'a', 'b' => 'b', 'c' => 'ts', 'è' => 'tS', 'æ' => 'tS\'', 'd' => 'd', |
| 165 |
'd¾' => 'dZ', 'ð'=> 'dZ\'', 'e' => 'e', 'f' => 'f', 'g' => 'g', 'h' => 'x', |
| 166 |
'i' => 'i', 'j' => 'j', 'k' => 'k', 'l' => 'l', 'lj'=> 'L', 'm' => 'm', |
| 167 |
'n' => 'n', 'nj'=> 'J', 'o' => 'o', 'p' => 'p', 'r' => 'r', 's' => 's', |
| 168 |
'¹' => 'S', 't' => 't', 'u' => 'u', 'v' => 'v', 'z' => 'z', '¾' => 'Z' |
| 169 |
}; |
| 170 |
|
| 171 |
my $durations = { |
| 172 |
'a' => 61, 'b' => 65, 'ts' => 113, 'tS' => 90, 'tS\''=> 98, 'd' => 54, |
| 173 |
'dZ' => 56, 'dZ\''=> 61, 'e' => 53, 'f' => 86, 'g' => 56, 'x' => 68, |
| 174 |
'i' => 49, 'j' => 53, 'k' => 81, 'l' => 35, 'L' => 59, 'm' => 56, |
| 175 |
'n' => 45, 'J' => 60, 'o' => 54, 'p' => 85, 'r' => 25, 's' => 91, |
| 176 |
'S' => 99, 't' => 76, 'u' => 50, 'v' => 40, 'z' => 68, 'Z' => 74, |
| 177 |
# pause |
| 178 |
'_' => 150 } |
| 179 |
; |
| 180 |
|
| 181 |
my $token_to_grapheme = { |
| 182 |
'x' => 'ks', |
| 183 |
'q' => 'k', |
| 184 |
'w' => 'v', |
| 185 |
'y' => 'j', |
| 186 |
}; |
| 187 |
|
| 188 |
my $silence = { |
| 189 |
'word' => 40, |
| 190 |
'sent' => 100, |
| 191 |
'comma' => 180, |
| 192 |
'hard' => 100, |
| 193 |
'spell' => 150 |
| 194 |
}; |
| 195 |
|
| 196 |
my $recovery; |
| 197 |
foreach my $df (qw/ bp oo uks/) { |
| 198 |
$recovery->{$df}++; |
| 199 |
} |
| 200 |
|
| 201 |
sub speak_hr { |
| 202 |
|
| 203 |
my $text = shift @_ || 'ovo je na¹a difonska sinteza govora. kako vam se sviða? mo¾da èak i radi!'; |
| 204 |
|
| 205 |
my $first = 1; |
| 206 |
my $zarez = 0; |
| 207 |
|
| 208 |
my $speed = 1.2; |
| 209 |
|
| 210 |
# FIXME: lj, nj, d¾ |
| 211 |
my @chars = split(//, $text); |
| 212 |
|
| 213 |
my @pho = ({ char => '_', dur => [50] }); |
| 214 |
|
| 215 |
my ($g,$f) = ('',''); |
| 216 |
|
| 217 |
foreach my $i ( 0 .. $#chars ) { |
| 218 |
my $c = $chars[$i]; |
| 219 |
|
| 220 |
$g .= $c; |
| 221 |
|
| 222 |
if (defined( $phonemes->{$c} )) { |
| 223 |
$c = $phonemes->{$c}; |
| 224 |
} elsif (defined( $token_to_grapheme->{$c} )) { |
| 225 |
$c = $token_to_grapheme->{$c}; |
| 226 |
} else { |
| 227 |
$pho[ $#pho ]->{dur}->[0] = $silence->{word}; |
| 228 |
next; |
| 229 |
} |
| 230 |
|
| 231 |
my $d = $durations->{$c} || $silence->{word}; |
| 232 |
|
| 233 |
my @dur = ( $d ); |
| 234 |
|
| 235 |
if ($first) { |
| 236 |
$first = 0; |
| 237 |
|
| 238 |
push @dur, ( 10, 120 ); |
| 239 |
} |
| 240 |
|
| 241 |
if ($zarez) { |
| 242 |
$zarez = 0; |
| 243 |
push @dur, ( 10, 100 ); |
| 244 |
} |
| 245 |
|
| 246 |
|
| 247 |
if ($c =~ m/[,\.!\? _]/) { |
| 248 |
|
| 249 |
my $from = $#{ $pho[ $i - 1 ]->{dur} }; |
| 250 |
$from = 3 if ($from > 3); |
| 251 |
|
| 252 |
my $tmpr = 80; |
| 253 |
|
| 254 |
foreach my $j ( -($from) .. 0 ) { |
| 255 |
$pho[ $i - 1 ]->{dur}->[$j] = $tmpr; |
| 256 |
$tmpr += 30; |
| 257 |
} |
| 258 |
|
| 259 |
if ($c =~ m/,/) { |
| 260 |
$zarez = 1; |
| 261 |
@dur = ( $silence->{comma} ); |
| 262 |
} else { |
| 263 |
@dur = ( $silence->{sentence} ); |
| 264 |
$first = 1; |
| 265 |
} |
| 266 |
|
| 267 |
} |
| 268 |
|
| 269 |
push @pho, { |
| 270 |
char => $c, |
| 271 |
dur => \@dur, |
| 272 |
}; |
| 273 |
|
| 274 |
$f .= $c; |
| 275 |
} |
| 276 |
|
| 277 |
push @pho, { char => '_', dur => [50] }; |
| 278 |
|
| 279 |
warn "# pho = ",dump(@pho),$/; |
| 280 |
|
| 281 |
my $out; |
| 282 |
|
| 283 |
my $last_c = ''; |
| 284 |
|
| 285 |
foreach my $p (@pho) { |
| 286 |
if (defined($recovery->{ $last_c . $p->{char} })) { |
| 287 |
$out .= '_ ' . ( $silence->{word} * $speed ) . "\n"; |
| 288 |
} |
| 289 |
$out .= $p->{char} . ' ' . join(' ', map { $_ * $speed } @{ $p->{dur} }) . "\n"; |
| 290 |
|
| 291 |
$last_c = $p->{char}; |
| 292 |
} |
| 293 |
|
| 294 |
return ($out, $g, $f); |
| 295 |
} |
| 296 |
|
| 297 |
my $mbrola = './bin/mbrola-linux-i386 ./cr1/cr1'; |
| 298 |
|
| 299 |
while ( defined ($_ = $term->readline($prompt)) ) { |
| 300 |
|
| 301 |
$term->addhistory($_); |
| 302 |
|
| 303 |
my ($out,$g,$f) = speak_hr( $_ ); |
| 304 |
|
| 305 |
open(my $fh, "| $mbrola - tmp.wav") || die "can't open $mbrola: $!"; |
| 306 |
|
| 307 |
print $OUT ">>> $g\n>>> $f\n$out\n"; |
| 308 |
|
| 309 |
print $fh $out || die "can't pipe to $mbrola"; |
| 310 |
close($fh) || die "error closing pipe to $mbrola"; |
| 311 |
|
| 312 |
system 'play tmp.wav'; |
| 313 |
|
| 314 |
} |
| 315 |
|
| 316 |
|
| 317 |
__END__ |
| 318 |
|
| 319 |
foreach my $c (split(//, $text)) { |
| 320 |
warn "# c = $c\n"; |
| 321 |
|
| 322 |
if (! defined($letters->{$c})) { |
| 323 |
print "_ 50\n"; |
| 324 |
next; |
| 325 |
} |
| 326 |
|
| 327 |
my $slova = $letters->{$c}->[0]; |
| 328 |
my $param = $letters->{$c}->[1]; |
| 329 |
warn "# ",dump($slova, $param), $/; |
| 330 |
foreach my $i ( 0 ... $#$slova ) { |
| 331 |
print $slova->[$i], " ", join(" ",@{ $param->[$i] }), "\n"; |
| 332 |
} |
| 333 |
|
| 334 |
} |
Parent Directory
| 
Revision Log