--- trunk/mbrola.pl	2006/08/05 10:54:32	1
+++ trunk/mbrola.pl	2006/08/08 15:06:33	5
@@ -7,11 +7,7 @@
 use strict;
 use Data::Dump qw/dump/;
 use Term::ReadLine;
-
-my $term = new Term::ReadLine 'Simple Perl calc';
-my $prompt = "Pričaj: ";
-my $OUT = $term->OUT || \*STDOUT;
-
+use File::Slurp;
 
 my $letters = {
 	'a' => [['a'],
@@ -194,7 +190,7 @@
 };
 
 my $recovery;
-foreach my $df (qw/ bp oo uks/) {
+foreach my $df (qw/bp oo uks/) {
 	$recovery->{$df}++;
 }
 
@@ -210,41 +206,44 @@
 	# FIXME: lj, nj, dľ
 	my @chars = split(//, $text);
 
-	my @pho = ({ char => '_', dur => [50] });
+	my @pho = ({ char => '_', dur => [ $silence->{word} ] });
 
 	my ($g,$f) = ('','');
 
-	foreach my $i ( 0 .. $#chars ) {
-		my $c = $chars[$i];
+	my $last_c = '';
+
+	my $i = 0;
+	while (@chars) {
+		my $c = shift @chars;
 
 		$g .= $c;
 
-		if (defined( $phonemes->{$c} )) {
-			$c = $phonemes->{$c};
-		} elsif (defined( $token_to_grapheme->{$c} )) {
-			$c = $token_to_grapheme->{$c};
-		} else {
-			$pho[ $#pho ]->{dur}->[0] = $silence->{word};
-			next;
+		if (defined( $phonemes->{ lc($c) } )) {
+			$c = $phonemes->{ lc($c) };
+		} elsif (defined( $token_to_grapheme->{ lc($c) } )) {
+			my $tmp = $token_to_grapheme->{ lc($c) };
+			if (length($tmp) > 1) {
+				my @tmp_c = split(//, $tmp);
+				warn "### $c --> $tmp\n";
+				$c = shift @tmp_c;
+				unshift @chars, ( @tmp_c );
+			} else {
+				$c = $tmp;
+			}
 		}
 
-		my $d = $durations->{$c} || $silence->{word};
-
-		my @dur = ( $d );
-
-		if ($first) {
-			$first = 0;
+		my $d = $durations->{$c} || $durations->{ lc($c) };
 
-			push @dur, ( 10, 120 );
-		}
-
-		if ($zarez) {
-			$zarez = 0;
-			push @dur, ( 10, 100 );
+		if (! $d) {
+			next if ($c =~ m/\s/);
+			warn "### skipped: $c\n";
+			$c = '_';
+			$d = $silence->{word};
 		}
 
+		my @dur = ( $d );
 
-		if ($c =~ m/[,\.!\? _]/) {
+		if ($last_c =~ m/[,\.!\? _]/) {
 
 			my $from = $#{ $pho[ $i - 1 ]->{dur} };
 			$from = 3 if ($from > 3);
@@ -256,16 +255,41 @@
 				$tmpr += 30;
 			}
 
-			if ($c =~ m/,/) {
-				$zarez = 1;
-				@dur = ( $silence->{comma} );
-			} else {
-				@dur = ( $silence->{sentence} );
-				$first = 1;
-			}
+			# begining of sentence
+			push @dur, ( 10, 120 );
+		}
 
+		if ($c =~ m/\s/) {
+			$pho[ $#pho ]->{dur}->[0] += $silence->{word};
+			$last_c = $c;
+			next;
+		} elsif ($c =~ m/[\.!\?]/) {
+			$pho[ $#pho ]->{dur}->[0] += $silence->{word};
+			push @pho, {
+				char => '_',
+				dur => [ $silence->{sent} ],
+			};
+			$last_c = $c;
+			next;
+		} elsif ($c =~ m/,/) {
+			push @dur, ( 10, 100 );
 		}
 
+		# same last chars? double duration
+		if ($last_c eq $c) {
+			$pho[ $#pho ]->{dur}->[0] *= 2;
+			next;
+		}
+
+		# fixup sequences that need special handling
+		if (defined($recovery->{ $last_c . $c })) {
+			push @pho, {
+				char => '_',
+				dur => [ $silence->{word} ],
+			};
+		}
+
+		$last_c = $c;
 		push @pho, {
 			char => $c,
 			dur => \@dur,
@@ -274,21 +298,14 @@
 		$f .= $c;
 	}
 
-	push @pho, { char => '_', dur => [50] };
+	push @pho, { char => '_', dur => [ $silence->{sent} ] };
 
-	warn "# pho = ",dump(@pho),$/;
+#	warn "# pho = ",dump(@pho),$/;
 
 	my $out;
 
-	my $last_c = '';
-
 	foreach my $p (@pho) {
-		if (defined($recovery->{ $last_c . $p->{char} })) {
-			$out .= '_ ' . ( $silence->{word} * $speed ) . "\n";
-		}
 		$out .= $p->{char} . ' ' . join(' ', map { $_ * $speed } @{ $p->{dur} }) . "\n";
-
-		$last_c = $p->{char};
 	}
 
 	return ($out, $g, $f);
@@ -296,21 +313,47 @@
 
 my $mbrola = './bin/mbrola-linux-i386 ./cr1/cr1';
 
-while ( defined ($_ = $term->readline($prompt)) ) {
+my $term = new Term::ReadLine 'Mbrola croatian speaker';
+my $OUT = $term->OUT || \*STDOUT;
 
-	$term->addhistory($_);
+sub play_speak_hr {
+	my $text = shift || return;
 
-	my ($out,$g,$f) = speak_hr( $_ );
+	my ($out,$g,$f) = speak_hr( $text );
 
 	open(my $fh, "| $mbrola - tmp.wav") || die "can't open $mbrola: $!";
 
-	print $OUT ">>> $g\n>>> $f\n$out\n";
+	print $OUT ">>> $g\n<<< $f\n";
 
 	print $fh $out || die "can't pipe to $mbrola";
 	close($fh) || die "error closing pipe to $mbrola";
 
+	$out =~ s/\n/ | /gs;
+	print $OUT "# $out\n";
+
 	system 'play tmp.wav';
+}
+
+
+if (my $path = shift @ARGV) {
+	my $text = read_file($path) || die "can't read $path: $!";
+
+	# strip html
+	$text =~ s!</?[^>]+>! !gs;
+
+	$text =~ s!\s+! !gs;
+
+	print "-- $text --";
 
+	play_speak_hr( $text );
+	exit;
+}
+
+my $prompt = "Pričaj: ";
+
+while ( defined ($_ = $term->readline($prompt)) ) {
+	$term->addhistory( $_ );
+	play_speak_hr( $_ );
 }