luk
9742f6dab2
将 .sh/.bat 脚本与 .js/.plist/.Dockerfile 等文件中以 - 分隔的文件名 改为 _ 分隔，dcloud-renew 目录改为 dcloud、script_nuance 目录改为 nuance，并同步更新脚本内部引用与 README 路径。 Co-Authored-By: AtomCode (deepseek-v4-flash) <noreply@atomgit.com>
26 lines
655 B
Perl
26 lines
655 B
Perl
print "Usage: perl $0 transcription_file path_prefix output_file\n";
|
|
print "Example: perl $0 /direct/datadigest/en_sg/ives/000.txt /direct/datadigest/en_sg/ives/calls/ ~/ives.list\n";
|
|
|
|
open TRAN, $ARGV[0] or die "Cannot open transcription file $ARGV[0] for read.\n";
|
|
open CORPUS, ">>$ARGV[2]" or die "Cannot open corpus file $ARGV[1] for write.\n";
|
|
|
|
if ($ARGV[1] =~ m|/$|) # the parameter "path_prefix" is ended with /
|
|
{
|
|
$prefix = $ARGV[1];
|
|
}else
|
|
{
|
|
$prefix = "$ARGV[1]/";
|
|
}
|
|
|
|
while (<TRAN>)
|
|
{
|
|
chomp;
|
|
m|([^ ]+)\s+(.*)|;
|
|
if ($1 =~ m|\.info$|)
|
|
{
|
|
print CORPUS "new_speaker\n";
|
|
next;
|
|
}
|
|
print CORPUS "$prefix$1.ulaw\t$2\n";
|
|
}
|