func main() {
var src, ext, reg string
flag.StringVar(&src, "src", "./", "Source directory of the corpus")
flag.StringVar(&ext, "ext", "wav", "extension of the corpus with default (wav)")
flag.StringVar(®, "reg", "", "regular expression for searching files)")
flag.Parse()
fmt.Printf("src:%s\next:%s\nreg:%s\n", src, ext, reg)
if !kaldi.DirExist(src) {
fmt.Printf("Error: No Exist src directory: %s\n", src)
return
}
pattern := "*" + reg + "*." + ext
file_cmd := kaldi.JoinArgs(
"find",
src+"/",
"-type f ",
"-iname "+pattern)
fmt.Println("Find: ", file_cmd)
output, _ := kaldi.BashOutput(file_cmd)
files := strings.Split(string(output[:]), "\n")
total_len := 0.0
count := 0
for _, file := range files {
cmd := kaldi.JoinArgs(
"sox",
strings.Trim(file, "\n "),
"-n stat",
"2>&1",
"|awk '/Length/ {print $3}'")
output, err := kaldi.BashOutput(cmd)
if err != nil {
fmt.Println("Missed file:", file)
continue
}
length, err := strconv.ParseFloat(strings.Trim(string(output[:]), "\n \t"), 64)
if err != nil {
fmt.Println("Missed file:", file)
continue
}
count++
total_len += length
}
fmt.Println("Total Count: ", len(files))
fmt.Println("Total Len(hours): ", total_len/3600)
fmt.Println("Wave Count: ", count)
fmt.Println("Missed Count: ", len(files)-count)
}
func main() {
fmt.Println("update-feats-scp")
var data string
flag.StringVar(&data, "data", "", "data dir")
flag.Parse()
param := flag.Arg(0)
fmt.Println("param", param)
if !kaldi.DirExist(param) {
fmt.Println("No exist param dir:", param)
return
}
Clear(param)
filepath.Walk(param, DirCheck)
if data != "" {
ConstructData(data, param)
}
}
func deal_const(src, dst string) {
file := file_files(dst, "*")
if len(file) < 1 {
fmt.Println("No scp files, maybe something wrong!")
}
for _, f := range file {
name := path.Base(f)
parts := strings.Split(name, ".")
dir := path.Join(path.Dir(f), strings.Join(parts[:len(parts)-1], "."))
if kaldi.DirExist(dir) {
f_src := path.Join(src, path.Base(dir), parts[len(parts)-1])
os.Rename(f_src, path.Join(dir, parts[len(parts)-1]))
}
}
}
func main() {
fmt.Println("mk-mfcc")
flag.Parse()
if flag.NArg() < 2 {
log.Println("No enough args!")
return
}
src := flag.Arg(0)
dst := flag.Arg(1)
fmt.Println(src, dst)
if !kaldi.DirExist(src) {
log.Println("no exist source dir :", src)
return
}
var wg sync.WaitGroup
filepath.Walk(src, func(path string, info os.FileInfo, err error) error {
if !info.IsDir() {
return nil
}
wg.Add(1)
go func(dir string) {
defer wg.Done()
files, err := filepath.Glob(dir + "/*.wav")
if err != nil {
log.Println(err)
return
}
if len(files) == 0 {
return
}
rel, rerr := filepath.Rel(src, dir)
if rerr != nil {
log.Println(err)
return
}
dst_dir := filepath.Join(dst, rel)
MakeMfcc(files, dst_dir)
}(path)
return nil
})
wg.Wait()
}