日语语音合成系统

#! /usr/bin/python3

import PySimpleGUI as sg
import glob, os, tempfile, subprocess
import winsound
import datetime
# -----------------------------------


root=os.path.abspath(os.path.dirname(__file__))
os.chdir(root)

#open_jtalk-1.11，词典，声音模型需要自行下载
OJT_COMMAND = "open_jtalk.exe"
VOICE_DIR = "C:/open_talk/open_jtalk-1.11/voice/"
DIC_DIR = "C:/open_talk/open_jtalk-1.11/dic"
SUPPORT_PLAY = True


AUDIO_BUFFER_SIZE = 4096

PLAY_COMMAND = ["play/", "-q"]



TITLE = "OpenJTalk FrontEnd-SG"

CAPTION = 0	
MIN_VAL = 1	
MAX_VAL = 2	
DEF_VAL = 3	
RES_VAL = 4	
PARAM   = 5	
KEY     = 6	
param_info = [
    ("α ",      0.0,  1.0, 0.5, 0.01, "-a",  "-ALPHA-"),
    ("β  ",     0.0,  1.0, 0.0, 0.01, "-b",  "-BETA-"),
    ("速度 ",    0.0,  2.0, 1.0, 0.01, "-r",  "-SPEED-"),
    ("ﾊｰﾌﾄｰﾝ", -10.0, 10.0, 0.0, 0.1,  "-fm", "-HALFTONE-"),
    ("境界  ",   0.0,  1.0, 0.5, 0.01, "-u",  "-MSD-"),
    ("GV0   ",   0.0,  5.0, 1.0, 0.01, "-jm", "-GV0-"),
    ("GV1    ",  0.0,  5.0, 1.0, 0.01, "-jf", "-GV1-"),
    ("音量    ", 0.0, 20.0, 1.0, 0.1,  "-g",  "-VOLUME-"),
]


def splitMessage(value):
    return [n.strip() for n in value.splitlines() if n.strip()]

def makeOJTCommand(values, output_file_path, audio_buffer):
    cmd = [OJT_COMMAND,
           "-x", DIC_DIR,
           "-m", os.path.join(VOICE_DIR, values["-MODEL-"])]

    if output_file_path is not None:
        cmd = cmd + ["-ow", output_file_path]
    if audio_buffer is not None:
        cmd = cmd + ["-z", str(audio_buffer)]

    for n in param_info:
        auto_key = "-AUTO" + n[KEY]
        if auto_key in values and values[auto_key]:
            continue
        cmd.append(n[PARAM])
        cmd.append(str(values[n[KEY]]))

    return cmd

voice = [
    os.path.relpath(f, VOICE_DIR)
    for f
    in glob.iglob(os.path.join(VOICE_DIR, "**/*.htsvoice"), recursive=True)
]

if not voice:
    sg.Popup("音響モデルファイルが無い。", title=TITLE)
    exit()

frame_layout = [
    [sg.Text("音響モデル"), sg.Combo(voice, default_value=voice[0], readonly=True, key="-MODEL-")],
    [sg.Text(n[CAPTION], size=(7, 1), justification="right") for n in param_info],
    [sg.Slider((n[MIN_VAL], n[MAX_VAL]),
               orientation="vertical",
               default_value=n[DEF_VAL],
               resolution=n[RES_VAL],
               key=n[KEY])
     for n in param_info],
    [sg.Checkbox("auto", default=True, key="-AUTO-ALPHA-")]
]

layout = [
    [sg.Multiline(size=(64,10),default_text='おはようございます',key="-MES-")],
    [sg.Frame("パラメータ", frame_layout)],
    [sg.Submit(button_text="再生"),
     sg.Submit(button_text="保存"),
     sg.InputText('voice', size=(6,1),key="-PREFIX-")
    ]
]

window = sg.Window(TITLE, layout)

while True:
    event, values = window.read()

    if event is None:
        break

    if event == "再生":
        mes = splitMessage(values["-MES-"])

        if SUPPORT_PLAY:
            current_time = str(datetime.datetime.strftime(datetime.datetime.now(),'%H-%M-%S'))
            path = os.path.abspath(os.path.dirname(__file__))+"/play/audio_"+current_time+".wav"
            #print(path)
            cmd = makeOJTCommand(values, path, AUDIO_BUFFER_SIZE)
            #print(cmd)
            
            for m in mes:
                c = subprocess.Popen(cmd, stdin=subprocess.PIPE)
                c.stdin.write(m.encode('shift-jis'))
                c.stdin.close()
                c.wait()
            #winsound.PlaySound('open_jtalk.wav', winsound.SND_FILENAME)
        else:
            path = "open_jtalk.wav"

         
            try:
                cmd = makeOJTCommand(values, path, None)
                print (cmd)

                c = subprocess.Popen(cmd,stdin=subprocess.PIPE)

                data = (mes[0].encode('shift-jis'))
                print (data)


                c.stdin.write(data)
                c.stdin.close()
                c.wait()

                winsound.PlaySound('open_jtalk.wav', winsound.SND_FILENAME)
            
            finally:
                print ("error!")
        
    elif event == "保存":
        save_dir = sg.PopupGetFolder("保存先", title=TITLE)
        if save_dir:

            prefix = values["-PREFIX-"]
            mes = splitMessage(values["-MES-"])
            no = 0
            for m in mes:

                output_path = os.path.join(save_dir,"{0}{1:03d}.wav".format(prefix, no))
                no += 1
                cmd = makeOJTCommand(values, output_path, None)
                
                c = subprocess.Popen(cmd, stdin=subprocess.PIPE)
                c.stdin.write(m.encode('utf-8'))
                c.stdin.close()
                c.wait()
            sg.popup("完了")
        

window.close()

配置下载