ppp-audio-tagger/window.py

import tkinter as tk
import vlc
from os import listdir, remove
from os.path import isfile, join
import json
import shutil

# Set up IBM speech to text
from ibm_watson import SpeechToTextV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('b373X-km7u5pAaz2JoizXigcVFZFEB8CIntgYgWzbCQ4')
speech_to_text = SpeechToTextV1(
    authenticator=authenticator
)

speech_to_text.set_service_url('https://api.us-south.speech-to-text.watson.cloud.ibm.com/instances/0e7a3edf-309c-4e64-b345-8251781245e4')
speech_to_text.set_disable_ssl_verification(True)

# Set up Arpabet dict
def ARPA(text):
	if len(text) == 0:
		return "␤"
	out = ''
	for word_ in text.split(" "):
		word=word_; end_chars = ''
		while any(elem in word for elem in r"!?,.;") and len(word) > 1:
			if word[-1] == '!': end_chars = '!' + end_chars; word = word[:-1]
			if word[-1] == '?': end_chars = '?' + end_chars; word = word[:-1]
			if word[-1] == ',': end_chars = ',' + end_chars; word = word[:-1]
			if word[-1] == '.': end_chars = '.' + end_chars; word = word[:-1]
			if word[-1] == ';': end_chars = ';' + end_chars; word = word[:-1]
			else: break
		try: word_arpa = thisdict[word.upper()]
		except: word_arpa = ''
		if len(word_arpa)!=0: word = "{" + str(word_arpa) + "}"
		out = (out + " " + word + end_chars).strip()
	if out[-1] != "␤": out = out + "␤"
	return out

thisdict = {}   # And load it
for line in reversed((open('merged.dict_1.1.txt', "r").read()).splitlines()):
    thisdict[(line.split(" ",1))[0]] = (line.split(" ",1))[1].strip()

# Stuff for dataset path
IDX = 0
MYPATH = "./dataset/in"
OUTPATH = './dataset/'

# f = open("{}filelist.txt".format(OUTPATH), "w")
# f.write("")
# f.close()

# f = open("{}filelist_arpa.txt".format(OUTPATH), "w")
# f.write("")
# f.close()

onlyfiles = [f for f in listdir(MYPATH) if isfile(join(MYPATH, f))]
if len(onlyfiles) > 0:
        img = onlyfiles[0]

def keyEvent(event):
	global IDX
	global v
	if (event.keysym == 'Left'):
		IDX = IDX - 1

		if IDX < 0:
			IDX = len(onlyfiles) - 1

		w.config(text = onlyfiles[IDX])
		v.set("")

	if (event.keysym == 'Right'):
		IDX = IDX + 1

		if IDX >= len(onlyfiles):
			IDX = 0

		w.config(text = onlyfiles[IDX])
		v.set("")

def comp_s(event):
	global v
	print(v.get())
	v.set("")

def playTrack(event):
	player = vlc.MediaPlayer("{}/{}".format(MYPATH, onlyfiles[IDX]))
	player.play()

def transcribeTrack(event):
	global v
	with open("{}/{}".format(MYPATH, onlyfiles[IDX]),
               'rb') as audio_file:
		speech_recognition_results = speech_to_text.recognize(
        	audio=audio_file,
	        content_type='audio/wav',
    	    word_alternatives_threshold=0.9,
	    ).get_result()
	# print(json.dumps(speech_recognition_results, indent=2))
	v.set(speech_recognition_results["results"][0]["alternatives"][0]["transcript"])

def saveTrack(event):
	f = open("{}filelist.txt".format(OUTPATH), "a")
	f.write("wavs/out/{}|{}\n".format(onlyfiles[IDX], o1.cget('text')))
	f.close()

	f = open("{}filelist_arpa.txt".format(OUTPATH), "a")
	f.write("wavs/out/{}|{}\n".format(onlyfiles[IDX], o2.cget('text')))
	f.close()

	move_audio_file(IDX, 'out')

def discardTrack(event):
	move_audio_file(IDX, 'discard')

def callback(sv):
	o1.config(text = "{}␤".format(sv.get().strip()), wraplength=500)
	o2.config(text = ARPA(sv.get().strip()), wraplength=500)

def move_audio_file(idx, destination):
	file_path = onlyfiles[idx]
	onlyfiles.pop(idx)
	idx = idx - 1
	if idx < 0:
		idx = len(onlyfiles) - 1

	shutil.copyfile(MYPATH + '/' + file_path, OUTPATH + destination + '/' + file_path)
	remove(MYPATH + '/' + file_path)
	global v
	w.config(text = onlyfiles[idx])
	v.set("")


mw = tk.Tk()

mw.title('Training Data - {} remaining'.format(len(onlyfiles)))
mw.geometry('500x200')
mw.configure(bg='black')


toolbar = tk.Frame(master=mw, width='500', height='24', borderwidth=2, bg='slategray4', relief='raised')
toolbar.pack()
play_btn = tk.Button(toolbar, text="(C-p) Play track", command=None)
play_btn.pack(side='left')
save_btn = tk.Button(toolbar, text="(C-s) Save track", command=None)
save_btn.pack(side='left')
transcribe_btn = tk.Button(toolbar, text="(C-t) Transcribe track", command=None)
transcribe_btn.pack(side='left')
discard_btn = tk.Button(toolbar, text="(C-d) Discard track", command=None)
discard_btn.pack(side='left')

back = tk.Frame(master=mw, bg='black')
back.bind('<Key>', keyEvent)
back.bind("<Control-p>", playTrack)
back.bind("<Control-t>", transcribeTrack)
back.bind("<Control-s>", saveTrack)
back.bind("<Control-d>", discardTrack)
back.pack_propagate(0)
back.pack(fill=tk.BOTH, expand=1)
back.focus_set()

# canvas = tk.Canvas(master=back, width='1280', height='720', bg='black', highlightthickness=0)  
# canvas.pack()

w = tk.Label(master=back, text=onlyfiles[IDX], bg='black', fg='white')
w.pack()

v = tk.StringVar()
v.trace("w", lambda name, index, mode, sv=v: callback(v))
e = tk.Entry(master=back, textvariable=v, width='500', bg='black', fg='white', highlightbackground='grey')
e.pack()
e.bind('<Return>', comp_s)
e.bind("<Control-p>", playTrack)
e.bind("<Control-t>", transcribeTrack)
e.bind("<Control-s>", saveTrack)
e.bind("<Control-d>", discardTrack)

o1 = tk.Label(master=back, text="text here", bg='black', fg='white')
o1.pack()
o2 = tk.Label(master=back, text="arpabet here", bg='black', fg='white')
o2.pack()

mw.mainloop()