Skip to content

polly_lipsync

Purpose

Shows how to use the AWS SDK for Python (Boto3) with Amazon Polly and Tkinter to create a lip-sync application that displays an animated face speaking along with the speech synthesized by Amazon Polly. Lip-sync is accomplished by requesting a list of visemes from Amazon Polly that match up with the synthesized speech.

PollyMouth

A Tkinter application that lets a user enter text, select an Amazon Polly voice, and hear the text spoken by the selected voice while an animated face lip-syncs along with it.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
class PollyMouth:
    """
    A Tkinter application that lets a user enter text, select an Amazon Polly voice,
    and hear the text spoken by the selected voice while an animated face lip-syncs
    along with it.
    """

    # A dictionary of visemes mapped to image file names.
    lips = {
        'p': {'name': '.media/lips_m.png'},
        't': {'name': '.media/lips_c.png'},
        'S': {'name': '.media/lips_ch.png'},
        'T': {'name': '.media/lips_th.png'},
        'f': {'name': '.media/lips_f.png'},
        'k': {'name': '.media/lips_c.png'},
        'i': {'name': '.media/lips_e.png'},
        'r': {'name': '.media/lips_r.png'},
        's': {'name': '.media/lips_c.png'},
        'u': {'name': '.media/lips_w.png'},
        '@': {'name': '.media/lips_u.png'},
        'a': {'name': '.media/lips_a.png'},
        'e': {'name': '.media/lips_a.png'},
        'E': {'name': '.media/lips_u.png'},
        'o': {'name': '.media/lips_o.png'},
        'O': {'name': '.media/lips_u.png'},
        'sil': {'name': '.media/lips_sil.png'}
    }

    def __init__(self, polly_wrapper):
        """
        Initializes the main Tkinter window and adds all of the widgets needed for
        the application.

        :param polly_wrapper: An object that can call Amazon Polly API functions.
        """
        self.polly_wrapper = polly_wrapper
        self.app = tkinter.Tk()
        self.app.title("Amazon Polly Lip Sync")
        self.app.resizable(False, False)

        self.load_lips()

        choices_frame = tkinter.Frame(self.app)

        self.sayit_label = tkinter.Label(
            self.app, wraplength=410,
            text="Write some text in the box below, then click 'Say it!' "
                 "to hear and see your text.")
        self.sayit_txt = tkinter.Text(self.app, width=50, height=16)

        self.engine_label = tkinter.Label(choices_frame, text='Engine:')
        self.engine_var = tkinter.StringVar(choices_frame, 'neural')
        self.engine_options = tkinter.OptionMenu(
            choices_frame, self.engine_var, *sorted(polly_wrapper.get_voice_engines()),
            command=self.change_engine)

        self.language_label = tkinter.Label(choices_frame, text='Language:')
        self.language_var = tkinter.StringVar(choices_frame, 'US English')
        self.language_choices = polly_wrapper.get_languages(self.engine_var.get())
        self.language_options = tkinter.OptionMenu(
            choices_frame, self.language_var, *sorted(self.language_choices),
            command=self.change_language)

        self.voice_label = tkinter.Label(choices_frame, text='Voice:')
        self.voice_var = tkinter.StringVar(choices_frame, 'Joanna')
        self.voice_choices = polly_wrapper.get_voices(
            self.engine_var.get(), self.language_choices[self.language_var.get()])
        self.voice_options = tkinter.OptionMenu(
            choices_frame, self.voice_var, *sorted(self.voice_choices))

        self.face_canvas = tkinter.Canvas(
            choices_frame, height=100, width=200, bg='white')
        self.sayit_button = tkinter.Button(
            self.app, text="Say it!", command=self.say_it)

        self.loading_text = tkinter.Label(self.app, bg='white')

        self.app.geometry("635x320")

        self.sayit_label.grid(row=0)
        self.sayit_txt.grid(row=1, column=0)
        self.sayit_txt.focus_set()
        self.sayit_button.grid(row=2, pady=10, columnspan=2)
        self.sayit_button.configure(width=85, padx=10)

        choices_frame.grid(row=1, column=1, sticky=tkinter.N)
        self.engine_label.grid(row=0, column=0, sticky=tkinter.N, pady=10)
        self.engine_options.grid(row=0, column=1, sticky=tkinter.NW, padx=5, pady=10)
        self.engine_options.configure(width=18)
        self.language_label.grid(row=1, column=0, sticky=tkinter.N, pady=10)
        self.language_options.grid(row=1, column=1, sticky=tkinter.NW, padx=5, pady=10)
        self.language_options.configure(width=18)
        self.voice_label.grid(row=2, column=0, sticky=tkinter.N, pady=10)
        self.voice_options.grid(row=2, column=1, sticky=tkinter.NW, padx=5, pady=10)
        self.voice_options.configure(width=18)
        self.face_canvas.grid(row=3, columnspan=2, padx=10)

        self.face_canvas.create_image(100, 60, image=self.lips['sil']['image'])

        self.app.mainloop()

    def load_lips(self):
        """
        Loads lip-sync images either from a local '.media' folder or from GitHub
        and saves image data in a dictionary of visemes.
        """
        if os.path.isdir('.media'):
            logger.info("Found .media folder. Loading images from the local folder.")
            for viseme in self.lips:
                self.lips[viseme]['image'] = tkinter.PhotoImage(
                    file=self.lips[viseme]['name'])
        else:
            logger.info("No local .media folder. Trying to load images from GitHub.")
            for viseme in self.lips:
                url = GITHUB_URL + self.lips[viseme]['name']
                resp = requests.get(url)
                img = resp.content if resp.status_code == 200 else b''
                if resp.status_code != 200:
                    logger.warning("Couldn't load image from %s.", url)
                self.lips[viseme]['image'] = tkinter.PhotoImage(data=img)

    def change_engine(self, engine):
        """
        Handles the event that is fired when the selected engine type is changed in
        the UI. Updates the lists of available languages and voices that are supported
        for the selected engine type.

        :param engine: The newly selected engine type.
        """
        self.language_choices = self.polly_wrapper.get_languages(engine)
        lang_menu = self.language_options['menu']
        lang_menu.delete(0, 'end')
        sorted_choices = sorted(self.language_choices)
        for lang in sorted_choices:
            lang_menu.add_command(
                label=lang, command=lambda l=lang: self.change_language(l))
        self.change_language(sorted_choices[0])

    def change_language(self, language):
        """
        Handles the event that is fired when the selected language is changed in the
        UI. Updates the list of available voices that are available for the selected
        language and engine type.

        :param language: The newly selected language.
        """
        self.language_var.set(language)
        self.voice_choices = self.polly_wrapper.get_voices(
            self.engine_var.get(), self.language_choices[language])
        voice_menu = self.voice_options['menu']
        voice_menu.delete(0, 'end')
        sorted_choices = sorted(self.voice_choices)
        for voice in sorted_choices:
            voice_menu.add_command(
                label=voice, command=lambda v=voice: self.voice_var.set(v))
        self.voice_var.set(sorted_choices[0])

    def animate_lips(self, start_time, viseme, viseme_iter):
        """
        Animates the face that lip-syncs along with the synthesized speech. This
        uses the list of visemes and their associated timings that is returned from
        Amazon Polly. The image associated with a viseme is displayed and the next
        viseme is scheduled to display at the time indicated in the list of visemes.

        :param start_time: The time the animation is started. This is used to
                           calculate the time to wait until the next viseme image
                           is displayed.
        :param viseme: The current viseme to display.
        :param viseme_iter: An iterator that yields visemes from the list returned
                            from Amazon Polly.
        """
        try:
            mouth = self.lips.get(viseme['value'], self.lips['sil'])
            self.face_canvas.create_image(
                100, 60, image=mouth['image'])
            self.app.update()
            next_viseme = next(viseme_iter)
            next_time = start_time + next_viseme['time']
            cur_time = time.time_ns() // 1000000  # milliseconds
            wait_time = max(0, next_time - cur_time)
            logger.info("Vis: %s, cur_time %s, wait_time %s", mouth,
                        cur_time - start_time, wait_time)
            self.app.after(
                wait_time, self.animate_lips, start_time, next_viseme, viseme_iter)
        except StopIteration:
            pass

    def long_text_wait_callback(self, task_type, task_status):
        """
        A callback function that displays status while waiting for an asynchronous
        long text speech synthesis task to complete.

        :param task_type: The type of synthesis task (either 'speech' or 'viseme').
        :param task_status: The status of the task.
        """
        self.loading_text.grid(row=0, rowspan=4, columnspan=2, sticky=tkinter.NSEW)
        self.loading_text.configure(
            text=f"Waiting for {task_type}. Current status: {task_status}.")
        self.app.update()
        if task_status in ('completed', 'failed'):
            self.app.after(1000)
            self.loading_text.grid_forget()

    def say_it(self):
        """
        Gets synthesized speech and visemes from Amazon Polly, stores the audio in
        a temporary file, and plays the sound and lip-sync animation.

        When the text is too long for synchronous synthesis, this function displays a
        dialog that asks the user for an Amazon Simple Storage Service (Amazon S3)
        bucket to use for output storage, starts an asynchronous synthesis task, and
        waits for the task to complete.
        """
        audio_stream = None
        visemes = []
        try:
            audio_stream, visemes = self.polly_wrapper.synthesize(
                self.sayit_txt.get(1.0, tkinter.END),
                self.engine_var.get(),
                self.voice_choices[self.voice_var.get()],
                'mp3',
                self.language_choices[self.language_var.get()],
                True)
        except ClientError as error:
            if error.response['Error']['Code'] == 'TextLengthExceededException':
                bucket_name = tkinter.simpledialog.askstring(
                    "Text too long",
                    "The text is too long for synchronous synthesis. To start an\n"
                    "asynchronous job, enter the name of an existing Amazon S3\n"
                    "bucket to use for speech synthesis output and click OK.",
                    parent=self.app)
                if bucket_name:
                    audio_stream, visemes = self.polly_wrapper.do_synthesis_task(
                        self.sayit_txt.get(1.0, tkinter.END),
                        self.engine_var.get(),
                        self.voice_choices[self.voice_var.get()],
                        'mp3',
                        bucket_name,
                        self.language_choices[self.language_var.get()],
                        True,
                        self.long_text_wait_callback)

        logger.debug("Visemes: %s.", json.dumps(visemes))

        if audio_stream is not None:
            with TemporaryDirectory() as tempdir:
                speech_file_name = tempdir + '/speech.mp3'
                with open(speech_file_name, 'wb') as speech_file:
                    speech_file.write(audio_stream.read())
                silence = '.media/silence.mp3'
                if not os.path.isdir('.media'):
                    silence = GITHUB_URL + silence
                # Play a short silent audio file to ensure playsound is loaded and
                # ready. Without this, the audio tends to lag behind viseme playback.
                # playsound(silence)
                playsound(speech_file_name, block=False)
                start_time = time.time_ns() // 1000000
                self.app.after(
                    0, self.animate_lips, start_time, {'value': 'sil'}, iter(visemes))

__init__(polly_wrapper)

Initializes the main Tkinter window and adds all of the widgets needed for the application.

:param polly_wrapper: An object that can call Amazon Polly API functions.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
def __init__(self, polly_wrapper):
    """
    Initializes the main Tkinter window and adds all of the widgets needed for
    the application.

    :param polly_wrapper: An object that can call Amazon Polly API functions.
    """
    self.polly_wrapper = polly_wrapper
    self.app = tkinter.Tk()
    self.app.title("Amazon Polly Lip Sync")
    self.app.resizable(False, False)

    self.load_lips()

    choices_frame = tkinter.Frame(self.app)

    self.sayit_label = tkinter.Label(
        self.app, wraplength=410,
        text="Write some text in the box below, then click 'Say it!' "
             "to hear and see your text.")
    self.sayit_txt = tkinter.Text(self.app, width=50, height=16)

    self.engine_label = tkinter.Label(choices_frame, text='Engine:')
    self.engine_var = tkinter.StringVar(choices_frame, 'neural')
    self.engine_options = tkinter.OptionMenu(
        choices_frame, self.engine_var, *sorted(polly_wrapper.get_voice_engines()),
        command=self.change_engine)

    self.language_label = tkinter.Label(choices_frame, text='Language:')
    self.language_var = tkinter.StringVar(choices_frame, 'US English')
    self.language_choices = polly_wrapper.get_languages(self.engine_var.get())
    self.language_options = tkinter.OptionMenu(
        choices_frame, self.language_var, *sorted(self.language_choices),
        command=self.change_language)

    self.voice_label = tkinter.Label(choices_frame, text='Voice:')
    self.voice_var = tkinter.StringVar(choices_frame, 'Joanna')
    self.voice_choices = polly_wrapper.get_voices(
        self.engine_var.get(), self.language_choices[self.language_var.get()])
    self.voice_options = tkinter.OptionMenu(
        choices_frame, self.voice_var, *sorted(self.voice_choices))

    self.face_canvas = tkinter.Canvas(
        choices_frame, height=100, width=200, bg='white')
    self.sayit_button = tkinter.Button(
        self.app, text="Say it!", command=self.say_it)

    self.loading_text = tkinter.Label(self.app, bg='white')

    self.app.geometry("635x320")

    self.sayit_label.grid(row=0)
    self.sayit_txt.grid(row=1, column=0)
    self.sayit_txt.focus_set()
    self.sayit_button.grid(row=2, pady=10, columnspan=2)
    self.sayit_button.configure(width=85, padx=10)

    choices_frame.grid(row=1, column=1, sticky=tkinter.N)
    self.engine_label.grid(row=0, column=0, sticky=tkinter.N, pady=10)
    self.engine_options.grid(row=0, column=1, sticky=tkinter.NW, padx=5, pady=10)
    self.engine_options.configure(width=18)
    self.language_label.grid(row=1, column=0, sticky=tkinter.N, pady=10)
    self.language_options.grid(row=1, column=1, sticky=tkinter.NW, padx=5, pady=10)
    self.language_options.configure(width=18)
    self.voice_label.grid(row=2, column=0, sticky=tkinter.N, pady=10)
    self.voice_options.grid(row=2, column=1, sticky=tkinter.NW, padx=5, pady=10)
    self.voice_options.configure(width=18)
    self.face_canvas.grid(row=3, columnspan=2, padx=10)

    self.face_canvas.create_image(100, 60, image=self.lips['sil']['image'])

    self.app.mainloop()

animate_lips(start_time, viseme, viseme_iter)

Animates the face that lip-syncs along with the synthesized speech. This uses the list of visemes and their associated timings that is returned from Amazon Polly. The image associated with a viseme is displayed and the next viseme is scheduled to display at the time indicated in the list of visemes.

:param start_time: The time the animation is started. This is used to calculate the time to wait until the next viseme image is displayed. :param viseme: The current viseme to display. :param viseme_iter: An iterator that yields visemes from the list returned from Amazon Polly.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
def animate_lips(self, start_time, viseme, viseme_iter):
    """
    Animates the face that lip-syncs along with the synthesized speech. This
    uses the list of visemes and their associated timings that is returned from
    Amazon Polly. The image associated with a viseme is displayed and the next
    viseme is scheduled to display at the time indicated in the list of visemes.

    :param start_time: The time the animation is started. This is used to
                       calculate the time to wait until the next viseme image
                       is displayed.
    :param viseme: The current viseme to display.
    :param viseme_iter: An iterator that yields visemes from the list returned
                        from Amazon Polly.
    """
    try:
        mouth = self.lips.get(viseme['value'], self.lips['sil'])
        self.face_canvas.create_image(
            100, 60, image=mouth['image'])
        self.app.update()
        next_viseme = next(viseme_iter)
        next_time = start_time + next_viseme['time']
        cur_time = time.time_ns() // 1000000  # milliseconds
        wait_time = max(0, next_time - cur_time)
        logger.info("Vis: %s, cur_time %s, wait_time %s", mouth,
                    cur_time - start_time, wait_time)
        self.app.after(
            wait_time, self.animate_lips, start_time, next_viseme, viseme_iter)
    except StopIteration:
        pass

change_engine(engine)

Handles the event that is fired when the selected engine type is changed in the UI. Updates the lists of available languages and voices that are supported for the selected engine type.

:param engine: The newly selected engine type.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
def change_engine(self, engine):
    """
    Handles the event that is fired when the selected engine type is changed in
    the UI. Updates the lists of available languages and voices that are supported
    for the selected engine type.

    :param engine: The newly selected engine type.
    """
    self.language_choices = self.polly_wrapper.get_languages(engine)
    lang_menu = self.language_options['menu']
    lang_menu.delete(0, 'end')
    sorted_choices = sorted(self.language_choices)
    for lang in sorted_choices:
        lang_menu.add_command(
            label=lang, command=lambda l=lang: self.change_language(l))
    self.change_language(sorted_choices[0])

change_language(language)

Handles the event that is fired when the selected language is changed in the UI. Updates the list of available voices that are available for the selected language and engine type.

:param language: The newly selected language.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
def change_language(self, language):
    """
    Handles the event that is fired when the selected language is changed in the
    UI. Updates the list of available voices that are available for the selected
    language and engine type.

    :param language: The newly selected language.
    """
    self.language_var.set(language)
    self.voice_choices = self.polly_wrapper.get_voices(
        self.engine_var.get(), self.language_choices[language])
    voice_menu = self.voice_options['menu']
    voice_menu.delete(0, 'end')
    sorted_choices = sorted(self.voice_choices)
    for voice in sorted_choices:
        voice_menu.add_command(
            label=voice, command=lambda v=voice: self.voice_var.set(v))
    self.voice_var.set(sorted_choices[0])

load_lips()

Loads lip-sync images either from a local '.media' folder or from GitHub and saves image data in a dictionary of visemes.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
def load_lips(self):
    """
    Loads lip-sync images either from a local '.media' folder or from GitHub
    and saves image data in a dictionary of visemes.
    """
    if os.path.isdir('.media'):
        logger.info("Found .media folder. Loading images from the local folder.")
        for viseme in self.lips:
            self.lips[viseme]['image'] = tkinter.PhotoImage(
                file=self.lips[viseme]['name'])
    else:
        logger.info("No local .media folder. Trying to load images from GitHub.")
        for viseme in self.lips:
            url = GITHUB_URL + self.lips[viseme]['name']
            resp = requests.get(url)
            img = resp.content if resp.status_code == 200 else b''
            if resp.status_code != 200:
                logger.warning("Couldn't load image from %s.", url)
            self.lips[viseme]['image'] = tkinter.PhotoImage(data=img)

long_text_wait_callback(task_type, task_status)

A callback function that displays status while waiting for an asynchronous long text speech synthesis task to complete.

:param task_type: The type of synthesis task (either 'speech' or 'viseme'). :param task_status: The status of the task.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
def long_text_wait_callback(self, task_type, task_status):
    """
    A callback function that displays status while waiting for an asynchronous
    long text speech synthesis task to complete.

    :param task_type: The type of synthesis task (either 'speech' or 'viseme').
    :param task_status: The status of the task.
    """
    self.loading_text.grid(row=0, rowspan=4, columnspan=2, sticky=tkinter.NSEW)
    self.loading_text.configure(
        text=f"Waiting for {task_type}. Current status: {task_status}.")
    self.app.update()
    if task_status in ('completed', 'failed'):
        self.app.after(1000)
        self.loading_text.grid_forget()

say_it()

Gets synthesized speech and visemes from Amazon Polly, stores the audio in a temporary file, and plays the sound and lip-sync animation.

When the text is too long for synchronous synthesis, this function displays a dialog that asks the user for an Amazon Simple Storage Service (Amazon S3) bucket to use for output storage, starts an asynchronous synthesis task, and waits for the task to complete.

Source code in backend/app/utils/tts/backends/resources/aws_example_code/polly_lipsync.py 
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
def say_it(self):
    """
    Gets synthesized speech and visemes from Amazon Polly, stores the audio in
    a temporary file, and plays the sound and lip-sync animation.

    When the text is too long for synchronous synthesis, this function displays a
    dialog that asks the user for an Amazon Simple Storage Service (Amazon S3)
    bucket to use for output storage, starts an asynchronous synthesis task, and
    waits for the task to complete.
    """
    audio_stream = None
    visemes = []
    try:
        audio_stream, visemes = self.polly_wrapper.synthesize(
            self.sayit_txt.get(1.0, tkinter.END),
            self.engine_var.get(),
            self.voice_choices[self.voice_var.get()],
            'mp3',
            self.language_choices[self.language_var.get()],
            True)
    except ClientError as error:
        if error.response['Error']['Code'] == 'TextLengthExceededException':
            bucket_name = tkinter.simpledialog.askstring(
                "Text too long",
                "The text is too long for synchronous synthesis. To start an\n"
                "asynchronous job, enter the name of an existing Amazon S3\n"
                "bucket to use for speech synthesis output and click OK.",
                parent=self.app)
            if bucket_name:
                audio_stream, visemes = self.polly_wrapper.do_synthesis_task(
                    self.sayit_txt.get(1.0, tkinter.END),
                    self.engine_var.get(),
                    self.voice_choices[self.voice_var.get()],
                    'mp3',
                    bucket_name,
                    self.language_choices[self.language_var.get()],
                    True,
                    self.long_text_wait_callback)

    logger.debug("Visemes: %s.", json.dumps(visemes))

    if audio_stream is not None:
        with TemporaryDirectory() as tempdir:
            speech_file_name = tempdir + '/speech.mp3'
            with open(speech_file_name, 'wb') as speech_file:
                speech_file.write(audio_stream.read())
            silence = '.media/silence.mp3'
            if not os.path.isdir('.media'):
                silence = GITHUB_URL + silence
            # Play a short silent audio file to ensure playsound is loaded and
            # ready. Without this, the audio tends to lag behind viseme playback.
            # playsound(silence)
            playsound(speech_file_name, block=False)
            start_time = time.time_ns() // 1000000
            self.app.after(
                0, self.animate_lips, start_time, {'value': 'sil'}, iter(visemes))