text-generation-webui/extensions/silero_tts/script.py

225 lines
9.4 KiB
Python
Raw Normal View History

import random
2023-03-12 17:36:18 +01:00
import time
2023-02-14 19:06:06 +01:00
from pathlib import Path
2023-02-24 23:00:11 +01:00
import gradio as gr
2023-03-22 19:55:03 +01:00
import torch
2023-05-11 20:37:04 +02:00
from extensions.silero_tts import tts_preprocessor
2023-08-13 06:12:15 +02:00
from modules import chat, shared, ui_chat
from modules.utils import gradio
2023-02-14 19:06:06 +01:00
torch._C._jit_set_profiling_mode(False)
2023-02-14 19:06:06 +01:00
params = {
'activate': True,
'speaker': 'en_56',
2023-02-14 19:06:06 +01:00
'language': 'en',
'model_id': 'v3_en',
'sample_rate': 48000,
'device': 'cpu',
2023-03-11 15:17:13 +01:00
'show_text': False,
'autoplay': True,
'voice_pitch': 'medium',
'voice_speed': 'medium',
2023-04-07 17:15:57 +02:00
'local_cache_path': '' # User can override the default cache path to something other via settings.json
2023-02-14 19:06:06 +01:00
}
2023-03-11 15:17:13 +01:00
2023-02-14 19:06:06 +01:00
current_params = params.copy()
2023-02-26 06:26:15 +01:00
voices_by_gender = ['en_99', 'en_45', 'en_18', 'en_117', 'en_49', 'en_51', 'en_68', 'en_0', 'en_26', 'en_56', 'en_74', 'en_5', 'en_38', 'en_53', 'en_21', 'en_37', 'en_107', 'en_10', 'en_82', 'en_16', 'en_41', 'en_12', 'en_67', 'en_61', 'en_14', 'en_11', 'en_39', 'en_52', 'en_24', 'en_97', 'en_28', 'en_72', 'en_94', 'en_36', 'en_4', 'en_43', 'en_88', 'en_25', 'en_65', 'en_6', 'en_44', 'en_75', 'en_91', 'en_60', 'en_109', 'en_85', 'en_101', 'en_108', 'en_50', 'en_96', 'en_64', 'en_92', 'en_76', 'en_33', 'en_116', 'en_48', 'en_98', 'en_86', 'en_62', 'en_54', 'en_95', 'en_55', 'en_111', 'en_3', 'en_83', 'en_8', 'en_47', 'en_59', 'en_1', 'en_2', 'en_7', 'en_9', 'en_13', 'en_15', 'en_17', 'en_19', 'en_20', 'en_22', 'en_23', 'en_27', 'en_29', 'en_30', 'en_31', 'en_32', 'en_34', 'en_35', 'en_40', 'en_42', 'en_46', 'en_57', 'en_58', 'en_63', 'en_66', 'en_69', 'en_70', 'en_71', 'en_73', 'en_77', 'en_78', 'en_79', 'en_80', 'en_81', 'en_84', 'en_87', 'en_89', 'en_90', 'en_93', 'en_100', 'en_102', 'en_103', 'en_104', 'en_105', 'en_106', 'en_110', 'en_112', 'en_113', 'en_114', 'en_115']
voice_pitches = ['x-low', 'low', 'medium', 'high', 'x-high']
voice_speeds = ['x-slow', 'slow', 'medium', 'fast', 'x-fast']
2023-02-14 19:06:06 +01:00
2023-03-11 15:17:13 +01:00
# Used for making text xml compatible, needed for voice pitch and speed control
table = str.maketrans({
"<": "&lt;",
">": "&gt;",
"&": "&amp;",
"'": "&apos;",
'"': "&quot;",
})
2023-03-11 15:17:13 +01:00
def xmlesc(txt):
return txt.translate(table)
2023-02-14 19:06:06 +01:00
def load_model():
2023-04-07 17:15:57 +02:00
torch_cache_path = torch.hub.get_dir() if params['local_cache_path'] == '' else params['local_cache_path']
model_path = torch_cache_path + "/snakers4_silero-models_master/src/silero/model/" + params['model_id'] + ".pt"
if Path(model_path).is_file():
print(f'\nUsing Silero TTS cached checkpoint found at {torch_cache_path}')
model, example_text = torch.hub.load(repo_or_dir=torch_cache_path + '/snakers4_silero-models_master/', model='silero_tts', language=params['language'], speaker=params['model_id'], source='local', path=model_path, force_reload=True)
else:
print(f'\nSilero TTS cache not found at {torch_cache_path}. Attempting to download...')
model, example_text = torch.hub.load(repo_or_dir='snakers4/silero-models', model='silero_tts', language=params['language'], speaker=params['model_id'])
2023-02-14 19:06:06 +01:00
model.to(params['device'])
return model
def remove_tts_from_history(history):
for i, entry in enumerate(history['internal']):
history['visible'][i] = [history['visible'][i][0], entry[1]]
return history
def toggle_text_in_history(history):
for i, entry in enumerate(history['visible']):
visible_reply = entry[1]
if visible_reply.startswith('<audio'):
if params['show_text']:
reply = history['internal'][i][1]
history['visible'][i] = [history['visible'][i][0], f"{visible_reply.split('</audio>')[0]}</audio>\n\n{reply}"]
else:
history['visible'][i] = [history['visible'][i][0], f"{visible_reply.split('</audio>')[0]}</audio>"]
return history
def state_modifier(state):
if not params['activate']:
return state
state['stream'] = False
return state
def input_modifier(string, state):
if not params['activate']:
return string
2023-03-11 15:17:13 +01:00
2023-05-21 18:24:54 +02:00
shared.processing_message = "*Is recording a voice message...*"
return string
def history_modifier(history):
# Remove autoplay from the last reply
2023-05-21 18:24:54 +02:00
if len(history['internal']) > 0:
history['visible'][-1] = [
history['visible'][-1][0],
history['visible'][-1][1].replace('controls autoplay>', 'controls>')
2023-05-21 17:11:59 +02:00
]
2023-02-14 19:06:06 +01:00
2023-05-21 18:24:54 +02:00
return history
2023-02-14 19:06:06 +01:00
def output_modifier(string, state):
global model, current_params, streaming_state
2023-02-14 19:06:06 +01:00
for i in params:
if params[i] != current_params[i]:
model = load_model()
current_params = params.copy()
break
if not params['activate']:
return string
original_string = string
string = tts_preprocessor.preprocess(string)
2023-02-14 19:06:06 +01:00
if string == '':
string = '*Empty reply, try regenerating*'
else:
output_file = Path(f'extensions/silero_tts/outputs/{state["character_menu"]}_{int(time.time())}.wav')
prosody = '<prosody rate="{}" pitch="{}">'.format(params['voice_speed'], params['voice_pitch'])
silero_input = f'<speak>{prosody}{xmlesc(string)}</prosody></speak>'
model.save_wav(ssml_text=silero_input, speaker=params['speaker'], sample_rate=int(params['sample_rate']), audio_path=str(output_file))
autoplay = 'autoplay' if params['autoplay'] else ''
string = f'<audio src="file/{output_file.as_posix()}" controls {autoplay}></audio>'
if params['show_text']:
string += f'\n\n{original_string}'
shared.processing_message = "*Is typing...*"
2023-02-14 19:06:06 +01:00
return string
2023-04-07 17:15:57 +02:00
def setup():
global model
model = load_model()
def random_sentence():
with open(Path("extensions/silero_tts/harvard_sentences.txt")) as f:
return random.choice(list(f))
def voice_preview(preview_text):
global model, current_params, streaming_state
for i in params:
if params[i] != current_params[i]:
model = load_model()
current_params = params.copy()
break
string = tts_preprocessor.preprocess(preview_text or random_sentence())
output_file = Path('extensions/silero_tts/outputs/voice_preview.wav')
prosody = f"<prosody rate=\"{params['voice_speed']}\" pitch=\"{params['voice_pitch']}\">"
silero_input = f'<speak>{prosody}{xmlesc(string)}</prosody></speak>'
model.save_wav(ssml_text=silero_input, speaker=params['speaker'], sample_rate=int(params['sample_rate']), audio_path=str(output_file))
return f'<audio src="file/{output_file.as_posix()}?{int(time.time())}" controls autoplay></audio>'
def custom_css():
path_to_css = Path(__file__).parent.resolve() / 'style.css'
return open(path_to_css, 'r').read()
2023-02-24 23:00:11 +01:00
def ui():
# Gradio elements
with gr.Accordion("Silero TTS"):
with gr.Row():
activate = gr.Checkbox(value=params['activate'], label='Activate TTS')
autoplay = gr.Checkbox(value=params['autoplay'], label='Play TTS automatically')
show_text = gr.Checkbox(value=params['show_text'], label='Show message text under audio player')
voice = gr.Dropdown(value=params['speaker'], choices=voices_by_gender, label='TTS voice')
with gr.Row():
v_pitch = gr.Dropdown(value=params['voice_pitch'], choices=voice_pitches, label='Voice pitch')
v_speed = gr.Dropdown(value=params['voice_speed'], choices=voice_speeds, label='Voice speed')
with gr.Row():
preview_text = gr.Text(show_label=False, placeholder="Preview text", elem_id="silero_preview_text")
preview_play = gr.Button("Preview")
preview_audio = gr.HTML(visible=False)
with gr.Row():
convert = gr.Button('Permanently replace audios with the message texts')
convert_cancel = gr.Button('Cancel', visible=False)
convert_confirm = gr.Button('Confirm (cannot be undone)', variant="stop", visible=False)
2023-08-13 06:12:15 +02:00
# Convert history with confirmation
convert_arr = [convert_confirm, convert, convert_cancel]
convert.click(lambda: [gr.update(visible=True), gr.update(visible=False), gr.update(visible=True)], None, convert_arr)
convert_confirm.click(
lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)], None, convert_arr).then(
remove_tts_from_history, gradio('history'), gradio('history')).then(
chat.save_persistent_history, gradio('history', 'character_menu', 'mode'), None).then(
chat.redraw_html, gradio(ui_chat.reload_arr), gradio('display'))
convert_cancel.click(lambda: [gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)], None, convert_arr)
# Toggle message text in history
show_text.change(
lambda x: params.update({"show_text": x}), show_text, None).then(
toggle_text_in_history, gradio('history'), gradio('history')).then(
chat.save_persistent_history, gradio('history', 'character_menu', 'mode'), None).then(
chat.redraw_html, gradio(ui_chat.reload_arr), gradio('display'))
2023-02-24 23:00:11 +01:00
# Event functions to update the parameters in the backend
activate.change(lambda x: params.update({"activate": x}), activate, None)
autoplay.change(lambda x: params.update({"autoplay": x}), autoplay, None)
2023-02-24 23:00:11 +01:00
voice.change(lambda x: params.update({"speaker": x}), voice, None)
v_pitch.change(lambda x: params.update({"voice_pitch": x}), v_pitch, None)
v_speed.change(lambda x: params.update({"voice_speed": x}), v_speed, None)
# Play preview
preview_text.submit(voice_preview, preview_text, preview_audio)
preview_play.click(voice_preview, preview_text, preview_audio)