broadfield's picture
Create app.py
c1e96e4 verified
raw
history blame
3.18 kB
import gradio as gr
import subprocess
import os
import json
import uuid
import requests
from pypipertts import PyPiper
pp=PyPiper()
def init():
key_list=pp.key_list
return(gr.update(label="Voice",choices=key_list,value="en_US-joe-medium",interactive=True))
def load_mod(instr="en_US-joe-medium"):
pp.load_mod(instr=instr)
return pp.json_ob
def save_set(model,length,noise,width,sen_pause):
if not os.path.isdir(f'{os.getcwd()}/saved'):
os.mkdir(f'{os.getcwd()}/saved')
set_json={"model":model,"length":length,"noise":noise,"width":width,"pause":sen_pause}
file_name=f'{model}__{length}__{noise}__{width}__{sen_pause}'.replace(".","_")
with open(f'{os.getcwd()}/saved/{file_name}.json','w') as file:
file.write(json.dumps(set_json,indent=4))
file.close()
return(f'{file_name}.json')
def load_set(set_file):
with open(set_file,'r') as file:
set_json=json.loads(file.read())
file.close()
return(gr.update(value=set_json['model']),gr.update(value=set_json['length']),
gr.update(value=set_json['noise']),gr.update(value=set_json['width']),
gr.update(value=set_json['pause']))
def exp(exp_file):
txt="""PiperTTS is a powerful text-to-speech TTS node designed to convert written text into high-quality spoken audio. This node leverages advanced voice synthesis models to generate natural-sounding speech, making it an invaluable tool for AI developers looking to add a vocal element to their projects."""
exp_file=f"./example/en_US-ljspeech-high_2__21_0__88_0__22_2__6.json"
return(gr.update(value=txt),gr.update(value=exp_file))
with gr.Blocks() as b:
state1=gr.State()
with gr.Row():
with gr.Column(scale=1):
exp1=gr.Button("Example 1")
json_ob=gr.JSON()
with gr.Column(scale=2):
in_txt=gr.Textbox(label="Text",lines=10)
names=gr.Dropdown()
sub_btn=gr.Button()
out_aud=gr.Audio(streaming=True, autoplay=True)
with gr.Column(scale=1):
with gr.Accordion("Control"):
length=gr.Slider(label="Length", minimum=0.01, maximum=25.0, value=5)
noise=gr.Slider(label="Noise", minimum=0.01, maximum=3.0, value=0.1)
width=gr.Slider(label="Noise Width", minimum=0.01, maximum=5.0, value=0.3)
sen_pause=gr.Slider(label="Sentence Pause", minimum=0.1, maximum=50.0, value=5.0)
with gr.Tab("Save Settings"):
save_file=gr.File()
save_btn=gr.Button("Save")
save_btn.click(save_set,[names,length,noise,width,sen_pause],save_file)
with gr.Tab("Load Settings"):
load_file=gr.File()
load_file.change(load_set,load_file,[names,length,noise,width,sen_pause])
#load_btn=gr.Button("Load").click(load_set,load_file,[names,length,noise,width,sen_pause])
exp1.click(exp,None,[in_txt,load_file])
names.change(load_mod,names,json_ob)
sub_btn.click(pp.tts,[in_txt,names,length,noise,width,sen_pause],out_aud)
b.load(init,None,names)
b.launch()