app.py

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"):
    load_mes=gr.Info(f"""Loading Model...<br>{instr}""",duration=2)
    pp.load_mod(instr=instr) 
    with open(pp.json_ob,'r') as f:
        #json_ob=json.dumps(f.read(),indent=4)
        json_ob=f.read()
    load_mes=gr.Info(f"Model Loaded<br>{instr}",duration=2)
    return 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'{os.getcwd()}/saved/{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']))
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."""

def button_on(stream):
    if stream==True:
        return gr.update(interactive=True,visible=True),gr.update(interactive=False,visible=False)
    if stream==False:
        return gr.update(interactive=False,visible=False),gr.update(interactive=True,visible=True)
def clear_aud():
    return None
with gr.Blocks() as b:
    gr.HTML("""<h1 style='font-size:xxx-large;font-weight:900;'>Piper Fast TTS</h1>
    <h4>Piper: <a href='https://github.com/rhasspy/piper' target='_blank'>https://github.com/rhasspy/piper</a></h4>
    <h4>PyPiperTTS: <a href='https://github.com/broadfield-dev/PyPiperTTS' target='_blank'>https://github.com/broadfield-dev/PyPiperTTS</a></h4>
    """)
    with gr.Row():
        with gr.Column(scale=2):
            in_txt=gr.Textbox(label="Text",lines=10)
            names=gr.Dropdown()
            with gr.Row():
                stream_btn=gr.Button("Stream",interactive=True,visible=True)
                sub_btn=gr.Button(interactive=False,visible=False)
                cancel_btn=gr.Button("Stop")
            out_aud=gr.Audio(streaming=True, autoplay=True)        
        with gr.Column(scale=1):
            with gr.Accordion("Control"):
                stream=gr.Checkbox(label="Stream",info="Streaming is fast, but lower quality",value=True,interactive=True)
                length=gr.Slider(label="Length", minimum=0.01, maximum=10.0, value=1)
                noise=gr.Slider(label="Noise", minimum=0.01, maximum=3.0, value=1)
                width=gr.Slider(label="Noise Width", minimum=0.01, maximum=3.0, value=1)
                sen_pause=gr.Slider(label="Sentence Pause", minimum=0.1, maximum=10.0, value=1)
                with gr.Tab("Save Settings"):
                    save_btn=gr.Button("Save")
                    save_file=gr.File()
                with gr.Tab("Load Settings"):
                    load_file=gr.File()
        with gr.Column(scale=1):

            with gr.Accordion("Model Config"):
                json_ob=gr.JSON(label="JSON")
    f1=stream.change(button_on,stream,[stream_btn,sub_btn])
    f2=save_btn.click(save_set,[names,length,noise,width,sen_pause],save_file)
    f3=load_file.change(load_set,load_file,[names,length,noise,width,sen_pause])
    f4=names.change(load_mod,names,json_ob).then(clear_aud,None,out_aud)
    f5=stream_btn.click(clear_aud,None,out_aud)
    f6=stream_btn.click(pp.stream_tts,[in_txt,names,length,noise,width,sen_pause],out_aud)
    f7=sub_btn.click(clear_aud,None,out_aud)
    f8=sub_btn.click(pp.tts,[in_txt,names,length,noise,width,sen_pause],out_aud)
    cancel_btn.click(None,None,None,cancels=[f1,f2,f3,f4,f5,f6,f7,f8])
    b.load(init,None,names)

b.queue(default_concurrency_limit=20).launch(max_threads=40)