app.py

import os
import gradio as gr

from langchain import PromptTemplate
from langchain.chains import LLMChain
from langchain.llms import OpenAI

from gradio_client import Client
eleven = Client("https://elevenlabs-tts.hf.space/")

import wave
from pydub import AudioSegment

openai_api_key = os.environ.get("OPENAI_API_KEY")

llm = OpenAI(temperature=0.9, max_tokens=-1)

def split_text_into_sentences(text):
    sentences = []
    current_sentence = ''
    words = text.split()

    for word in words:
        current_sentence += ' ' + word
        if word.endswith('.'):
            sentences.append(current_sentence.strip())
            current_sentence = ''

    if current_sentence:
        sentences.append(current_sentence.strip())

    return sentences

def join_wav_files(input_files, output_file):
    # Open the first input file to get its parameters
    with wave.open(input_files[0], 'rb') as first_file:
        # Get the audio parameters from the first file
        params = first_file.getparams()

        # Create a new wave file for writing the joined audio
        with wave.open(output_file, 'wb') as output:
            output.setparams(params)

            # Iterate over the input files and write their audio data to the output file
            for input_file in input_files:
                with wave.open(input_file, 'rb') as input:
                    output.writeframes(input.readframes(input.getnframes()))

def generate_story(text, lang, age):
    """Generate a story using the langchain library and OpenAI's GPT-3 model."""
    prompt = PromptTemplate(
        input_variables=["text", "lang", "age"],
        template=""" 
        You are a fun and seasoned storyteller. 
        Generate a short bedtime story for a {age} years old audience about {text}.
        Your story must be written in {lang}.
        Use short sentences. The story is not too long, but not too short either.
        """
    )
    story = LLMChain(llm=llm, prompt=prompt)
    story_result = story.run(text=text, lang=lang, age=age)
    print(story_result)
    print("""
    —
    Cutting text in chunks
    —
    """)
    input_waves = []
    
    text_chunks = split_text_into_sentences(story_result)
    for chunk in text_chunks:
        print(chunk)
        result = eleven.predict(
				chunk,	# str representing input in 'Input Text (250 characters max)' Textbox component
				"Bella",	# str representing input in 'Voice' Dropdown component
				"eleven_multilingual_v1",	# str representing input in 'Model' Radio component
				fn_index=0
        )
        print(result)
        input_waves.append(result)
    
    output_wav = 'output.wav'

    join_wav_files(input_waves, output_wav)

    # Load the WAV file
    wav_file = AudioSegment.from_file("output.wav", format="wav")

    # Export it as an MP3 file
    mp3_file = "output.mp3"
    wav_file.export(mp3_file, format="mp3")
    
    return story_result, 'output.mp3'

def app(text, lang, age):
    story = generate_story(text, lang, age)
    return story

css = """
#col-container {max-width: 510px; margin-left: auto; margin-right: auto;}
a {text-decoration-line: underline; font-weight: 600;}
"""
with gr.Blocks(css=css) as demo:
    with gr.Column(elem_id="col-container"):
        gr.HTML("""<div style="text-align: center; max-width: 700px; margin: 0 auto;">
                <div
                style="
                    display: inline-flex;
                    align-items: center;
                    gap: 0.8rem;
                    font-size: 1.75rem;
                "
                >
                <h1 style="font-weight: 900; margin-bottom: 7px; margin-top: 5px;">
                    Bedtime Story
                </h1>
                </div>
                <p style="margin-bottom: 10px; font-size: 94%">
                Generate a bedtime story for a 5 years old audience who needs to get some sleep 😴  
                </p>
            </div>""")
        
        with gr.Group():           
            text = gr.Textbox(label="Subject: what the story should be about ?", info="Will generate and tell a story about {your text input}")
            with gr.Row():
                lang = gr.Dropdown(label="Pick a language", choices=["English", "French", "German", "Hindi", "Italian", "Polish", "Portuguese", "Spanish"], value="English")
                age = gr.Dropdown(label="Age target", choices=["3","4","5","6","7"], value="5")
        submit_btn = gr.Button('Submit')
        
        audio = gr.Audio(label="The story audio told")
        story = gr.Textbox(label="The story text")

    submit_btn.click(fn=app, inputs=[text, lang, age], outputs=[story, audio])
        
demo.launch()