app.py

import gradio as gr
import whisper
from transformers import pipeline

model = whisper.load_model("base")
sentiment_analysis = pipeline("sentiment-analysis", framework="pt", model="SamLowe/roberta-base-go_emotions")


import gradio as gr
import whisper
from transformers import pipeline
import gradio as gr
import pandas as pd
from io import StringIO
import os,re
from langchain.llms import OpenAI
import pandas as pd

from langchain.document_loaders import UnstructuredPDFLoader

from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain

from langchain.embeddings.openai import OpenAIEmbeddings
from langchain.vectorstores import Chroma
from langchain.text_splitter import CharacterTextSplitter
from langchain.llms import OpenAI
from langchain.chains import RetrievalQA
from langchain.document_loaders import TextLoader
from langchain.prompts import PromptTemplate
from langchain.callbacks.stdout import StdOutCallbackHandler
from langchain.chat_models.openai import ChatOpenAI
from langchain.prompts.prompt import PromptTemplate

from langchain.llms import OpenAI
from langchain.prompts import PromptTemplate
from langchain.chains import LLMChain
model = whisper.load_model("base")
sentiment_analysis = pipeline("sentiment-analysis", framework="pt", model="SamLowe/roberta-base-go_emotions")

def predict(text):
    # loader = UnstructuredPDFLoader(file_obj.orig_name)
    # data = loader.load()
    # text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
    # texts = text_splitter.split_documents(data)    
  
    # embeddings = OpenAIEmbeddings()
    # docsearch = Chroma.from_documents(texts, embeddings)
    # qa = RetrievalQA.from_chain_type(llm=OpenAI(), chain_type="map_reduce", retriever=docsearch.as_retriever())
    
    prompt_template = """Ignore all previous instructions. You are the world's hearing aid company markerting agent.
    I am going to give you a text of a customer. Analyze it and you have 4 products in list which you have to suggest to the customer: 
    ampli-mini it is mainly works for Maximum comfort and discretion, ampli-connect it is mainly works for Connected to the things you love, 
    ampli-energy it is mainly works for Full of energy, like you, ampli-easy it is mainly works for Allow yourself to hear well.
    You can also be creative, funny, or show emotions at time.
    also share the book a appointment link of your company https://www.amplifon.com/uk/book-an-appointment
Question: {question}
Product details:"""

    prompt_template_lang = """
    You are the world's best languages translator. Will give you some text or paragraph which you have to convert into Tamil, Hindi, Kannada
    and French.
    Input Text: {text}
    Tamil: 
    Hindi:
    Kannada:
    French:
    """
    PROMPT = PromptTemplate(
        template=prompt_template, input_variables=["question"]
    )
    PROMPT_lang = PromptTemplate(
        template=prompt_template_lang, input_variables=["text"]
    )
    # chain_type_kwargs = {"prompt": PROMPT}
    # qa = RetrievalQA.from_chain_type(llm=OpenAI(), chain_type="stuff", retriever=docsearch.as_retriever(), chain_type_kwargs=chain_type_kwargs)

    #Actually, Hi, how are you doing? Actually, I am looking for the hearing aid for my grandfather. He has like age around 62, 65 year old and one of the like major thing that I am looking for the hearing aid product which is like maximum comfort. So if you have anything in that category, so can you please tell me? Thank you.

    llm = OpenAI()
    # prompt = PromptTemplate(
    #     input_variables=["product"],
    #     template="What is a good name for a company that makes {product}?",
    # )
    chain = LLMChain(llm=llm, prompt=PROMPT)
    chain_lang = LLMChain(llm=llm, prompt=PROMPT_lang)
    
    resp = chain.run(question=text)
    resp_lang = chain_lang.run(text=resp)
    
    # print(resp)
    
    # response = []
    # category = ["ampli-mini", "ampli-connect", "ampli-energy", "ampli-easy"]
    # for value in category:

        # response.append({value:ai(qa, value)})

    # html_output = ""
    # for obj in response:
    # # Loop through the key-value pairs in the object
    #     for key, value in obj.items():
    #         value = re.sub(r'[\d\.]+', '', value)
    #         value_list = value.strip().split('\n')
    #         value_html = "<ol>"
    #         for item in value_list:
    #             value_html += "<li>{}</li>".format(item.strip())
    #         value_html += "</ol>"
    #         html_output += "<h2>{}</h2>".format(key)
    #         html_output += value_html

    
    return [resp, resp_lang]
    
# def ai(qa,category):
#     query = "please suggest "+ category +" interview questions"
#     data = list(filter(None, qa.run(query).split('\n')))
#     results = list(filter(lambda x: x != ' ', data))
#     results = "\n".join(results)
    
#     return results


def analyze_sentiment(text):
    results = sentiment_analysis(text)
    sentiment_results = {result['label']: result['score'] for result in results}
    return sentiment_results

def get_sentiment_emoji(sentiment):
    # Define the emojis corresponding to each sentiment
    emoji_mapping = {
        "disappointment": "😞",
        "sadness": "😢",
        "annoyance": "😠",
        "neutral": "😐",
        "disapproval": "👎",
        "realization": "😮",
        "nervousness": "😬",
        "approval": "👍",
        "joy": "😄",
        "anger": "😡",
        "embarrassment": "😳",
        "caring": "🤗",
        "remorse": "😔",
        "disgust": "🤢",
        "grief": "😥",
        "confusion": "😕",
        "relief": "😌",
        "desire": "😍",
        "admiration": "😌",
        "optimism": "😊",
        "fear": "😨",
        "love": "❤️",
        "excitement": "🎉",
        "curiosity": "🤔",
        "amusement": "😄",
        "surprise": "😲",
        "gratitude": "🙏",
        "pride": "🦁"
    }
    return emoji_mapping.get(sentiment, "")

def display_sentiment_results(sentiment_results, option):
    sentiment_text = ""
    for sentiment, score in sentiment_results.items():
        emoji = get_sentiment_emoji(sentiment)
        if option == "Sentiment Only":
            sentiment_text += f"{sentiment} {emoji}\n"
        elif option == "Sentiment + Score":
            sentiment_text += f"{sentiment} {emoji}: {score}\n"
    return sentiment_text

def inference(audio, sentiment_option):
    audio = whisper.load_audio(audio)
    audio = whisper.pad_or_trim(audio)

    mel = whisper.log_mel_spectrogram(audio).to(model.device)

    _, probs = model.detect_language(mel)
    lang = max(probs, key=probs.get)

    options = whisper.DecodingOptions(fp16=False)
    result = whisper.decode(model, mel, options)

    sentiment_results = analyze_sentiment(result.text)
    print(result.text)
    prediction  = predict(result.text)
    sentiment_output = display_sentiment_results(sentiment_results, sentiment_option)

    return lang.upper(), result.text, sentiment_output, prediction[0], prediction[1]



title = """<h1 align="center">🎤 Multilingual ASR 💬</h1>"""
description = """
💻 This demo showcases a general-purpose speech recognition model called Whisper. It is trained on a large dataset of diverse audio and supports multilingual speech recognition, speech translation, and language identification tasks.<br><br>
<br>
⚙️ Components of the tool:<br>
<br>
&nbsp;&nbsp;&nbsp;&nbsp; - Real-time multilingual speech recognition<br>
&nbsp;&nbsp;&nbsp;&nbsp; - Language identification<br>
&nbsp;&nbsp;&nbsp;&nbsp; - Sentiment analysis of the transcriptions<br>
<br>
🎯 The sentiment analysis results are provided as a dictionary with different emotions and their corresponding scores.<br>
<br>
😃 The sentiment analysis results are displayed with emojis representing the corresponding sentiment.<br>
<br>
✅ The higher the score for a specific emotion, the stronger the presence of that emotion in the transcribed text.<br>
<br>
❓ Use the microphone for real-time speech recognition.<br>
<br>
⚡️ The model will transcribe the audio and perform sentiment analysis on the transcribed text.<br>
"""

custom_css = """
#banner-image {
    display: block;
    margin-left: auto;
    margin-right: auto;
}
#chat-message {
    font-size: 14px;
    min-height: 300px;
}
"""

block = gr.Blocks(css=custom_css)

with block:
    gr.HTML(title)

    with gr.Row():
        with gr.Column():
            gr.HTML(description)

    with gr.Group():
        with gr.Box():
            audio = gr.Audio(
                label="Input Audio",
                show_label=False,
                source="microphone",
                type="filepath"
            )

            sentiment_option = gr.Radio(
                choices=["Sentiment Only", "Sentiment + Score"],
                label="Select an option",
                default="Sentiment Only"
            )

            btn = gr.Button("Transcribe")

        lang_str = gr.Textbox(label="Language")

        text = gr.Textbox(label="Transcription")

        sentiment_output = gr.Textbox(label="Sentiment Analysis Results", output=True)

        prediction = gr.Textbox(label="Prediction")
        
        language_translation = gr.Textbox(label="Language Translation")
        

        btn.click(inference, inputs=[audio, sentiment_option], outputs=[lang_str, text, sentiment_output, prediction,language_translation])

        gr.HTML('''
        <div class="footer">
            <p>Model by <a href="https://github.com/openai/whisper" style="text-decoration: underline;" target="_blank">OpenAI</a>
            </p>
        </div>
        ''') 

block.launch()