app.py

import os
import openai
import whisper
import gradio as gr

openai.api_key = os.environ.get('SessionToken')

whisper_model = whisper.load_model("small")

conversation = ""
user_name = "MH"
bot_name = "bbDemo"

def chat_hf(audio):
    conversation = ""
    try:
        whisper_text = translate(audio)
        user_input = whisper_text

        # Conversation route
        prompt = user_name + ": " + user_input + "\n" + bot_name+ ": "
        conversation += prompt  # allows for context
        # fetch the response from open AI api
        response = openai.Completion.create(engine='text-davinci-003', prompt=conversation, max_tokens=50)
        response_str = response["choices"][0]["text"].replace("\n", "")
        response_str = response_str.split(user_name + ": ", 1)[0].split(bot_name + ": ", 1)[0]
        
        conversation += response_str + "\n"
    
        gpt_response = response_str

    except:
        # Conversation route
        whisper_text = translate(audio)
        user_input = whisper_text
        prompt = user_name + ": " + user_input + "\n" + bot_name+ ": "
        conversation += prompt  # allows for context
        # fetch the response from open AI api
        response = openai.Completion.create(engine='text-davinci-003', prompt=conversation, max_tokens=1024)
        response_str = response["choices"][0]["text"].replace("\n", "")
        response_str = response_str.split(user_name + ": ", 1)[0].split(bot_name + ": ", 1)[0]
        
        conversation += response_str + "\n"
    
        gpt_response = response_str
        print("Error")
 
    
    return whisper_text, gpt_response


def translate(audio):
    print("""
    —
    Sending audio to Whisper ...
    —
    """)
    
    audio = whisper.load_audio(audio)
    audio = whisper.pad_or_trim(audio)
    
    mel = whisper.log_mel_spectrogram(audio).to(whisper_model.device)
    
    _, probs = whisper_model.detect_language(mel)
    
    transcript_options = whisper.DecodingOptions(task="transcribe", fp16 = False)    
    transcription = whisper.decode(whisper_model, mel, transcript_options)
    
    print("language spoken: " + transcription.language)
    print("transcript: " + transcription.text)
    print("———————————————————————————————————————————")  

    return transcription.text

title = """
    <div style="text-align: center; max-width: 500px; margin: 0 auto;">
        <div
        style="
            display: inline-flex;
            align-items: center;
            gap: 0.8rem;
            font-size: 1.75rem;
            margin-bottom: 10px;
        "
        >
        <h1 style="font-weight: 600; margin-bottom: 7px;">
            Whisper to chatGPT
        </h1>
        </div>
        <p style="margin-bottom: 10px;font-size: 94%;font-weight: 100;line-height: 1.5em;">
        Chat with GPT with your voice in your native language!
        <br />
        </p>
        <p style="font-size: 94%">
            <span style="display: flex;align-items: center;justify-content: center;height: 30px;">
            </span>
        </p>
    </div>
"""

article = """
    <div class="footer">
        <p><a href="https://chat.openai.com/chat" target="_blank">chatGPT</a> 
        by <a href="https://openai.com/" style="text-decoration: underline;" target="_blank">OpenAI</a> - 
        Gradio Demo by 🤗 <a href="https://englishphonetics.net/" target="_blank">Eriberto Oliveira</a>
        </p>
    </div>
"""

css = '''
    #col-container {max-width: 700px; margin-left: auto; margin-right: auto;}
    a {text-decoration-line: underline; font-weight: 600;}
    .footer {
            margin-bottom: 45px;
            margin-top: 35px;
            text-align: center;
            border-bottom: 1px solid #e5e5e5;
        }
        .footer>p {
            font-size: .8rem;
            display: inline-block;
            padding: 0 10px;
            transform: translateY(10px);
            background: white;
        }
        .dark .footer {
            border-color: #303030;
        }
        .dark .footer>p {
            background: #0b0f19;
        }
'''
 

with gr.Blocks(css=css) as demo:
    
    with gr.Column(elem_id="col-container"):
        
        gr.HTML(title)
        
        with gr.Row():
            record_input = gr.Audio(source="microphone",type="filepath", show_label=False)
            send_btn = gr.Button("Send my message !")
   
    with gr.Column():
        audio_translation = gr.Textbox(type="text",label="Whisper transcription")
        gpt_response = gr.Textbox(type="text",label="chatGPT response")

        gr.HTML(article)
    
    send_btn.click(chat_hf, inputs=[record_input], outputs=[audio_translation, gpt_response])

demo.queue(max_size=32, concurrency_count=20).launch(debug=True)