app.py

import gradio as gr
#from pyChatGPT import ChatGPT
import os
import requests
api = os.environ.get('API_ENDPOINT')
#session_token = os.environ.get('SessionToken') 
#cf_clearance_token = os.environ.get('ClearanceToken')
#cf_bm_token = os.environ.get('cf_bm_token')
whisper = gr.Interface.load(name="spaces/sanchit-gandhi/whisper-large-v2")
 
def call_api(message):
    response = requests.get(f'{api}?q={message}')
    if response.status_code == 200:
        
        return str(response.text).split('\n', 2)[2]
    else:
        return """Sorry, I'm quite busy right now, but please try again later :)"""

def chat_hf(audio, task):

    try:
        whisper_text = translate(audio, task)
        if whisper_text == "ERROR: You have to either use the microphone or upload an audio file":
            gpt_response = "MISSING AUDIO: Record your voice by clicking the microphone button, do not forget to stop recording before sending your message ;)"
        else:
            gpt_response = call_api(whisper_text)
        #api = ChatGPT(session_token, cf_clearance_token, cf_bm_token)
        #api = ChatGPT(session_token)
        #api.refresh_auth()  # refresh the authorization token
        #if reset_conversation:
        #   
        #    api.reset_conversation()  # reset the conversation
        #resp = api.send_message(whisper_text)
        #gpt_response = resp['message']

    except:
        
        
        gpt_response = """Sorry, I'm quite busy right now, but please try again later :)"""
    
    print(f"""
    {whisper_text}
    ————
    {gpt_response}
    """)
    
    return whisper_text, gpt_response


def translate(audio, task):

    if task == "transcribe":
        text_result = whisper(audio, None, "transcribe", fn_index=0)
    else:
        text_result = whisper(audio, None, "translate", fn_index=0)
    
    return text_result

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 />If you need a custom session key, see  
        <a href="https://youtu.be/TdNSj_qgdFk" target="_blank">Bhavesh Baht video for reference</a>
        </p>-->
        <!--<p style="font-size: 94%">
            You can skip the queue by duplicating this space: 
            <span style="display: flex;align-items: center;justify-content: center;height: 30px;">
            <a href="https://huggingface.co/nightfury/whisperAI?duplicate=true"><img src="https://img.shields.io/badge/-Duplicate%20Space-blue?labelColor=white&style=flat&logo=data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAP5JREFUOE+lk7FqAkEURY+ltunEgFXS2sZGIbXfEPdLlnxJyDdYB62sbbUKpLbVNhyYFzbrrA74YJlh9r079973psed0cvUD4A+4HoCjsA85X0Dfn/RBLBgBDxnQPfAEJgBY+A9gALA4tcbamSzS4xq4FOQAJgCDwV2CPKV8tZAJcAjMMkUe1vX+U+SMhfAJEHasQIWmXNN3abzDwHUrgcRGmYcgKe0bxrblHEB4E/pndMazNpSZGcsZdBlYJcEL9Afo75molJyM2FxmPgmgPqlWNLGfwZGG6UiyEvLzHYDmoPkDDiNm9JR9uboiONcBXrpY1qmgs21x1QwyZcpvxt9NS09PlsPAAAAAElFTkSuQmCC&logoWidth=14" alt="Duplicate Space"></a>       
            </span>
        </p>-->
    </div>
"""

article = """
    <p style="font-size: 0.8em;line-height: 1.2em;">Note: this demo is not able to sustain a conversation from earlier responses. 
    For more detailed results and dialogue, you should use the official ChatGPT interface.
    <br />—
    <br/>Also, be aware that audio records from iOS devices will not be decoded as expected by Gradio. For the best experience, record your voice from a computer instead of your smartphone ;)</p>
    <div class="footer">
        <p>Whisper & 
        <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://huggingface.co/nightfury/" target="_blank">Nightfury</a>
        </p>
    </div>
"""

css = '''
    #col-container, #col-container-2 {max-width: 510px; margin-left: auto; margin-right: auto;}
    a {text-decoration-line: underline; font-weight: 600;}
    div#record_btn > .mt-6 {
        margin-top: 0!important;
    }
    div#record_btn > .mt-6 button {
        width: 100%;
        height: 40px;
    }
    .footer {
            margin-bottom: 45px;
            margin-top: 10px;
            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,elem_id="record_btn")
            task = gr.Radio(choices=["transcribe","translate"], value="transcribe", show_label=False)
            
        with gr.Row():
            #reset_conversation = gr.Checkbox(label="Reset conversation?", value=False)
            send_btn = gr.Button("Send my request !")
        #custom_token = gr.Textbox(label='If it fails, use your own session token', placeholder="your own session token", max_lines=3)   
    
    with gr.Column(elem_id="col-container-2"):
        audio_translation = gr.Textbox(type="text",label="Whisper transcript")
        gpt_response = gr.Textbox(type="text",label="chatGPT response")

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

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