File size: 2,775 Bytes
c688a3b
 
6d7dbcd
 
 
180d8aa
 
6d7dbcd
c688a3b
 
 
 
 
 
 
 
6d7dbcd
 
 
 
 
 
 
5f3392e
e7b338c
180d8aa
 
 
 
 
 
f7709d8
 
 
 
180d8aa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9fdfed4
c52d065
9fdfed4
c52d065
9fdfed4
c688a3b
 
5f3392e
e7b338c
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
import openai
import os
import random
import streamlit as st

from langchain.prompts import PromptTemplate
from modules.gpt_modules import gpt_call
from dotenv import dotenv_values

"""
apt-get update
apt-get install ffmpeg
"""

config = dotenv_values(".env")

if config:
    openai.organization = config.get('OPENAI_ORGANIZATION')
    openai.api_key = config.get('OPENAI_API_KEY')
else:
    openai.organization = st.secrets['OPENAI_ORGANIZATION'] #config.get('OPENAI_ORGANIZATION')
    openai.api_key = st.secrets['OPENAI_API_KEY'] #config.get('OPENAI_API_KEY')


def debate_in_sound(audio):
    os.rename(audio, audio + '.wav')
    file = open(audio + '.wav', "rb")

    # user_words
    user_prompt = openai.Audio.transcribe("whisper-1", file).text

    print("**************************************")
    print("user_audio transcription", user_prompt)
    print("**************************************")

    # 일단 테스트를 위해 고정함
    debate_subject = "In 2050, AI robots are able to replicate the appearance, conversation, and reaction to emotions of human beings. However, their intelligence still does not allow them to sense emotions and feelings such as pain, happiness, joy, and etc."

    debate_role = [
            "pro side", 
            "con side",
        ]
    user_debate_role = random.choice(debate_role)
    bot_debate_role = "".join([role for role in debate_role if role != user_debate_role])

    debate_preset = "\n".join([
            "Debate Rules: ",
            "1) This debate will be divided into pro and con",
            "2) You must counter user's arguments",
            "3) Answer logically with an introduction, body, and conclusion.\n", #add this one.
            "User debate role: " + user_debate_role,
            "Bot debate roles: " + bot_debate_role + "\n",
            "Debate subject: " + debate_subject
        ])
    
    prompt_template = PromptTemplate(
                input_variables=["prompt"],
                template="\n".join([
                    debate_preset, #persona
                    "User: {prompt}",
                    "Bot: "
                    ])
            )
    bot_prompt = prompt_template.format(
                prompt=user_prompt
            )
    response = gpt_call(bot_prompt)

    return response


def whisper_transcribe(audio):

    audio_file= open("audio/audio.wav", "rb")
    result = openai.Audio.transcribe("whisper-1", audio_file).text
    audio_file.close()

    return result


# gr.Interface(
#     title = 'Whisper Audio to Text with Speaker Recognition', 
#     fn=debate,
#     inputs=[
#         gr.inputs.Audio(source="microphone", type="filepath"),
#         #gr.inputs.Number(default=2, label="Number of Speakers")
#     ],
#     outputs="text"
#   ).launch()