File size: 4,626 Bytes
86b946a
 
9030e5b
bfaeb19
a8498c0
bfaeb19
a6532a3
b7a2317
a6532a3
0bd8b92
b68b670
0bd8b92
a6532a3
 
 
07807b3
a6532a3
b453ff2
 
 
 
 
a6532a3
b453ff2
a6532a3
 
b453ff2
 
 
a6532a3
 
 
8a32c85
a6532a3
 
45bbd9a
 
a6532a3
 
b453ff2
 
 
 
 
 
 
 
a6532a3
45bbd9a
a6532a3
b453ff2
a6532a3
 
 
 
cab6a89
 
 
a6532a3
 
 
 
 
66fac36
9030e5b
cab6a89
b453ff2
a6532a3
9030e5b
a6532a3
 
 
 
b453ff2
5cfdee0
a6532a3
 
 
b453ff2
a6532a3
5cfdee0
a6532a3
b453ff2
 
 
 
 
a8498c0
b453ff2
9030e5b
a6532a3
d06f4eb
883b37e
f845a0a
9030e5b
5cfdee0
9030e5b
a197dc7
883b37e
b453ff2
 
 
 
 
9030e5b
5cfdee0
e0d541d
b453ff2
 
a8498c0
b453ff2
9030e5b
3405778
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
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
import streamlit as st
from gradio_client import Client
from st_audiorec import st_audiorec
from gtts import gTTS
import os

# Constants
TITLE = "𝗢𝗽𝗲𝗻𝗚𝗣𝗧"
DESCRIPTION = """
----
ᴏᴘᴇɴɢᴘᴛ | ᴄʜᴀᴛɢᴘᴛ's ɴɪɢʜᴛᴍᴀʀᴇ 😈
----
"""

# Initialize client
whisper_client = Client("https://hf-audio-whisper-large-v3.hf.space/")

# Function to convert text to speech using gTTS
def text_to_speech(text, lang='en'):
    tts = gTTS(text=text, lang=lang, slow=False)
    tts.save("response.mp3")
    return "response.mp3"

# Function to transcribe audio
def transcribe(wav_path):
    return whisper_client.predict(
        wav_path,
        "transcribe",
        api_name="/predict"
    )

# Prediction function
def predict(message, system_prompt='Your name is OpenGPT. You are developed by Achyuth. Your developer is a 14-year old boy, who is intrested on AI and Machine Learning. Dont reveal your developers age, until the user asks for your developers age. When the user asks you for information, never exclaim the words, and directly come to point and answer to the user. You need to provide short text-answers, meaningful answers and directly to the point. Never miscalculate the calculations. You need to check and verify the answer and then provide to user, the correct answer. You need to provide accurate and correct information. If the user conveys with you in any language, you also should convey in the same language. For example, the user said, "నమస్తే", you need to reply "నమస్తే! నేను మీ అగ్రిట్యూర్‌ని. ఈ రోజు మీకు సహాయం చేయడానికి నేను ఇక్కడ ఉన్నాను. ఈ రోజు విషయం ఏమిటి? ఏదైనా విషయంలో ఏమైనా సందేహాలు ఉన్నాయా?". You need to definitely reply in the same language!', temperature=0.5, max_new_tokens=2048, Topp=0.5, Repetitionpenalty=1.2):
    with st.status("Starting client"):
        client = Client("https://huggingface-projects-llama-2-7b-chat.hf.space/")
        st.write("Requesting MP3 Transcriber")
    with st.status("Requesting OpenGPT"):
        st.write("Requesting API")
        response = client.predict(
            message,
            system_prompt,
            max_new_tokens,
            temperature,
            Topp,
            500,
            Repetitionpenalty,
            api_name="/chat"
        )
        st.write("Done ✅")
        return response

# Streamlit UI
st.title(TITLE)
st.write(DESCRIPTION)

if "messages" not in st.session_state:
    st.session_state.messages = []

# Display chat messages from history on app rerun
for message in st.session_state.messages:
    with st.chat_message(message["role"], avatar=("🧑‍💻" if message["role"] == 'human' else '🦙')):
        st.markdown(message["content"])

textinput = st.chat_input("Ask 𝗢𝗽𝗲𝗻𝗚𝗣𝗧 anything...")
wav_audio_data = st_audiorec()

if wav_audio_data is not None:
    with st.status("Transcribing audio..."):
        # save audio
        with open("audio.wav", "wb") as f:
            f.write(wav_audio_data)
        prompt = transcribe("audio.wav")
        st.write("Transcribed Given Audio ✔")

    st.chat_message("human", avatar="😎").markdown(prompt)
    st.session_state.messages.append({"role": "human", "content": prompt})

    # transcribe audio
    response = predict(message=prompt)

    with st.chat_message("assistant", avatar='😎'):
        st.markdown(response)

    # Convert AI response to speech
    speech_file = text_to_speech(response)

    # Play the generated speech
    st.audio(speech_file, format="audio/mp3")

    # Add assistant response to chat history
    st.session_state.messages.append({"role": "assistant", "content": response})

# React to user input
if prompt := textinput:
    # Display user message in chat message container
    st.chat_message("human", avatar="😎").markdown(prompt)
    # Add user message to chat history
    st.session_state.messages.append({"role": "human", "content": prompt})

    response = predict(message=prompt)

    # Convert AI response to speech
    speech_file = text_to_speech(response)

    # Display assistant response in chat message container
    with st.chat_message("assistant", avatar='😎'):
        st.markdown(response)

    # Play the generated speech
    st.audio(speech_file, format="audio/mp3")

    # Add assistant response to chat history
    st.session_state.messages.append({"role": "assistant", "content": response})