Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from pathlib import Path
|
| 2 |
+
|
| 3 |
+
import streamlit as st
|
| 4 |
+
from audiorecorder import audiorecorder
|
| 5 |
+
from openai import OpenAI
|
| 6 |
+
|
| 7 |
+
open_ai = OpenAI(
|
| 8 |
+
api_key="sk-dW5stjNmgnwJb00p-Hk-oQCwGhHPNMoT9Nrx4yQvH4T3BlbkFJgoCDc7RdDufCuXRQNCpcq_m4oUswu217zkojwaxw4A")
|
| 9 |
+
|
| 10 |
+
st.set_page_config(
|
| 11 |
+
layout="centered",
|
| 12 |
+
page_icon="🤖",
|
| 13 |
+
page_title="Voice Chat"
|
| 14 |
+
)
|
| 15 |
+
|
| 16 |
+
st.title("Voice Chat")
|
| 17 |
+
|
| 18 |
+
if 'chat' not in st.session_state:
|
| 19 |
+
st.session_state.chat = []
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
def recordAudio():
|
| 23 |
+
audio = audiorecorder("Click to record", "Click to stop recording")
|
| 24 |
+
if len(audio) > 0:
|
| 25 |
+
audio.export("audio.wav", format="wav")
|
| 26 |
+
userPrompt = speechToText()
|
| 27 |
+
st.text(userPrompt)
|
| 28 |
+
st.session_state.chat.append({
|
| 29 |
+
'role': 'user',
|
| 30 |
+
'content': userPrompt
|
| 31 |
+
})
|
| 32 |
+
|
| 33 |
+
answer = getAnswer(userPrompt)
|
| 34 |
+
st.session_state.chat.append({
|
| 35 |
+
'role': 'assistant',
|
| 36 |
+
'content': answer
|
| 37 |
+
})
|
| 38 |
+
textToSpeech(answer)
|
| 39 |
+
st.audio("reply.wav", autoplay=True)
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
def speechToText():
|
| 43 |
+
audio_file = open("audio.wav", "rb")
|
| 44 |
+
transcription = open_ai.audio.transcriptions.create(
|
| 45 |
+
model="whisper-1",
|
| 46 |
+
file=audio_file
|
| 47 |
+
)
|
| 48 |
+
return transcription.text
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
def textToSpeech(text):
|
| 52 |
+
speech_file_path = Path(__file__).parent / "reply.wav"
|
| 53 |
+
response = open_ai.audio.speech.create(
|
| 54 |
+
model="tts-1",
|
| 55 |
+
voice="alloy",
|
| 56 |
+
input=text
|
| 57 |
+
)
|
| 58 |
+
|
| 59 |
+
response.stream_to_file(speech_file_path)
|
| 60 |
+
|
| 61 |
+
|
| 62 |
+
def getAnswer(question):
|
| 63 |
+
response = open_ai.chat.completions.create(
|
| 64 |
+
model="gpt-4o-mini",
|
| 65 |
+
messages=[
|
| 66 |
+
{"role": "system", "content": "You are a helpful assistant who can uderstand native urdu andwho provide very very short response in urdu"},
|
| 67 |
+
*st.session_state.chat,
|
| 68 |
+
{"role": "user", "content": question}
|
| 69 |
+
]
|
| 70 |
+
)
|
| 71 |
+
|
| 72 |
+
return response.choices[0].message.content
|
| 73 |
+
|
| 74 |
+
|
| 75 |
+
recordAudio()
|