Update app.py
Browse files
app.py
CHANGED
@@ -3,14 +3,12 @@
|
|
3 |
import os
|
4 |
import chromadb
|
5 |
import streamlit as st
|
6 |
-
from base64 import b64decode
|
7 |
from langchain_huggingface import HuggingFaceEmbeddings
|
8 |
from langchain_chroma import Chroma
|
9 |
from langchain_groq import ChatGroq
|
10 |
from langchain.memory import ConversationBufferMemory
|
11 |
from langchain.chains import ConversationalRetrievalChain
|
12 |
from PyPDF2 import PdfReader
|
13 |
-
from streamlit_audio_recorder import st_audio_recorder
|
14 |
from groq import Groq
|
15 |
|
16 |
# Clear ChromaDB cache to fix tenant issue
|
@@ -19,7 +17,7 @@ chromadb.api.client.SharedSystemClient.clear_system_cache()
|
|
19 |
# Ensure required environment variables are set
|
20 |
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
|
21 |
if not GROQ_API_KEY:
|
22 |
-
st.error("GROQ_API_KEY is not set. Please configure it in
|
23 |
st.stop()
|
24 |
|
25 |
# Initialize Groq Client for transcription and LLM
|
@@ -53,18 +51,6 @@ def chat_chain(vectorstore):
|
|
53 |
)
|
54 |
return chain
|
55 |
|
56 |
-
# Function to record audio using streamlit_audio_recorder
|
57 |
-
def record_audio():
|
58 |
-
st.write("Click below to record your audio:")
|
59 |
-
audio_bytes = st_audio_recorder()
|
60 |
-
if audio_bytes:
|
61 |
-
audio_file_path = "recorded_audio.wav"
|
62 |
-
with open(audio_file_path, "wb") as f:
|
63 |
-
f.write(audio_bytes)
|
64 |
-
st.success("Audio recorded successfully!")
|
65 |
-
return audio_file_path
|
66 |
-
return None
|
67 |
-
|
68 |
# Transcribe audio using Groq Whisper
|
69 |
def transcribe_audio(file_path):
|
70 |
"""Transcribe audio using Groq's Whisper model."""
|
@@ -87,7 +73,7 @@ if uploaded_files:
|
|
87 |
chain = chat_chain(vectorstore)
|
88 |
st.success("PDFs processed! Ready to chat.")
|
89 |
|
90 |
-
input_method = st.radio("Choose Input Method", ["Text Input", "Audio
|
91 |
|
92 |
# Text Input Mode
|
93 |
if input_method == "Text Input":
|
@@ -97,13 +83,17 @@ if uploaded_files:
|
|
97 |
response = chain({"question": query})["answer"]
|
98 |
st.write(f"**Response:** {response}")
|
99 |
|
100 |
-
# Audio Input Mode
|
101 |
-
elif input_method == "Audio
|
102 |
-
|
103 |
-
if
|
104 |
-
|
|
|
|
|
|
|
|
|
105 |
st.write("Transcribing audio...")
|
106 |
-
transcription = transcribe_audio(
|
107 |
st.write(f"**You said:** {transcription}")
|
108 |
|
109 |
with st.spinner("Generating response..."):
|
|
|
3 |
import os
|
4 |
import chromadb
|
5 |
import streamlit as st
|
|
|
6 |
from langchain_huggingface import HuggingFaceEmbeddings
|
7 |
from langchain_chroma import Chroma
|
8 |
from langchain_groq import ChatGroq
|
9 |
from langchain.memory import ConversationBufferMemory
|
10 |
from langchain.chains import ConversationalRetrievalChain
|
11 |
from PyPDF2 import PdfReader
|
|
|
12 |
from groq import Groq
|
13 |
|
14 |
# Clear ChromaDB cache to fix tenant issue
|
|
|
17 |
# Ensure required environment variables are set
|
18 |
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
|
19 |
if not GROQ_API_KEY:
|
20 |
+
st.error("GROQ_API_KEY is not set. Please configure it in environment variables.")
|
21 |
st.stop()
|
22 |
|
23 |
# Initialize Groq Client for transcription and LLM
|
|
|
51 |
)
|
52 |
return chain
|
53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
54 |
# Transcribe audio using Groq Whisper
|
55 |
def transcribe_audio(file_path):
|
56 |
"""Transcribe audio using Groq's Whisper model."""
|
|
|
73 |
chain = chat_chain(vectorstore)
|
74 |
st.success("PDFs processed! Ready to chat.")
|
75 |
|
76 |
+
input_method = st.radio("Choose Input Method", ["Text Input", "Audio File Upload"])
|
77 |
|
78 |
# Text Input Mode
|
79 |
if input_method == "Text Input":
|
|
|
83 |
response = chain({"question": query})["answer"]
|
84 |
st.write(f"**Response:** {response}")
|
85 |
|
86 |
+
# Audio Input Mode (File Upload)
|
87 |
+
elif input_method == "Audio File Upload":
|
88 |
+
uploaded_audio = st.file_uploader("Upload an audio file (.wav, .mp3)", type=["wav", "mp3"])
|
89 |
+
if uploaded_audio:
|
90 |
+
audio_file_path = "uploaded_audio.wav"
|
91 |
+
with open(audio_file_path, "wb") as f:
|
92 |
+
f.write(uploaded_audio.read())
|
93 |
+
|
94 |
+
st.audio(audio_file_path, format="audio/wav")
|
95 |
st.write("Transcribing audio...")
|
96 |
+
transcription = transcribe_audio(audio_file_path)
|
97 |
st.write(f"**You said:** {transcription}")
|
98 |
|
99 |
with st.spinner("Generating response..."):
|