Added Hindi voice assistant

Hindi Voice Assistant/app.py

@@ -0,0 +1,62 @@
+import streamlit as st
+import os
+from utils import get_answer, text_to_speech, autoplay_audio, speech_to_text
+from audio_recorder_streamlit import audio_recorder
+from streamlit_float import *
+
+float_init()
+
+def initialize_session_state():
+    if "messages" not in st.session_state:
+        st.session_state.messages = [
+            {"role": "assistant",
+             "content": "हेल्लो, मैं आपकी कैसे मदद करूं?"}]
+
+initialize_session_state()
+
+st.title("Hindi Voice Assistant")
+
+footer_container = st.container()
+with footer_container:
+    audio_bytes = audio_recorder()
+
+
+for message in st.session_state.messages:
+    with st.chat_message(message['role']):
+        st.write(message['content'])
+
+# Take input from user, convert speech to text, give that text as input to model and display the chat
+if audio_bytes:
+    # write the audio bytes to a file
+    with st.spinner("Transcribing..."):
+        webm_file_path = "temp_audio.mp3"
+        with open(webm_file_path,"wb") as f:
+            f.write(audio_bytes)
+
+        transcript = speech_to_text(webm_file_path)
+        if transcript:
+            st.session_state.messages.append({'role': 'user', 'content': transcript})
+
+        with st.chat_message('user'):
+            st.write(transcript)
+        os.remove(webm_file_path)
+
+# If the last message is not from assistant, than we have to response
+if st.session_state.messages[-1]['role'] != 'assistant':
+    with st.chat_message('assistant'):
+        with st.spinner("Thinking...."):
+            final_response = get_answer(st.session_state.messages)
+
+        with st.spinner("Generating audio response...."):
+            audio_file = text_to_speech(final_response)
+            autoplay_audio(audio_file)
+
+        st.write(final_response)
+        st.session_state.messages.append({
+            "role":"assistant",
+            "content": final_response
+        })
+
+        os.remove(audio_file)
+
+footer_container.float("bottom: 0rem;")

Hindi Voice Assistant/requirements.txt

@@ -0,0 +1,6 @@
+streamlit
+groq
+streamlit_float
+audio_recorder_streamlit
+dotenv
+gtts

Hindi Voice Assistant/utils.py

@@ -0,0 +1,57 @@
+from groq import Groq
+import streamlit as st
+import base64
+import os
+from dotenv import load_dotenv, dotenv_values
+from gtts import gTTS
+import tempfile
+
+load_dotenv()
+
+client = Groq(api_key=os.getenv("API_key"),)
+
+def speech_to_text(audio):
+  with open(audio, 'rb') as audio_file:
+    transcription = client.audio.transcriptions.create(
+        file= audio_file,
+        model = "whisper-large-v3",
+        language='hi',
+        response_format='text'
+    )
+    return transcription
+
+def get_answer(message):
+  system_message = [
+    {
+    "role" : "system",
+    "content": """You are a helpfull AI chatbot, that answers questions, in Hindi language, asked by User.
+    1. You must Avoid discussing sensitive, offensive, or harmful content. Refrain from engaging in any form of discrimination, harassment, or inappropriate behavior.
+    3. If the user expresses gratitude or indicates the end of the conversation, respond with a polite farewell.
+    """
+    }
+  ]
+  messages = system_message + message
+
+  response = client.chat.completions.create(
+           model='llama3-8b-8192',
+           messages = messages
+    )
+  return response.choices[0].message.content
+
+
+def text_to_speech(input_text):
+    tts = gTTS(input_text, lang='hi')
+    webm_file_path = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False).name
+    tts.save(webm_file_path)
+    return webm_file_path
+
+def autoplay_audio(file_path):
+    with open(file_path, 'rb') as f:
+      data = f.read()
+    b64 = base64.b64encode(data).decode('utf-8')
+    md = f"""
+    <audio autoplay>
+    <source src='data:audio/mp3;base64,{b64}' type='audio/mp3'>
+    </audio>
+    """
+    st.markdown(md, unsafe_allow_html=True)