Update app.py

app.py

@@ -1,64 +1,95 @@
+import os
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
-
-
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+import requests
+from sentence_transformers import SentenceTransformer, util
+import torch
+import json
+import urllib.parse
+
+# Fetch Hugging Face API Token securely from environment variables
+HF_API_TOKEN = os.getenv("HF_API_TOKEN")  # This fetches the token securely
+
+WHISPER_API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large"
+LLAMA_API_URL = "https://api-inference.huggingface.co/models/meta-llama/Llama-2-7b-chat-hf"
+
+# Load SentenceTransformer model for retrieval
+retriever_model = SentenceTransformer("distiluse-base-multilingual-cased-v2")
+
+# Load dataset
+with open("qa_dataset.json", "r", encoding="utf-8") as f:
+    qa_data = json.load(f)
+
+# Function to transcribe audio using Whisper
+def transcribe_audio(audio_file):
+    headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
+    with open(audio_file, "rb") as f:
+        response = requests.post(WHISPER_API_URL, headers=headers, data=f)
+    return response.json()["text"]
+
+# Function to generate TTS audio URL (Google Translate API for Tamil Voice)
+def get_tts_audio_url(text, lang="ta"):
+    safe_text = text.replace(" ", "+")
+    return f"https://translate.google.com/translate_tts?ie=UTF-8&q={safe_text}&tl={lang}&client=tw-ob"
+
+# Function to retrieve a relevant response from the Q&A dataset using SentenceTransformer
+def get_bot_response(query):
+    query_embedding = retriever_model.encode(query, convert_to_tensor=True)
+    qa_embeddings = retriever_model.encode([qa["question"] for qa in qa_data], convert_to_tensor=True)
+
+    scores = util.pytorch_cos_sim(query_embedding, qa_embeddings)
+    best_idx = torch.argmax(scores)
+
+    top_qa = qa_data[best_idx]
+    prompt = f"User asked: {query}\nRelevant FAQ: {top_qa['question']}\nAnswer: {top_qa['answer']}\nNow generate a helpful and fluent Tamil response to the user query."
+
+    # Use LLaMA for generating the refined response
+    headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
+    payload = {
+        "inputs": prompt,
+        "parameters": {"temperature": 0.7, "max_new_tokens": 150, "return_full_text": False},
+    }
+    response = requests.post(LLAMA_API_URL, headers=headers, json=payload)
+    result = response.json()
+
+    if isinstance(result, list) and "generated_text" in result[0]:
+        return result[0]["generated_text"]
+    else:
+        return "மன்னிக்கவும், நான் இந்த கேள்விக்கு பதில் தர முடியவில்லை."
+
+# Gradio interface function
+def chatbot(audio, message, history, system_message, max_tokens, temperature, top_p):
+    if audio is not None:
+        # Save the audio file temporarily
+        with open("temp.wav", "wb") as f:
+            f.write(audio.read())
+
+        # Transcribe the audio using Whisper
+        transcript = transcribe_audio("temp.wav")
+        message = transcript  # Use the transcript as the input message
+
+    # Get the bot response using the text input or transcribed audio
+    response = get_bot_response(message)
+
+    # Generate the TTS audio URL
+    audio_url = get_tts_audio_url(response)
+
+    return response, audio_url
+
+# Define Gradio interface
+demo = gr.Interface(
+    fn=chatbot,
+    inputs=[
+        gr.Audio(source="microphone", type="file", label="Speak to the Bot"),
+        gr.Textbox(value="How can I help you?", label="Text Input (optional)"),
+        gr.State(),
+        gr.Textbox(value="You are a friendly chatbot.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
     ],
+    outputs=[gr.Textbox(label="Response"), gr.Audio(label="Bot's Voice Response (Tamil)")],
+    live=True,
 )
 
-
 if __name__ == "__main__":
     demo.launch()