Upload 2 files

app.py

@@ -0,0 +1,95 @@
+import os, sys, time, json, asyncio, tempfile
+import gradio as gr
+import soundfile as sf, edge_tts, requests
+from groq import Groq
+from dotenv import load_dotenv
+
+# ───── Load .env and keys ───────────────────────────────────
+load_dotenv()
+HF_TOKEN     = os.getenv("Hugging_Face_API")
+WHISPER_URL  = "https://api-inference.huggingface.co/models/openai/whisper-large-v3"
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
+VOICE_ID     = "ur-PK-UzmaNeural"
+MAX_TOKENS   = 256
+
+client = Groq(api_key=GROQ_API_KEY)
+
+# ───── Transcribe audio ─────────────────────────────────────
+def whisper_transcribe(audio_path: str) -> str:
+    with open(audio_path, "rb") as f:
+        audio = f.read()
+    r = requests.post(
+        WHISPER_URL,
+        headers={
+            "Authorization": f"Bearer {HF_TOKEN}",
+            "Content-Type": "audio/wav"
+        },
+        data=audio, timeout=120)
+    try:
+        txt = r.json().get("text", "")
+        print("📝 Transcription:", txt)
+        return txt
+    except json.JSONDecodeError:
+        print("❌ Whisper JSON error:", r.text)
+        return ""
+
+# ───── Generate Urdu response ───────────────────────────────
+def generate_urdu(prompt: str) -> str:
+    if not prompt.strip():
+        return ""
+    messages = [
+        {"role": "system", "content": "براہ کرم ہمیشہ سادہ اور فصیح اردو میں جواب دیں۔"},
+        {"role": "user", "content": prompt.strip()}
+    ]
+    try:
+        res = client.chat.completions.create(
+            model="llama-3.1-8b-instant",
+            messages=messages,
+            temperature=0.7,
+            max_tokens=MAX_TOKENS
+        )
+        reply = res.choices[0].message.content.strip()
+        if "۔" in reply:
+            reply = reply.rsplit("۔", 1)[0] + "۔"
+        print("🤖 Reply:", reply)
+        return reply
+    except Exception as e:
+        print("❌ Groq error:", e, file=sys.stderr)
+        return ""
+
+# ───── Generate TTS audio ───────────────────────────────────
+async def speak(text: str) -> str:
+    if not text:
+        return ""
+    tmp = tempfile.NamedTemporaryFile(suffix=".mp3", delete=False)
+    tmp_path = tmp.name
+    tmp.close()
+    await edge_tts.Communicate(text=text, voice=VOICE_ID).save(tmp_path)
+    return tmp_path
+
+# ───── Complete pipeline ────────────────────────────────────
+async def process_pipeline(audio_file):
+    if not audio_file:
+        return "📭 Empty input", None
+
+    transcription = whisper_transcribe(audio_file)
+    reply = generate_urdu(transcription)
+    audio_output_path = await speak(reply)
+    return reply, audio_output_path
+
+# ───── Gradio Interface ─────────────────────────────────────
+with gr.Blocks() as demo:
+    gr.Markdown("## 🗣️ Urdu Voice Chatbot with LLaMA-3.1 + Whisper + Edge TTS")
+
+    with gr.Row():
+        with gr.Column():
+            audio_input = gr.Audio(type="filepath", label="🎙️ Record your question")
+            submit_btn = gr.Button("▶️ Process")
+        with gr.Column():
+            output_text = gr.Textbox(label="🤖 Urdu Response")
+            output_audio = gr.Audio(label="🔊 Spoken Response", autoplay=True)
+
+    submit_btn.click(fn=process_pipeline, inputs=audio_input, outputs=[output_text, output_audio])
+
+# Run
+demo.launch()

requirements.txt

@@ -0,0 +1,6 @@
+gradio
+groq
+python-dotenv
+soundfile
+requests
+edge-tts