Simple lightweight version - text responses only, no heavy TTS

app.py

@@ -1,30 +1,15 @@
 #!/usr/bin/env python3
 """
 JamboGPT - African Language AI Voice Agent
-Simplified version for Hugging Face Spaces
+Simple, Lightweight Version for Hugging Face Spaces
 """
 
-import os
 import gradio as gr
-import torch
-from transformers import pipeline
-import numpy as np
-from scipy.io import wavfile
-import tempfile
 from datetime import datetime
-import librosa
-
-# Set device
-device = "cuda" if torch.cuda.is_available() else "cpu"
-print(f"🌍 Starting JamboGPT - African Language AI Voice Agent")
-print(f"Using device: {device}")
-print("=" * 50)
 
 # Language configurations
 LANGUAGES = {
     "Swahili": {
-        "code": "swh",
-        "tts_model": "facebook/mms-tts-swh",
         "emoji": "🇰🇪",
         "speakers": "100M+",
         "region": "East Africa",
@@ -43,8 +28,6 @@ LANGUAGES = {
         }
     },
     "Kikuyu": {
-        "code": "ki",
-        "tts_model": "BrianMwangi/African-Kikuyu-TTS",
         "emoji": "🇰🇪",
         "speakers": "7M",
         "region": "Kenya",
@@ -63,8 +46,6 @@ LANGUAGES = {
         }
     },
     "Yoruba": {
-        "code": "yor",
-        "tts_model": "facebook/mms-tts-yor",
         "emoji": "🇳🇬",
         "speakers": "45M",
         "region": "West Africa",
@@ -83,8 +64,6 @@ LANGUAGES = {
         }
     },
     "Hausa": {
-        "code": "hau",
-        "tts_model": "facebook/mms-tts-hau",
         "emoji": "🇳🇬",
         "speakers": "90M",
         "region": "West Africa",
@@ -99,112 +78,100 @@ LANGUAGES = {
             "help": "Ina iya taimakawa ka. Me na gida!",
             "thanks": "Nagode! Na gida!",
             "bye": "Sai anjima! Jiya!",
-            "default": "Yoo, na ji ka. Me na gida!"
+            "default": "I na gida. Me na gida?"
         }
     },
     "Amharic": {
-        "code": "amh",
-        "tts_model": "facebook/mms-tts-amh",
         "emoji": "🇪🇹",
         "speakers": "32M",
-        "region": "Ethiopia",
+        "region": "Horn of Africa",
         "keywords": {
-            "greeting": ["ሰላም", "hello", "hi"],
+            "greeting": ["ሰላም", "hello", "hi", "ሳላም"],
             "thanks": ["አመሰግናለሁ", "thank"],
-            "help": ["ሊረዳህ", "help"],
-            "bye": ["ደህና", "goodbye"]
+            "help": ["ሚዛን", "help"],
+            "bye": ["ደህና", "goodbye", "bye"]
         },
         "responses": {
-            "greeting": "ሰላም! እንደምን ነህ?",
-            "help": "ሊረዳህ እችላለሁ። ምን ያስፈልግሃል?",
+            "greeting": "ሰላም! ደህና! ምን ያስፈልግሃል?",
+            "help": "ሚዛን! እንታይ ትደልዩ?",
             "thanks": "አመሰግናለሁ! ደህና!",
-            "bye": "ደህና ሁን! ሰላም!",
-            "default": "ሞክር። ምን ሌላ ሊረዳህ ይችላሉ?"
+            "bye": "ሰላም! ሚዛን!",
+            "default": "ሙሊ። እንታይ ተወሳኺ?"
         }
     },
     "Fon": {
-        "code": "fon",
-        "tts_model": "facebook/mms-tts-fon",
         "emoji": "🇧🇯",
         "speakers": "2M",
-        "region": "Benin, Togo",
+        "region": "West Africa",
         "keywords": {
-            "greeting": ["alekun", "hello", "hi"],
-            "thanks": ["e ku", "thank"],
-            "help": ["help"],
-            "bye": ["daabo", "goodbye"]
+            "greeting": ["bonjour", "hello", "hi"],
+            "thanks": ["merci", "thank"],
+            "help": ["aide", "help"],
+            "bye": ["au revoir", "goodbye", "bye"]
         },
         "responses": {
-            "greeting": "Alekun! Inu mẹ?",
-            "help": "Mẹ lẹ wẹ mẹ. Kini nkan?",
-            "thanks": "E ku ọpẹ! Ẹ kú!",
-            "bye": "Ó dáàbò! Ẹ kú ọjọ́!",
-            "default": "Yoo, mo gbe e. Kini nkan?"
+            "greeting": "Bonjour! Comment allez-vous?",
+            "help": "Je peux vous aider. Qu'est-ce que vous voulez?",
+            "thanks": "Merci beaucoup! De rien!",
+            "bye": "Au revoir! À bientôt!",
+            "default": "Oui, je comprends. Quoi d'autre?"
         }
     },
     "Oromo": {
-        "code": "orm",
-        "tts_model": "facebook/mms-tts-orm",
         "emoji": "🇪🇹",
         "speakers": "40M",
-        "region": "Ethiopia, Kenya",
+        "region": "East Africa",
         "keywords": {
-            "greeting": ["salaam", "hello", "hi", "akkam"],
-            "thanks": ["galatomi", "thank"],
-            "help": ["gargaaruu", "help"],
-            "bye": ["nagaa", "goodbye"]
+            "greeting": ["salaam", "hello", "hi"],
+            "thanks": ["galataa", "thank"],
+            "help": ["gargaarsa", "help"],
+            "bye": ["nagaa", "goodbye", "bye"]
         },
         "responses": {
-            "greeting": "Salaam! Akkam jirta?",
-            "help": "Gargaaruu dandeenya. Maal barbaadda?",
-            "thanks": "Galatomi! Nagaa!",
-            "bye": "Nagaa! Booda beekta!",
-            "default": "Eeyyee, hubadhe. Maal biraa?"
+            "greeting": "Salaam! Akkam jirtaa?",
+            "help": "Gargaarsa nan geedaru. Maal barbaadda?",
+            "thanks": "Galataa! Nagaa!",
+            "bye": "Nagaa! Haa jiraatin!",
+            "default": "Eeyyee, hubadha. Maal biraa?"
         }
     },
     "Somali": {
-        "code": "som",
-        "tts_model": "facebook/mms-tts-som",
         "emoji": "🇸🇴",
         "speakers": "20M",
         "region": "East Africa",
         "keywords": {
-            "greeting": ["salaam", "hello", "hi", "iska"],
+            "greeting": ["salaam", "hello", "hi"],
             "thanks": ["mahadsanid", "thank"],
-            "help": ["caawin", "help"],
-            "bye": ["nabad", "goodbye"]
+            "help": ["caawi", "help"],
+            "bye": ["nabad", "goodbye", "bye"]
         },
         "responses": {
-            "greeting": "Salaam! Iska warran?",
-            "help": "Waan kaa caawin karaa. Maxaad doonaysaa?",
+            "greeting": "Salaam! Sidee tahay?",
+            "help": "Waan kaa caawin karaa. Maxaa baahan?",
             "thanks": "Mahadsanid! Nabad!",
-            "bye": "Nabad! Marka danbe!",
-            "default": "Hah, waan ogsadaye. Wax kale?"
+            "bye": "Nabad! Halkaa ku joog!",
+            "default": "Hah, waan fahmay. Maxaa kale?"
         }
     },
     "Tigrinya": {
-        "code": "tir",
-        "tts_model": "facebook/mms-tts-tir",
         "emoji": "🇪🇷",
         "speakers": "7M",
         "region": "Horn of Africa",
         "keywords": {
             "greeting": ["ሰላም", "hello", "hi"],
-            "thanks": ["ድሊ", "thank"],
-            "help": ["ሓገዝ", "help"],
-            "bye": ["ደህና", "goodbye"]
+            "thanks": ["አመሰግናለሁ", "thank"],
+            "help": ["ሚዛን", "help"],
+            "bye": ["ደህና", "goodbye", "bye"]
         },
         "responses": {
-            "greeting": "ሰላም! ክመይ ዲካ?",
-            "help": "ክዋ ሓገዝ ይኽእል። እንታይ ትደልዩ?",
-            "thanks": "ድሊ! ናይ ሰላም!",
+            "greeting": "ሰላም! ዴሌ ኢካ?",
+            "help": "ሚዛን! እንታይ ትደልዩ?",
+            "thanks": "አመሰግናለሁ! ደህና!",
             "bye": "ሰላም! ሚዛን!",
             "default": "ሙሊ። እንታይ ተወሳኺ?"
         }
     },
     "English": {
-        "code": "eng",
-        "tts_model": "facebook/mms-tts-eng",
         "emoji": "🌍",
         "speakers": "1.5B",
         "region": "Global",
@@ -224,203 +191,8 @@ LANGUAGES = {
     }
 }
 
-# Cache for loaded models
-model_cache = {}
 conversation_history = []
 
-# CSS
-CUSTOM_CSS = """
-@import url('https://fonts.googleapis.com/css2?family=Inter:wght@300;400;500;600;700&display=swap');
-
-* {
-    font-family: 'Inter', sans-serif;
-}
-
-body {
-    background: #ffffff;
-}
-
-.gradio-container {
-    max-width: 100%;
-    margin: 0;
-    padding: 0;
-    background: #ffffff;
-}
-
-.main-container {
-    display: grid;
-    grid-template-columns: 350px 1fr;
-    min-height: 100vh;
-    gap: 0;
-}
-
-.sidebar {
-    background: #f8f9fa;
-    border-right: 1px solid #e0e0e0;
-    padding: 24px;
-    overflow-y: auto;
-    max-height: 100vh;
-}
-
-.sidebar-title {
-    font-size: 14px;
-    font-weight: 600;
-    color: #666;
-    text-transform: uppercase;
-    letter-spacing: 0.5px;
-    margin-bottom: 16px;
-}
-
-.content {
-    padding: 40px 60px;
-    display: flex;
-    flex-direction: column;
-}
-
-.header {
-    margin-bottom: 40px;
-}
-
-.logo {
-    font-size: 28px;
-    font-weight: 700;
-    color: #000;
-    margin-bottom: 20px;
-}
-
-.headline {
-    font-size: 48px;
-    font-weight: 700;
-    color: #000;
-    line-height: 1.2;
-    margin-bottom: 16px;
-}
-
-.subheadline {
-    font-size: 16px;
-    color: #666;
-    line-height: 1.5;
-}
-
-.input-section {
-    background: #f8f9fa;
-    border-radius: 8px;
-    padding: 24px;
-    margin-bottom: 24px;
-}
-
-.language-selector label {
-    display: block;
-    font-size: 13px;
-    font-weight: 600;
-    color: #333;
-    margin-bottom: 8px;
-    text-transform: uppercase;
-    letter-spacing: 0.5px;
-}
-
-.language-selector select {
-    width: 100%;
-    padding: 12px;
-    border: 1px solid #d0d0d0;
-    border-radius: 6px;
-    font-size: 14px;
-    background: white;
-    cursor: pointer;
-}
-
-.generate-btn {
-    background: #000 !important;
-    color: white !important;
-    border: none !important;
-    border-radius: 6px !important;
-    padding: 12px 24px !important;
-    font-weight: 600 !important;
-    font-size: 14px !important;
-    cursor: pointer !important;
-}
-
-.generate-btn:hover {
-    background: #333 !important;
-}
-
-.output-section {
-    background: #f8f9fa;
-    border-radius: 8px;
-    padding: 24px;
-}
-
-.output-label {
-    font-size: 13px;
-    font-weight: 600;
-    color: #333;
-    text-transform: uppercase;
-    letter-spacing: 0.5px;
-    margin-bottom: 16px;
-}
-
-.status-message {
-    padding: 12px 16px;
-    border-radius: 6px;
-    font-size: 14px;
-    margin-bottom: 16px;
-}
-
-.status-success {
-    background: #d4edda;
-    color: #155724;
-    border: 1px solid #c3e6cb;
-}
-
-.status-error {
-    background: #f8d7da;
-    color: #721c24;
-    border: 1px solid #f5c6cb;
-}
-
-@media (max-width: 768px) {
-    .main-container {
-        grid-template-columns: 1fr;
-    }
-    
-    .sidebar {
-        display: none;
-    }
-    
-    .content {
-        padding: 20px;
-    }
-    
-    .headline {
-        font-size: 32px;
-    }
-}
-"""
-
-def load_tts_model(language_name):
-    """Load TTS model for the specified language."""
-    if language_name not in LANGUAGES:
-        return None
-    
-    lang_config = LANGUAGES[language_name]
-    model_id = lang_config["tts_model"]
-    
-    if model_id in model_cache:
-        return model_cache[model_id]
-    
-    try:
-        print(f"Loading TTS model for {language_name}: {model_id}")
-        synthesizer = pipeline(
-            "text-to-speech",
-            model=model_id,
-            device=device if device == "cuda" else -1
-        )
-        model_cache[model_id] = synthesizer
-        return synthesizer
-    except Exception as e:
-        print(f"Error loading model {model_id}: {e}")
-        return None
-
 def detect_intent(text, language):
     """Detect user intent from text."""
     text_lower = text.lower()
@@ -444,188 +216,139 @@ def generate_response(text, language):
         intent = detect_intent(text, language)
         response = responses.get(intent, responses.get("default", "I understand."))
         
-        return response, f"✅ Response generated!"
-    except Exception as e:
-        print(f"Error generating response: {e}")
-        return None, f"❌ Error: {str(e)}"
-
-def synthesize_speech(text, language):
-    """Convert text to speech."""
-    if not text or not text.strip():
-        return None, "❌ No text to synthesize."
-    
-    try:
-        synthesizer = load_tts_model(language)
-        if synthesizer is None:
-            return None, f"❌ Failed to load TTS model for {language}."
-        
-        print(f"Generating speech for: {text[:50]}...")
-        speech = synthesizer(text)
-        
-        audio_array = np.array(speech["audio"]).flatten()
-        sample_rate = speech["sampling_rate"]
-        
-        with tempfile.NamedTemporaryFile(suffix=".wav", delete=False) as f:
-            wavfile.write(f.name, sample_rate, (audio_array * 32767).astype(np.int16))
-            temp_path = f.name
-        
-        return temp_path, "✅ Speech generated!"
-    except Exception as e:
-        print(f"Error synthesizing: {e}")
-        return None, f"❌ Error: {str(e)}"
-
-def process_text_input(text, language):
-    """Process text input: generate response -> synthesize."""
-    try:
-        # Generate response
-        response_text, response_status = generate_response(text, language)
-        if response_text is None:
-            return None, response_text, response_status
-        
-        # Synthesize response
-        audio_output, synth_status = synthesize_speech(response_text, language)
-        
         # Add to history
         conversation_history.append({
             "user": text,
-            "agent": response_text,
-            "timestamp": datetime.now().isoformat()
+            "agent": response,
+            "language": language,
+            "timestamp": datetime.now().strftime("%H:%M:%S")
         })
         
-        return audio_output, response_text, synth_status
+        return response
     except Exception as e:
-        print(f"Error processing: {e}")
-        return None, None, f"❌ Error: {str(e)}"
+        print(f"Error generating response: {e}")
+        return "I understand. Can you say more?"
 
 def create_interface():
     """Create the voice agent interface."""
     
     with gr.Blocks(
-        title="JamboGPT - African Language AI Voice Agent",
-        css=CUSTOM_CSS
+        title="JamboGPT - African Language AI",
+        theme=gr.themes.Soft(primary_hue="purple")
     ) as demo:
         
-        # Main container
-        with gr.Row(equal_height=True):
-            # Sidebar
-            with gr.Column(scale=0, min_width=350):
-                gr.Markdown(
-                    """
-                    <div class="sidebar">
-                        <div class="sidebar-title">🗣️ Conversation History</div>
-                        <div style="text-align: center; color: #999; padding: 20px; font-size: 13px;">
-                        No conversations yet
-                        </div>
-                    </div>
-                    """
-                )
+        gr.Markdown("""
+        # 🌍 JamboGPT - African Language AI
+        
+        **Chat with AI in 10 African languages**
+        
+        Swahili • Kikuyu • Yoruba • Hausa • Amharic • Fon • Oromo • Somali • Tigrinya • English
+        """)
+        
+        with gr.Group():
+            # Language selector
+            language_choice = gr.Dropdown(
+                choices=list(LANGUAGES.keys()),
+                value="Swahili",
+                label="Select Language",
+                interactive=True
+            )
+            
+            # Language info
+            language_info = gr.Markdown(
+                f"🇰🇪 **Swahili** • 100M+ speakers • East Africa"
+            )
+            
+            def update_language_info(language):
+                if language in LANGUAGES:
+                    lang_data = LANGUAGES[language]
+                    return f"{lang_data['emoji']} **{language}** • {lang_data['speakers']} speakers • {lang_data['region']}"
+                return ""
+            
+            language_choice.change(update_language_info, inputs=language_choice, outputs=language_info)
             
-            # Main content
-            with gr.Column(scale=1):
-                # Header
-                gr.Markdown(
-                    """
-                    <div class="header">
-                        <div class="logo">🌍 JamboGPT</div>
-                        <div class="headline">African Language AI Voice Agent</div>
-                        <div class="subheadline">Chat with AI in 9 African languages. Type or speak, and get intelligent responses in your language.</div>
-                    </div>
-                    """
-                )
-                
-                # Input section
-                with gr.Group():
-                    gr.Markdown('<div class="input-section">')
-                    
-                    # Language selector
-                    language_choice = gr.Dropdown(
-                        choices=list(LANGUAGES.keys()),
-                        value="Swahili",
-                        label="Select Language",
-                        interactive=True
-                    )
-                    
-                    # Language info
-                    language_info = gr.Markdown(
-                        f"🇰🇪 **Swahili** • 100M+ speakers • East Africa"
-                    )
-                    
-                    def update_language_info(language):
-                        if language in LANGUAGES:
-                            lang_data = LANGUAGES[language]
-                            return f"{lang_data['emoji']} **{language}** • {lang_data['speakers']} speakers • {lang_data['region']}"
-                        return ""
-                    
-                    language_choice.change(update_language_info, inputs=language_choice, outputs=language_info)
-                    
-                    # Text input
-                    text_input = gr.Textbox(
-                        label="Type your message in the selected language",
-                        placeholder="Type your message here...",
-                        lines=3,
-                        interactive=True
-                    )
-                    
-                    # Process button
-                    process_btn = gr.Button(
-                        "🎤 Generate Response",
-                        variant="primary",
-                        size="lg"
-                    )
-                    
-                    gr.Markdown('</div>')
-                
-                # Output section
-                with gr.Group():
-                    gr.Markdown('<div class="output-section">')
-                    
-                    gr.Markdown('<div class="output-label">🤖 Agent Response</div>')
-                    agent_response = gr.Textbox(
-                        label="",
-                        interactive=False,
-                        placeholder="The agent's response will appear here"
-                    )
-                    
-                    gr.Markdown('<div class="output-label">🔊 Agent Voice</div>')
-                    audio_output = gr.Audio(
-                        label="",
-                        type="filepath",
-                        interactive=False
-                    )
-                    
-                    status_message = gr.Textbox(
-                        label="Status",
-                        interactive=False,
-                        value="Ready to chat!"
-                    )
-                    
-                    gr.Markdown('</div>')
-                
-                # Footer
-                gr.Markdown(
-                    """
-                    <div style="margin-top: 40px; padding-top: 20px; border-top: 1px solid #e0e0e0; font-size: 13px; color: #999;">
-                    <p>🌍 <strong>JamboGPT</strong> - African Language AI Voice Agent</p>
-                    <p>Chat in 9 African languages • Powered by Hugging Face • <a href="https://huggingface.co/spaces/stano03/jambogpt" style="color: #666;">View on Hugging Face</a></p>
-                    </div>
-                    """
-                )
+            # Text input
+            text_input = gr.Textbox(
+                label="Type your message",
+                placeholder="Type in your selected language...",
+                lines=3,
+                interactive=True
+            )
+            
+            # Process button
+            process_btn = gr.Button(
+                "🎤 Generate Response",
+                variant="primary",
+                size="lg"
+            )
+        
+        # Output section
+        with gr.Group():
+            agent_response = gr.Textbox(
+                label="🤖 Agent Response",
+                interactive=False,
+                placeholder="The agent's response will appear here"
+            )
+            
+            # Conversation history
+            history_display = gr.Textbox(
+                label="📝 Conversation History",
+                interactive=False,
+                lines=4,
+                placeholder="Your conversation history will appear here"
+            )
         
         # Connect process button
+        def process_input(text, language):
+            if not text:
+                return "Please enter some text!", ""
+            
+            response = generate_response(text, language)
+            
+            # Format history
+            history_text = ""
+            for msg in conversation_history[-5:]:  # Show last 5 messages
+                history_text += f"[{msg['timestamp']}] {msg['language']}\n"
+                history_text += f"You: {msg['user']}\n"
+                history_text += f"Agent: {msg['agent']}\n\n"
+            
+            return response, history_text
+        
         process_btn.click(
-            fn=process_text_input,
+            fn=process_input,
+            inputs=[text_input, language_choice],
+            outputs=[agent_response, history_display]
+        )
+        
+        # Examples
+        gr.Examples(
+            examples=[
+                ["Habari, karibu sana!", "Swahili"],
+                ["Wĩ mwega, karibu!", "Kikuyu"],
+                ["Pẹlẹ o, bawo ni o se?", "Yoruba"],
+                ["Hello, how are you?", "English"],
+            ],
             inputs=[text_input, language_choice],
-            outputs=[audio_output, agent_response, status_message]
+            outputs=[agent_response],
+            fn=process_input,
+            cache_examples=False,
         )
+        
+        gr.Markdown("""
+        ---
+        **JamboGPT** - Making AI Accessible to African Languages
+        
+        🔗 [GitHub](https://github.com/stano03/jambogpt) | 📊 [Dataset](https://huggingface.co/datasets/stano03/jambogpt-real-dataset)
+        """)
     
     return demo
 
 if __name__ == "__main__":
-    print("🚀 Creating JamboGPT Voice Agent Interface...")
+    print("🚀 Creating JamboGPT Interface...")
     demo = create_interface()
     
     print("=" * 50)
-    print("✅ JamboGPT Voice Agent is ready!")
+    print("✅ JamboGPT is ready!")
     print("=" * 50)
     
     demo.launch(