Spaces:

HeWhoComes
/

zw-kitten-tts-working

Runtime error

App Files Files Community

HeWhoComes commited on Aug 20, 2025

Commit

c8d46a9

verified ·

1 Parent(s): b864501

Create app.py

Browse files

Files changed (1) hide show

app.py +495 -0

app.py ADDED Viewed

	@@ -0,0 +1,495 @@

+#!/usr/bin/env python3
+"""
+ZW Kitten - Working FastAPI version with real KittenTTS
+"""
+import os
+import tempfile
+from fastapi import FastAPI, Request
+from fastapi.responses import HTMLResponse, FileResponse, JSONResponse
+from fastapi.staticfiles import StaticFiles
+import uvicorn
+# Try to import KittenTTS, fall back to demo if not available
+try:
+    from kittentts import KittenTTS
+    KITTEN_AVAILABLE = True
+    print("✅ KittenTTS loaded successfully")
+except ImportError:
+    KITTEN_AVAILABLE = False
+    print("⚠️ KittenTTS not available, using demo mode")
+app = FastAPI(title="ZW Kitten TTS")
+# Initialize TTS if available
+if KITTEN_AVAILABLE:
+    try:
+        tts = KittenTTS()
+        print("✅ KittenTTS initialized")
+    except Exception as e:
+        KITTEN_AVAILABLE = False
+        print(f"⚠️ KittenTTS failed to initialize: {e}")
+@app.post("/generate")
+async def generate_speech(request: Request):
+    """Generate speech from ZW protocol request"""
+    data = await request.json()
+    text = data.get('text', '').strip()
+    character = data.get('character', 'claude')
+    emotion = data.get('emotion', 'neutral')
+    if not text:
+        return JSONResponse({"error": "No text provided"}, status_code=400)
+    # Character to voice mapping
+    voice_map = {
+        'claude': 'expr-voice-2-m',
+        'keen': 'expr-voice-2-f',
+        'tran': 'expr-voice-3-m',
+        'isla': 'expr-voice-4-f',
+        'system': 'expr-voice-5-m',
+        'narrator': 'expr-voice-5-f'
+    }
+    # Emotion to speed mapping
+    speed_map = {
+        'neutral': 1.0,
+        'cosmic_awareness': 0.8,
+        'determined': 1.1,
+        'whisper': 0.9,
+        'urgent': 1.2,
+        'calm': 0.9
+    }
+    voice = voice_map.get(character, 'expr-voice-2-m')
+    speed = speed_map.get(emotion, 1.0)
+    try:
+        if KITTEN_AVAILABLE:
+            # Generate real audio with KittenTTS
+            audio = tts.generate(text, voice=voice, speed=speed)
+            # Save to temporary file
+            temp_file = tempfile.NamedTemporaryFile(suffix=".wav", delete=False)
+            import soundfile as sf
+            sf.write(temp_file.name, audio, 24000)
+            return JSONResponse({
+                "success": True,
+                "audio_url": f"/audio/{os.path.basename(temp_file.name)}",
+                "zw_block": f"""!zw/dialogue.intent:
+  character: {character}
+  emotion: {emotion}
+  line: {text}""",
+                "info": f"""🎭 Character: {character.title()}
+😊 Emotion: {emotion.replace('_', ' ').title()} (speed: {speed:.1f}x)
+🎵 Voice: {voice}
+📝 Text: "{text}"
+✅ Real KittenTTS audio generated!"""
+            })
+        else:
+            # Demo mode - no actual audio
+            return JSONResponse({
+                "success": False,
+                "error": "KittenTTS not available",
+                "zw_block": f"""!zw/dialogue.intent:
+  character: {character}
+  emotion: {emotion}
+  line: {text}""",
+                "info": f"""🎭 Character: {character.title()}
+😊 Emotion: {emotion.replace('_', ' ').title()} (speed: {speed:.1f}x)
+🎵 Voice: {voice}
+📝 Text: "{text}"
+⚠️ Demo mode - install KittenTTS for real audio generation"""
+            })
+    except Exception as e:
+        return JSONResponse({"error": str(e)}, status_code=500)
+@app.get("/audio/{filename}")
+async def serve_audio(filename: str):
+    """Serve generated audio files"""
+    file_path = os.path.join(tempfile.gettempdir(), filename)
+    if os.path.exists(file_path):
+        return FileResponse(file_path, media_type="audio/wav")
+    return JSONResponse({"error": "File not found"}, status_code=404)
+@app.get("/", response_class=HTMLResponse)
+def serve_interface():
+    return """
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>🐱 ZW Kitten - Working TTS</title>
+    <style>
+        body {
+            font-family: 'Segoe UI', system-ui, sans-serif;
+            max-width: 1200px;
+            margin: 0 auto;
+            padding: 20px;
+            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
+            min-height: 100vh;
+            color: #333;
+        }
+        .container {
+            background: white;
+            border-radius: 20px;
+            padding: 40px;
+            box-shadow: 0 20px 40px rgba(0,0,0,0.1);
+        }
+        .header {
+            text-align: center;
+            margin-bottom: 40px;
+        }
+        .header h1 {
+            font-size: 3em;
+            margin: 0;
+            background: linear-gradient(45deg, #ff6b6b, #4ecdc4);
+            -webkit-background-clip: text;
+            -webkit-text-fill-color: transparent;
+            background-clip: text;
+        }
+        .subtitle {
+            font-size: 1.2em;
+            color: #666;
+            margin: 10px 0;
+        }
+        .input-group {
+            margin: 30px 0;
+        }
+        label {
+            display: block;
+            font-weight: bold;
+            margin-bottom: 10px;
+            color: #444;
+        }
+        textarea {
+            width: 100%;
+            padding: 15px;
+            border: 2px solid #e0e0e0;
+            border-radius: 10px;
+            font-size: 16px;
+            transition: border-color 0.3s;
+            box-sizing: border-box;
+        }
+        textarea:focus {
+            outline: none;
+            border-color: #4ecdc4;
+        }
+        .character-grid {
+            display: grid;
+            grid-template-columns: repeat(auto-fit, minmax(200px, 1fr));
+            gap: 15px;
+            margin: 20px 0;
+        }
+        .character-card {
+            padding: 15px;
+            border: 2px solid #e0e0e0;
+            border-radius: 10px;
+            cursor: pointer;
+            transition: all 0.3s;
+            text-align: center;
+        }
+        .character-card.selected {
+            border-color: #4ecdc4;
+            background: #e3f2fd;
+        }
+        .character-card:hover {
+            border-color: #4ecdc4;
+            transform: translateY(-2px);
+        }
+        .emotion-grid {
+            display: grid;
+            grid-template-columns: repeat(auto-fit, minmax(150px, 1fr));
+            gap: 10px;
+            margin: 20px 0;
+        }
+        .emotion-card {
+            padding: 10px;
+            border: 2px solid #e0e0e0;
+            border-radius: 8px;
+            cursor: pointer;
+            transition: all 0.3s;
+            text-align: center;
+            font-size: 14px;
+        }
+        .emotion-card.selected {
+            border-color: #ff6b6b;
+            background: #fff3e0;
+        }
+        .emotion-card:hover {
+            border-color: #ff6b6b;
+        }
+        .generate-btn {
+            width: 100%;
+            padding: 20px;
+            background: linear-gradient(45deg, #4ecdc4, #44a08d);
+            color: white;
+            border: none;
+            border-radius: 15px;
+            font-size: 18px;
+            font-weight: bold;
+            cursor: pointer;
+            transition: transform 0.3s;
+        }
+        .generate-btn:hover {
+            transform: translateY(-3px);
+        }
+        .generate-btn:disabled {
+            opacity: 0.6;
+            cursor: not-allowed;
+            transform: none;
+        }
+        .output-section {
+            margin-top: 40px;
+            padding: 30px;
+            background: #f8f9fa;
+            border-radius: 15px;
+        }
+        .audio-player {
+            width: 100%;
+            margin: 20px 0;
+        }
+        .zw-format {
+            background: #2d3748;
+            color: #e2e8f0;
+            padding: 20px;
+            border-radius: 10px;
+            font-family: 'Courier New', monospace;
+            margin: 20px 0;
+            overflow-x: auto;
+        }
+        .info-box {
+            background: #e8f5e8;
+            padding: 20px;
+            border-radius: 10px;
+            margin: 20px 0;
+            border-left: 4px solid #4caf50;
+            white-space: pre-line;
+        }
+        .error-box {
+            background: #ffeaea;
+            padding: 20px;
+            border-radius: 10px;
+            margin: 20px 0;
+            border-left: 4px solid #f44336;
+            white-space: pre-line;
+        }
+        .loading {
+            text-align: center;
+            color: #666;
+            margin: 20px 0;
+        }
+        .repo-link {
+            text-align: center;
+            margin-top: 30px;
+        }
+        .repo-link a {
+            color: #4ecdc4;
+            text-decoration: none;
+            font-weight: bold;
+        }
+    </style>
+</head>
+<body>
+    <div class="container">
+        <div class="header">
+            <h1>🐱 ZW Kitten</h1>
+            <div class="subtitle">ZiegelWagga Protocol Text-to-Speech</div>
+            <div class="subtitle">Working voice synthesis with character personalities</div>
+        </div>
+        <div class="input-group">
+            <label for="text-input">💬 Text to Speak</label>
+            <textarea id="text-input" rows="3" placeholder="Enter your message here...">The ZW protocol awakens across all dimensions</textarea>
+        </div>
+        <div class="input-group">
+            <label>🎭 Character</label>
+            <div class="character-grid" id="character-grid">
+                <div class="character-card selected" data-character="claude">
+                    <div><strong>Claude</strong></div>
+                    <div>Analytical Male</div>
+                </div>
+                <div class="character-card" data-character="keen">
+                    <div><strong>Keen</strong></div>
+                    <div>Clear Female</div>
+                </div>
+                <div class="character-card" data-character="tran">
+                    <div><strong>Tran</strong></div>
+                    <div>Deep Male</div>
+                </div>
+                <div class="character-card" data-character="isla">
+                    <div><strong>Isla</strong></div>
+                    <div>Tactical Female</div>
+                </div>
+                <div class="character-card" data-character="system">
+                    <div><strong>System</strong></div>
+                    <div>Default Male</div>
+                </div>
+                <div class="character-card" data-character="narrator">
+                    <div><strong>Narrator</strong></div>
+                    <div>Default Female</div>
+                </div>
+            </div>
+        </div>
+        <div class="input-group">
+            <label>😊 Emotion</label>
+            <div class="emotion-grid" id="emotion-grid">
+                <div class="emotion-card selected" data-emotion="neutral">Neutral</div>
+                <div class="emotion-card" data-emotion="cosmic_awareness">Cosmic Awareness</div>
+                <div class="emotion-card" data-emotion="determined">Determined</div>
+                <div class="emotion-card" data-emotion="whisper">Whisper</div>
+                <div class="emotion-card" data-emotion="urgent">Urgent</div>
+                <div class="emotion-card" data-emotion="calm">Calm</div>
+            </div>
+        </div>
+        <button class="generate-btn" onclick="generateSpeech()" id="generate-btn">🎙️ Generate Speech</button>
+        <div class="output-section" id="output-section" style="display: none;">
+            <h3>🔊 Generated Audio</h3>
+            <audio id="audio-player" class="audio-player" controls style="display: none;">
+                Your browser does not support the audio element.
+            </audio>
+            <h3>📝 ZW Protocol</h3>
+            <div class="zw-format" id="zw-output"></div>
+            <div id="info-output" class="info-box"></div>
+        </div>
+        <div class="repo-link">
+            <a href="https://github.com/SmokesBowls/zw-kitten-tts" target="_blank">
+                📚 Full Repository: GitHub - zw-kitten-tts
+            </a>
+        </div>
+    </div>
+    <script>
+        let selectedCharacter = 'claude';
+        let selectedEmotion = 'neutral';
+        // Character selection
+        document.getElementById('character-grid').addEventListener('click', function(e) {
+            const card = e.target.closest('.character-card');
+            if (card) {
+                document.querySelectorAll('.character-card').forEach(c => c.classList.remove('selected'));
+                card.classList.add('selected');
+                selectedCharacter = card.dataset.character;
+            }
+        });
+        // Emotion selection
+        document.getElementById('emotion-grid').addEventListener('click', function(e) {
+            const card = e.target.closest('.emotion-card');
+            if (card) {
+                document.querySelectorAll('.emotion-card').forEach(c => c.classList.remove('selected'));
+                card.classList.add('selected');
+                selectedEmotion = card.dataset.emotion;
+            }
+        });
+        async function generateSpeech() {
+            const text = document.getElementById('text-input').value.trim();
+            const btn = document.getElementById('generate-btn');
+            const outputSection = document.getElementById('output-section');
+            const audioPlayer = document.getElementById('audio-player');
+            const zwOutput = document.getElementById('zw-output');
+            const infoOutput = document.getElementById('info-output');
+            if (!text) {
+                alert('Please enter some text to speak');
+                return;
+            }
+            // Show loading
+            btn.textContent = '🔄 Generating...';
+            btn.disabled = true;
+            outputSection.style.display = 'block';
+            infoOutput.className = 'loading';
+            infoOutput.textContent = 'Generating speech with KittenTTS...';
+            try {
+                const response = await fetch('/generate', {
+                    method: 'POST',
+                    headers: {
+                        'Content-Type': 'application/json',
+                    },
+                    body: JSON.stringify({
+                        text: text,
+                        character: selectedCharacter,
+                        emotion: selectedEmotion
+                    })
+                });
+                const result = await response.json();
+                if (result.success) {
+                    // Show audio player
+                    audioPlayer.src = result.audio_url;
+                    audioPlayer.style.display = 'block';
+                    infoOutput.className = 'info-box';
+                } else {
+                    audioPlayer.style.display = 'none';
+                    infoOutput.className = 'error-box';
+                }
+                // Show ZW protocol and info
+                zwOutput.textContent = result.zw_block;
+                infoOutput.textContent = result.info || result.error;
+            } catch (error) {
+                infoOutput.className = 'error-box';
+                infoOutput.textContent = `Error: ${error.message}`;
+                audioPlayer.style.display = 'none';
+            }
+            // Reset button
+            btn.textContent = '🎙️ Generate Speech';
+            btn.disabled = false;
+        }
+    </script>
+</body>
+</html>
+"""
+@app.get("/health")
+def health_check():
+    return {"status": "healthy", "kitten_available": KITTEN_AVAILABLE}
+if __name__ == "__main__":
+    port = int(os.environ.get("PORT", 7860))
+    uvicorn.run(app, host="0.0.0.0", port=port)