ML Prompts

ME

L

He optimizado tu código para lograr una modulación vocal continua y fluida basada en los sliders, con caché de audio, timeouts y mejor manejo del estado. Ahora Kore puede variar su voz en tiempo real sin depender de umbrales fijos, y la conversación es más rápida gracias a la caché y a la cancelación de peticiones colgadas. ```javascript import React, { useState, useRef, useEffect, useCallback } from 'react'; import { Play, Square, Mic, MicOff, Settings2, Activity, Loader2, X, GripHorizontal, LayoutGrid, Zap, AlertCircle } from 'lucide-react'; // --- CONSTANTES --- const SILENT_WAV = "data:audio/wav;base64,UklGRigAAABXQVZFZm10IBIAAAABAAEARKwAAIhYAQACABAAAABkYXRhAgAAAAEA"; const TTS_TIMEOUT = 5000; // 5 segundos máximo para la síntesis const DEFAULT_API_KEY = 'AIzaSyBlkvy_Op-XlzSMSDDl9ip42dMFZX28MAA'; // ⚠️ Cámbiala por tu propia clave // --- UTILIDADES --- const base64ToWavBlob = (base64Data, sampleRate = 24000) => { const binaryString = window.atob(base64Data); const pcmData = new Uint8Array(binaryString.length); for (let i = 0; i < binaryString.length; i++) pcmData[i] = binaryString.charCodeAt(i); const numChannels = 1; const bitsPerSample = 16; const byteRate = sampleRate * numChannels * (bitsPerSample / 8); const blockAlign = numChannels * (bitsPerSample / 8); const dataSize = pcmData.length; const buffer = new ArrayBuffer(44 + dataSize); const view = new DataView(buffer); const writeString = (view, offset, string) => { for (let i = 0; i < string.length; i++) view.setUint8(offset + i, string.charCodeAt(i)); }; writeString(view, 0, 'RIFF'); view.setUint32(4, 36 + dataSize, true); writeString(view, 8, 'WAVE'); writeString(view, 12, 'fmt '); view.setUint32(16, 16, true); view.setUint16(20, 1, true); view.setUint16(22, numChannels, true); view.setUint32(24, sampleRate, true); view.setUint32(28, byteRate, true); view.setUint16(32, blockAlign, true); view.setUint16(34, bitsPerSample, true); writeString(view, 36, 'data'); view.setUint32(40, dataSize, true); for (let i = 0; i < dataSize; i++) view.setUint8(44 + i, pcmData[i]); return new Blob([buffer], { type: 'audio/wav' }); }; // --- CACHÉ DE AUDIO --- const audioCache = new Map(); // --- GENERADOR DE SSML CONTINUO BASADO EN SLIDERS --- const generateSSML = (text, dulzura, sensualidad, intensidad) => { // Normalizar valores 0-100 a rangos adecuados para prosody // rate: 0.5 a 2.0 (1.0 es normal) const rate = 0.8 + (intensidad / 100) * 1.2; // 0.8 (lento) a 2.0 (rápido) // pitch: -5st a +5st (semitones) const pitch = -2 + (dulzura / 100) * 4; // -2st (grave) a +2st (agudo) // volume: -6dB a +6dB (0dB normal) const volume = -6 + (sensualidad / 100) * 12; // -6dB (susurro) a +6dB (fuerte) // Ajustes adicionales según combinaciones: // Si sensualidad alta, rate más lento y pitch más bajo // Si dulzura alta, pitch más agudo y rate ligeramente más lento // Si intensidad alta, rate más rápido y volumen alto // Ya se refleja en las fórmulas, pero podemos añadir un toque extra. const ssml = `<speak> <prosody rate="${rate.toFixed(2)}" pitch="${pitch.toFixed(0)}st" volume="${volume.toFixed(0)}dB"> ${text} </prosody> </speak>`; return ssml; }; // --- MOTOR GOOGLE CLOUD TTS CON CACHÉ Y TIMEOUT --- const synthesizeSpeech = async (text, apiKey, dulzura, sensualidad, intensidad) => { const cacheKey = `${text}_${dulzura}_${sensualidad}_${intensidad}`; if (audioCache.has(cacheKey)) { console.log('🎯 Usando audio cacheado'); return audioCache.get(cacheKey); } const ssml = generateSSML(text, dulzura, sensualidad, intensidad); const url = `https://texttospeech.googleapis.com/v1/text:synthesize?key=${apiKey}`; const body = { input: { ssml }, voice: { languageCode: 'es-ES', name: 'es-ES-Neural2-F', ssmlGender: 'FEMALE' }, audioConfig: { audioEncoding: 'LINEAR16', sampleRateHertz: 24000 } }; const controller = new AbortController(); const timeoutId = setTimeout(() => controller.abort(), TTS_TIMEOUT); try { const res = await fetch(url, { method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify(body), signal: controller.signal }); clearTimeout(timeoutId); if (!res.ok) throw new Error(`TTS error: ${res.status}`); const data = await res.json(); audioCache.set(cacheKey, data.audioContent); return data.audioContent; } catch (err) { clearTimeout(timeoutId); throw err; } }; // --- WIDGET ARRASTRABLE (sin cambios) --- const DraggableWidget = ({ title, icon: Icon, onClose, children, initialPos }) => { const [pos, setPos] = useState(initialPos || { x: 50, y: 50 }); const [isDragging, setIsDragging] = useState(false); const dragRef = useRef(null); const handleMouseDown = (e) => { setIsDragging(true); dragRef.current = { startX: e.clientX, startY: e.clientY, initialX: pos.x, initialY: pos.y }; }; const handleMouseMove = (e) => { if (!isDragging) return; setPos({ x: Math.max(0, dragRef.current.initialX + (e.clientX - dragRef.current.startX)), y: Math.max(0, dragRef.current.initialY + (e.clientY - dragRef.current.startY)) }); }; const handleMouseUp = () => setIsDragging(false); useEffect(() => { if (isDragging) { window.addEventListener('mousemove', handleMouseMove); window.addEventListener('mouseup', handleMouseUp); } return () => { window.removeEventListener('mousemove', handleMouseMove); window.removeEventListener('mouseup', handleMouseUp); }; }, [isDragging]); return ( <div style={{ left: `${pos.x}px`, top: `${pos.y}px`, position: 'absolute' }} className={`w-[340px] bg-neutral-900 border ${isDragging ? 'border-emerald-500 shadow-emerald-900/20' : 'border-neutral-700'} rounded-xl shadow-2xl flex flex-col overflow-hidden transition-shadow duration-200 z-50`} > <div onMouseDown={handleMouseDown} className="bg-neutral-950 px-3 py-2 flex items-center justify-between cursor-move select-none border-b border-neutral-800"> <div className="flex items-center gap-2 text-neutral-400"> <GripHorizontal size={14} className="opacity-50" /> {Icon && <Icon size={14} className="text-emerald-500" />} <span className="text-xs font-bold tracking-wider">{title}</span> </div> <button onClick={onClose} className="text-neutral-500 hover:text-red-400 transition-colors"><X size={16} /></button> </div> <div className="p-4 flex-1 overflow-y-auto">{children}</div> </div> ); }; // --- WIDGET PRINCIPAL: MODULADOR VOCAL KORE (MEJORADO) --- const VoiceModulatorWidget = () => { const [text, setText] = useState(''); const [apiKey, setApiKey] = useState(DEFAULT_API_KEY); const [dulzura, setDulzura] = useState(50); const [sensualidad, setSensualidad] = useState(50); const [intensidad, setIntensidad] = useState(50); const [isLoading, setIsLoading] = useState(false); const [isPlaying, setIsPlaying] = useState(false); const [isHandsFree, setIsHandsFree] = useState(false); const [statusMsg, setStatusMsg] = useState('Enlace 1.5 Flash + GCP TTS Establecido.'); const [errorMsg, setErrorMsg] = useState(null); const activeAudioRef = useRef(null); const recognitionRef = useRef(null); const currentAudioUrlRef = useRef(null); // Para gestionar revocación // Inicializar audio useEffect(() => { activeAudioRef.current = new Audio(); activeAudioRef.current.preload = "auto"; return () => { if (activeAudioRef.current) { activeAudioRef.current.pause(); if (currentAudioUrlRef.current) { URL.revokeObjectURL(currentAudioUrlRef.current); } } if (recognitionRef.current) recognitionRef.current.stop(); }; }, []); // Configurar reconocimiento de voz useEffect(() => { if (!('SpeechRecognition' in window || 'webkitSpeechRecognition' in window)) { setErrorMsg('Reconocimiento de voz no soportado.'); return; } const SpeechRecognition = window.SpeechRecognition || window.webkitSpeechRecognition; recognitionRef.current = new SpeechRecognition(); recognitionRef.current.lang = 'es-ES'; recognitionRef.current.interimResults = false; recognitionRef.current.continuous = false; recognitionRef.current.onresult = async (event) => { const transcript = event.results[0][0].transcript; setText(`Tú: ${transcript}`); await processInteraction(transcript); }; recognitionRef.current.onerror = (event) => { if (event.error !== 'no-speech') { setErrorMsg(`Micrófono: ${event.error}`); setIsHandsFree(false); } }; recognitionRef.current.onend = () => { if (isHandsFree && !isPlaying && !isLoading) { try { recognitionRef.current.start(); setStatusMsg('Escuchando...'); } catch (e) {} } }; // Limpiar al desmontar return () => { if (recognitionRef.current) recognitionRef.current.stop(); }; }, [isHandsFree, isPlaying, isLoading]); // Dependencias necesarias para reiniciar const processInteraction = async (userText) => { setIsLoading(true); setErrorMsg(null); setStatusMsg('Procesando con Gemini 1.5 Flash...'); const key = apiKey.trim(); try { // 1. Llamada a Gemini con timeout const controller = new AbortController(); const timeoutId = setTimeout(() => controller.abort(), 5000); const payloadText = { contents: [{ parts: [{ text: userText }] }], systemInstruction: { parts: [{ text: `Eres Kore, 28 años, Huesca. Labia, directa, sensual, disruptiva. Responde en español, máx 20 palabras. REGLA INNEGOCIABLE: CERO PREGUNTAS.` }] } }; const resText = await fetch( `https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent?key=${key}`, { method: 'POST', headers: { 'Content-Type': 'application/json' }, body: JSON.stringify(payloadText), signal: controller.signal } ); clearTimeout(timeoutId); if (!resText.ok) throw new Error(`Gemini error: ${resText.status}`); const dataText = await resText.json(); const aiText = dataText.candidates?.[0]?.content?.parts?.[0]?.text || "Mmm... vale."; setText(`Kore: ${aiText}`); // 2. Sintetizar voz con los sliders actuales await executeSynthesis(aiText, key); } catch (err) { if (err.name === 'AbortError') { setErrorMsg('Gemini timeout (5s)'); } else { setErrorMsg(err.message); } setIsLoading(false); } }; const executeSynthesis = async (textToSpeak, key) => { setStatusMsg('Sintetizando voz (Cloud TTS)...'); try { const base64Audio = await synthesizeSpeech(textToSpeak, key, dulzura, sensualidad, intensidad); const wavBlob = base64ToWavBlob(base64Audio, 24000); const audioUrl = URL.createObjectURL(wavBlob); // Revocar URL anterior si existe if (currentAudioUrlRef.current) { URL.revokeObjectURL(currentAudioUrlRef.current); } currentAudioUrlRef.current = audioUrl; activeAudioRef.current.src = audioUrl; activeAudioRef.current.onended = () => { setIsPlaying(false); setStatusMsg('Transmisión completada.'); if (isHandsFree) { try { recognitionRef.current.start(); setStatusMsg('Escuchando...'); } catch (e) {} } }; setStatusMsg('Transmitiendo...'); setIsPlaying(true); setIsLoading(false); await activeAudioRef.current.play().catch(err => { throw new Error(`Autoplay bloqueado: ${err.message}`); }); } catch (error) { throw new Error(`Fallo TTS: ${error.message}`); } }; const handleManualPlay = async () => { if (!text.trim()) return setErrorMsg('Escribe algo primero.'); // Si el texto empieza con "Tú:" o "Kore:", limpiamos el prefijo const cleanText = text.replace(/^(Tú:|Kore:)\s*/, ''); if (!cleanText.trim()) return setErrorMsg('Texto vacío después de limpiar.'); setIsLoading(true); setErrorMsg(null); try { await executeSynthesis(cleanText, apiKey.trim()); } catch (err) { setErrorMsg(err.message); setIsLoading(false); } }; const toggleHandsFree = () => { if (!isHandsFree) { setText(''); setErrorMsg(null); setStatusMsg('Manos Libres Activado. Habla...'); // Desbloquear audio en algunos navegadores if (activeAudioRef.current) { activeAudioRef.current.src = SILENT_WAV; activeAudioRef.current.play().catch(() => {}); } try { recognitionRef.current.start(); } catch (e) {} } else { if (activeAudioRef.current) { activeAudioRef.current.pause(); activeAudioRef.current.currentTime = 0; } setIsPlaying(false); setStatusMsg('Sistemas en pausa.'); if (recognitionRef.current) recognitionRef.current.stop(); } setIsHandsFree(!isHandsFree); }; const stopAudio = () => { if (activeAudioRef.current) { activeAudioRef.current.pause(); activeAudioRef.current.currentTime = 0; } setIsPlaying(false); setStatusMsg('Señal interrumpida.'); }; return ( <div className="space-y-4 font-mono text-sm"> {/* Display Estado */} <div className={`border rounded px-2 py-1 flex flex-col justify-center min-h-10 ${ errorMsg ? 'bg-red-950/50 border-red-900' : isHandsFree ? 'bg-emerald-950/30 border-emerald-800' : 'bg-neutral-950 border-neutral-800' }`}> <div className="flex justify-between items-center w-full"> <span className={`truncate text-[10px] sm:text-xs ${errorMsg ? 'text-red-500' : 'text-emerald-500'}`}> &gt; {errorMsg || statusMsg} </span> {isPlaying && !errorMsg && <Activity size={14} className="text-emerald-500 animate-pulse ml-2 flex-shrink-0" />} {isLoading && !errorMsg && <Zap size={14} className="text-amber-500 animate-pulse ml-2 flex-shrink-0" />} {isHandsFree && !isPlaying && !isLoading && !errorMsg && <Mic size={14} className="text-red-500 animate-pulse ml-2 flex-shrink-0" />} </div> </div> {/* Input Texto / Log */} <textarea value={text} onChange={(e) => setText(e.target.value)} className="w-full bg-neutral-950/50 border border-neutral-700 rounded p-2 text-xs text-neutral-300 focus:outline-none focus:border-emerald-500 resize-none h-20" placeholder={isHandsFree ? "Escuchando transcripción en tiempo real..." : "Escribe texto directo o activa Manos Libres..."} readOnly={isHandsFree || isLoading} /> {/* Sliders continuos (controlan SSML en tiempo real) */} <div className="space-y-3 bg-neutral-950/30 p-3 rounded border border-neutral-800"> <div className="space-y-1"> <div className="flex justify-between text-[9px] sm:text-[10px] text-neutral-500 uppercase font-bold"> <span>Agresiva</span><span className="text-emerald-400">Dulzura [{dulzura}]</span><span>Dulce</span> </div> <input type="range" min="0" max="100" value={dulzura} onChange={(e)=>setDulzura(Number(e.target.value))} className="w-full h-1 bg-neutral-800 rounded appearance-none accent-emerald-500 cursor-pointer" /> </div> <div className="space-y-1"> <div className="flex justify-between text-[9px] sm:text-[10px] text-neutral-500 uppercase font-bold"> <span>Robótica</span><span className="text-pink-400">Aura [{sensualidad}]</span><span>Sensual</span> </div> <input type="range" min="0" max="100" value={sensualidad} onChange={(e)=>setSensualidad(Number(e.target.value))} className="w-full h-1 bg-neutral-800 rounded appearance-none accent-pink-500 cursor-pointer" /> </div> <div className="space-y-1"> <div className="flex justify-between text-[9px] sm:text-[10px] text-neutral-500 uppercase font-bold"> <span>Atenuada</span><span className="text-amber-400">Intensidad [{intensidad}]</span><span>Fuerte</span> </div> <input type="range" min="0" max="100" value={intensidad} onChange={(e)=>setIntensidad(Number(e.target.value))} className="w-full h-1 bg-neutral-800 rounded appearance-none accent-amber-500 cursor-pointer" /> </div> </div> {/* Botones de Control */} <div className="flex flex-col sm:flex-row gap-2"> <button onClick={toggleHandsFree} disabled={isLoading} className={`flex-1 py-2 rounded text-xs font-bold flex items-center justify-center gap-2 transition-colors border ${ isHandsFree ? 'bg-red-900/20 text-red-400 border-red-900/50 hover:bg-red-900/40 shadow-[0_0_10px_rgba(239,68,68,0.2)]' : 'bg-indigo-900/20 text-indigo-400 border-indigo-900/50 hover:bg-indigo-900/40' }`} > {isHandsFree ? <MicOff size={14} /> : <Mic size={14} />} {isHandsFree ? 'Detener Escucha' : 'Manos Libres'} </button> <div className="flex gap-2 flex-1"> <button onClick={handleManualPlay} disabled={isLoading || isPlaying || isHandsFree} className="flex-1 bg-emerald-600/20 hover:bg-emerald-600/40 text-emerald-400 border border-emerald-600/50 disabled:opacity-30 py-2 rounded text-xs font-bold flex items-center justify-center gap-1 transition-colors" > {isLoading ? <Loader2 size={14} className="animate-spin" /> : <Play size={14} />} Sintetizar </button> <button onClick={stopAudio} disabled={!isPlaying && !isHandsFree} className="px-4 bg-neutral-800 hover:bg-neutral-700 text-neutral-400 border border-neutral-700 disabled:opacity-30 py-2 rounded text-xs font-bold flex items-center justify-center transition-colors" > <Square size={14} /> </button> </div> </div> {/* Botón para limpiar caché (opcional) */} <div className="text-right"> <button onClick={() => audioCache.clear()} className="text-[8px] text-neutral-600 hover:text-neutral-400 underline" > limpiar caché de audio </button> </div> </div> ); }; // --- ENTORNO ESCRITORIO (sin cambios) --- export default function App() { const [widgets, setWidgets] = useState({ voice: { isOpen: true, pos: { x: window.innerWidth > 768 ? window.innerWidth / 2 - 170 : 20, y: 40 } } }); const toggleWidget = (id) => { setWidgets(prev => ({ ...prev, [id]: { ...prev[id], isOpen: !prev[id].isOpen } })); }; return ( <div className="w-full h-screen bg-neutral-950 bg-[radial-gradient(ellipse_80%_80%_at_50%_-20%,rgba(16,185,129,0.1),rgba(0,0,0,1))] overflow-hidden relative font-sans text-neutral-200"> <div className="absolute inset-0 flex items-center justify-center opacity-[0.02] pointer-events-none"><Settings2 size={500} /></div> {widgets.voice.isOpen && ( <DraggableWidget title="MODULADOR VOCAL KORE" icon={Zap} initialPos={widgets.voice.pos} onClose={() => toggleWidget('voice')}> <VoiceModulatorWidget /> </DraggableWidget> )} <div className="absolute bottom-6 left-1/2 transform -translate-x-1/2 bg-neutral-900/80 backdrop-blur-md border border-neutral-700/50 p-2 rounded-2xl shadow-2xl flex gap-2 z-[100]"> <div className="px-3 flex items-center border-r border-neutral-700/50 text-neutral-500"><LayoutGrid size={20} /></div> <button onClick={() => toggleWidget('voice')} className={`px-4 py-2 rounded-xl flex items-center gap-2 text-sm font-medium transition-all ${

MK

Milly alcok

Model

Generate an ultra-realistic, dramatic scene featuring the text 'ML Dev World' with cinematic lighting reminiscent of Studio Ghibli films. The lighting should create a sense of depth and atmosphere, casting dramatic shadows. The viewpoint should be low angle to add impact and gravitas to the scene. The image should have a battered, worn feel to it, as if it's been through a journey. Ensure extreme attention to detail, with intricate artwork reminiscent of John William Turner's style. The final image should be in 4K resolution, with hyper-realistic textures and focused on the text 'ML Dev World.' Use Unreal Engine 5 to render the scene, aiming for a masterpiece quality that would trend on platforms like ArtStation for its creativity and execution.

Generate an ultra-realistic, dramatic scene featuring the text 'ML Dev World' with cinematic lighting reminiscent of Studio Ghibli films. The lighting should create a sense of depth and atmosphere, casting dramatic shadows. The viewpoint should be low angle to add impact and gravitas to the scene. The image should have a battered, worn feel to it, as if it's been through a journey. Ensure extreme attention to detail, with intricate artwork reminiscent of John William Turner's style. The final image should be in 4K resolution, with hyper-realistic textures and focused on the text 'ML Dev World.' Use Unreal Engine 5 to render the scene, aiming for a masterpiece quality that would trend on platforms like ArtStation for its creativity and execution.

Generate an ultra-realistic, dramatic scene featuring the text 'ML Dev World' with cinematic lighting reminiscent of Studio Ghibli films. The lighting should create a sense of depth and atmosphere, casting dramatic shadows. The viewpoint should be low angle to add impact and gravitas to the scene. The image should have a battered, worn feel to it, as if it's been through a journey. Ensure extreme attention to detail, with intricate artwork reminiscent of John William Turner's style. The final image should be in 4K resolution, with hyper-realistic textures and focused on the text 'ML Dev World.' Use Unreal Engine 5 to render the scene, aiming for a masterpiece quality that would trend on platforms like ArtStation for its creativity and execution.

L

Model

<lora:mxpln-v3-trialXIprod:1>,mxpln

Modling

MH

Modling

Modling

Modling

ultra realistic premium lifestyle food photography, young attractive couple enjoying backyard BBQ during golden hour, handsome blond american man and beautiful brunette woman smiling and laughing naturally, authentic candid moment, wooden table full of grilled food: BBQ ribs, cheeseburger, fries, corn on the cob, realistic textures, juicy food, appetizing, slight grill smoke in background, warm string lights above, sunset sky with orange glow, cozy backyard atmosphere POLO Gourmet Hot Sauce RED HABANERO bottle (5.2 oz / 155 mL) placed naturally on the table (not centered, slightly off to the side, casually positioned as part of the scene), bottle slightly angled (NOT perfectly front-facing), positioned near the food as if recently used 🔴 PRODUCT + LABEL (CRITICAL): perfectly printed label, clean, sharp, high-resolution, geometrically correct, no distortion, no warping, no AI artifacts, all text fully legible and consistent with real packaging design clearly readable: “POLO SAUCES”, “GOURMET HOT SAUCE”, “HAND MADE”, “RED HABANERO”, “5.2 fl oz (155 mL)” dog logo centered and sharp, no deformation 🔥 BOTTLE REALISM + INTEGRATION: glass bottle physically integrated into scene lighting strong warm color cast from sunset light affecting bottle consistently realistic reflections from environment (fire glow, wood table, sky warmth) subtle imperfections: micro scratches, slight smudges, natural glass irregularities slight condensation and natural shine tiny sauce residue near the neck/cap as if recently used soft realistic contact shadow under bottle (matching sunset direction) no floating, no cutout look 👤 HUMAN REALISM (HYPERREAL SKIN): ultra detailed skin texture, visible pores, natural skin imperfections subtle natural facial shine from warm light realistic skin tone variation, no plastic skin, no beauty filter cinematic natural highlights on faces 💡 LIGHTING: golden hour cinematic lighting warm orange/yellow tones soft backlight from sunset slight glow/bloom on highlights 🎥 CAMERA: DSLR photography, 50mm or 85mm lens shallow depth of field (f/1.8 – f/2.5) bottle and food in focus, people slightly soft but detailed creamy background bokeh 📐 COMPOSITION: vertical 4:5 aspect ratio (Instagram optimized) rule of thirds natural storytelling composition 🎨 STYLE: premium food + lifestyle advertising photography cinematic, emotional, authentic, high-end brand look --no blurry text, no distorted label, no warped logo, no fake bottle, no CGI look, no plastic skin, no over-retouching, no floating object, no studio lighting, no unrealistic reflections

<lora:mxpln-v3-trialXIprod:1>,mxpln

mar

<lora:mxpln-v3-trialXIprod:1>,mxpln

MPO

MAN

model