Create components.py

ui/components.py

@@ -0,0 +1,1447 @@
+# --- 기존 코드 유지 (원본) ---
+"""
+ACE-Step: A Step Towards Music Generation Foundation Model
+
+https://github.com/ace-step/ACE-Step
+
+Apache 2.0 License
+"""
+
+import gradio as gr
+import librosa
+import os
+import random
+import hashlib
+import numpy as np
+import json
+from typing import Dict, List, Tuple, Optional
+
+# 추가된 부분: OpenAI 모듈 임포트
+from openai import OpenAI
+
+# os.getenv("LLM_API")에서 API 키를 불러와서 사용
+client = OpenAI(api_key=os.getenv("LLM_API"))
+
+# 가사 생성을 위한 OpenAI 호출 함수
+def openai_generate_lyrics(topic: str) -> str:
+    """
+    사용자가 입력한 '주제(topic)'를 바탕으로 GPT-4.1-mini 모델을 호출하여
+    해당 언어(한국어/영어 등) 기준으로 [verse], [chorus], [bridge] 등의
+    구조를 포함한 노래 가사를 생성하여 반환합니다.
+    """
+    # 시스템 프롬프트(명령) : 노래 가사를 작사하는 전문가 역할
+    system_prompt = (
+        "너는 노래 가사를 작사하는 전문가 역할이다. 이용자가 입력하는 주제에 따라 "
+        "이에 관련된 노래 가사를 작성하라. 가사의 규칙은 \"[  ]\"로 구분하여, "
+        "다음 예시를 참조하라. \"\"\"[verse]\nNeon lights they flicker bright\nCity "
+        "hums in dead of night\nRhythms pulse through concrete veins\nLost in echoes "
+        "of refrains\n\n[verse]\nBassline groovin' in my chest\nHeartbeats match the "
+        "city's zest\nElectric whispers fill the air\nSynthesized dreams everywhere\n\n"
+        "[chorus]\nTurn it up and let it flow\nFeel the fire let it grow\nIn this rhythm "
+        "we belong\nHear the night sing out our song\n\n[verse]\nGuitar strings they "
+        "start to weep\nWake the soul from silent sleep\nEvery note a story told\nIn "
+        "this night we're bold and gold\n\n[bridge]\nVoices blend in harmony\nLost in "
+        "pure cacophony\nTimeless echoes timeless cries\nSoulful shouts beneath the "
+        "skies\n\n[verse]\nKeyboard dances on the keys\nMelodies on evening breeze\nCatch "
+        "the tune and hold it tight\nIn this moment we take flight\n\"\"\""
+    )
+    try:
+        # GPT-4.1-mini 모델에 system / user 메시지 형식으로 호출
+        response = client.responses.create(
+            model="gpt-4.1-mini",
+            input=[
+                {
+                    "role": "system",
+                    "content": [
+                        {
+                            "type": "input_text",
+                            "text": system_prompt
+                        }
+                    ]
+                },
+                {
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": "input_text",
+                            "text": topic
+                        }
+                    ]
+                }
+            ],
+            text={"format": {"type": "text"}},
+            reasoning={},
+            tools=[],
+            temperature=1,
+            max_output_tokens=2048,
+            top_p=1,
+            store=True
+        )
+        # response 구조상 마지막 assistant 응답이 input 리스트의 마지막 원소로 들어옴
+        # role이 assistant인 항목을 찾아서 'text' 필드를 추출
+        if "input" in response and len(response["input"]) > 0:
+            # 가장 마지막 메시지가 assistant일 것이라 가정
+            last_msg = response["input"][-1]
+            if last_msg.get("role") == "assistant":
+                # content 리스트 중 첫 번째에 우리가 원하는 가사가 들어있을 것
+                content = last_msg.get("content", [])
+                if len(content) > 0 and content[0].get("type") == "output_text":
+                    return content[0].get("text", "")
+        return "가사 생성 실패: 응답 형식을 파악할 수 없습니다."
+    except Exception as e:
+        return f"가사 생성 실패: {str(e)}"
+
+TAG_DEFAULT = "funk, pop, soul, rock, melodic, guitar, drums, bass, keyboard, percussion, 105 BPM, energetic, upbeat, groovy, vibrant, dynamic"
+LYRIC_DEFAULT = """[verse]
+Neon lights they flicker bright
+City hums in dead of night
+Rhythms pulse through concrete veins
+Lost in echoes of refrains
+
+[verse]
+Bassline groovin' in my chest
+Heartbeats match the city's zest
+Electric whispers fill the air
+Synthesized dreams everywhere
+
+[chorus]
+Turn it up and let it flow
+Feel the fire let it grow
+In this rhythm we belong
+Hear the night sing out our song
+
+[verse]
+Guitar strings they start to weep
+Wake the soul from silent sleep
+Every note a story told
+In this night we're bold and gold
+
+[bridge]
+Voices blend in harmony
+Lost in pure cacophony
+Timeless echoes timeless cries
+Soulful shouts beneath the skies
+
+[verse]
+Keyboard dances on the keys
+Melodies on evening breeze
+Catch the tune and hold it tight
+In this moment we take flight
+"""
+
+# 확장된 장르 프리셋 (기존 + 개���된 태그)
+GENRE_PRESETS = {
+    "Modern Pop": "pop, synth, drums, guitar, 120 bpm, upbeat, catchy, vibrant, female vocals, polished vocals, radio-ready, commercial, layered vocals",
+    "Rock": "rock, electric guitar, drums, bass, 130 bpm, energetic, rebellious, gritty, male vocals, raw vocals, power chords, driving rhythm",
+    "Hip Hop": "hip hop, 808 bass, hi-hats, synth, 90 bpm, bold, urban, intense, male vocals, rhythmic vocals, trap beats, punchy drums",
+    "Country": "country, acoustic guitar, steel guitar, fiddle, 100 bpm, heartfelt, rustic, warm, male vocals, twangy vocals, storytelling, americana",
+    "EDM": "edm, synth, bass, kick drum, 128 bpm, euphoric, pulsating, energetic, instrumental, progressive build, festival anthem, electronic",
+    "Reggae": "reggae, guitar, bass, drums, 80 bpm, chill, soulful, positive, male vocals, smooth vocals, offbeat rhythm, island vibes",
+    "Classical": "classical, orchestral, strings, piano, 60 bpm, elegant, emotive, timeless, instrumental, dynamic range, sophisticated harmony",
+    "Jazz": "jazz, saxophone, piano, double bass, 110 bpm, smooth, improvisational, soulful, male vocals, crooning vocals, swing feel, sophisticated",
+    "Metal": "metal, electric guitar, double kick drum, bass, 160 bpm, aggressive, intense, heavy, male vocals, screamed vocals, distorted, powerful",
+    "R&B": "r&b, synth, bass, drums, 85 bpm, sultry, groovy, romantic, female vocals, silky vocals, smooth production, neo-soul"
+}
+
+# 품질 프리셋 시스템 추가
+QUALITY_PRESETS = {
+    "Draft (Fast)": {
+        "infer_step": 50,
+        "guidance_scale": 10.0,
+        "scheduler_type": "euler",
+        "omega_scale": 5.0,
+        "use_erg_diffusion": False,
+        "use_erg_tag": True,
+        "description": "빠른 초안 생성 (1-2분)"
+    },
+    "Standard": {
+        "infer_step": 150,
+        "guidance_scale": 15.0,
+        "scheduler_type": "euler",
+        "omega_scale": 10.0,
+        "use_erg_diffusion": True,
+        "use_erg_tag": True,
+        "description": "표준 품질 (3-5분)"
+    },
+    "High Quality": {
+        "infer_step": 200,
+        "guidance_scale": 18.0,
+        "scheduler_type": "heun",
+        "omega_scale": 15.0,
+        "use_erg_diffusion": True,
+        "use_erg_tag": True,
+        "description": "고품질 생성 (8-12분)"
+    },
+    "Ultra (Best)": {
+        "infer_step": 299,
+        "guidance_scale": 20.0,
+        "scheduler_type": "heun",
+        "omega_scale": 20.0,
+        "use_erg_diffusion": True,
+        "use_erg_tag": True,
+        "description": "최고 품질 (15-20분)"
+    }
+}
+
+# 다중 시드 생성 설정
+MULTI_SEED_OPTIONS = {
+    "Single": 1,
+    "Best of 3": 3,
+    "Best of 5": 5,
+    "Best of 10": 10
+}
+
+class MusicGenerationCache:
+    """생성 결과 캐싱 시스템"""
+    def __init__(self):
+        self.cache = {}
+        self.max_cache_size = 50
+    
+    def get_cache_key(self, params):
+        # 중요한 파라미터만으로 해시 생성
+        key_params = {k: v for k, v in params.items() 
+                     if k in ['prompt', 'lyrics', 'infer_step', 'guidance_scale', 'audio_duration']}
+        return hashlib.md5(str(sorted(key_params.items())).encode()).hexdigest()[:16]
+    
+    def get_cached_result(self, params):
+        key = self.get_cache_key(params)
+        return self.cache.get(key)
+    
+    def cache_result(self, params, result):
+        if len(self.cache) >= self.max_cache_size:
+            oldest_key = next(iter(self.cache))
+            del self.cache[oldest_key]
+        
+        key = self.get_cache_key(params)
+        self.cache[key] = result
+
+# 전역 캐시 인스턴스
+generation_cache = MusicGenerationCache()
+
+def enhance_prompt_with_genre(base_prompt: str, genre: str) -> str:
+    """장르에 따른 스마트 프롬프트 확장"""
+    if genre == "Custom" or not genre:
+        return base_prompt
+    
+    # 장르별 추가 개선 태그
+    genre_enhancements = {
+        "Modern Pop": ["polished production", "mainstream appeal", "hook-driven"],
+        "Rock": ["guitar-driven", "powerful drums", "energetic performance"],
+        "Hip Hop": ["rhythmic flow", "urban atmosphere", "bass-heavy"],
+        "Country": ["acoustic warmth", "storytelling melody", "authentic feel"],
+        "EDM": ["electronic atmosphere", "build-ups", "dance-friendly"],
+        "Reggae": ["laid-back groove", "tropical vibes", "rhythmic guitar"],
+        "Classical": ["orchestral depth", "musical sophistication", "timeless beauty"],
+        "Jazz": ["musical complexity", "improvisational spirit", "sophisticated harmony"],
+        "Metal": ["aggressive energy", "powerful sound", "intense atmosphere"],
+        "R&B": ["smooth groove", "soulful expression", "rhythmic sophistication"]
+    }
+    
+    if genre in genre_enhancements:
+        additional_tags = ", ".join(genre_enhancements[genre])
+        return f"{base_prompt}, {additional_tags}"
+    
+    return base_prompt
+
+def calculate_quality_score(audio_path: str) -> float:
+    """간단한 품질 점수 계산 (실제 구현에서는 더 복잡한 메트릭 사용)"""
+    try:
+        y, sr = librosa.load(audio_path)
+        
+        # 기본 품질 메트릭
+        rms_energy = np.sqrt(np.mean(y**2))
+        spectral_centroid = np.mean(librosa.feature.spectral_centroid(y=y, sr=sr))
+        zero_crossing_rate = np.mean(librosa.feature.zero_crossing_rate(y))
+        
+        # 정규화된 점수 (0-100)
+        energy_score = min(rms_energy * 1000, 40)  # 0-40점
+        spectral_score = min(spectral_centroid / 100, 40)  # 0-40점  
+        clarity_score = min((1 - zero_crossing_rate) * 20, 20)  # 0-20점
+        
+        total_score = energy_score + spectral_score + clarity_score
+        return round(total_score, 1)
+    except:
+        return 50.0  # 기본값
+
+def update_tags_from_preset(preset_name):
+    if preset_name == "Custom":
+        return ""
+    return GENRE_PRESETS.get(preset_name, "")
+
+def update_quality_preset(preset_name):
+    """품질 프리셋 적용"""
+    if preset_name not in QUALITY_PRESETS:
+        return (100, 15.0, "euler", 10.0, True, True)
+    
+    preset = QUALITY_PRESETS[preset_name]
+    return (
+        preset.get("infer_step", 100),
+        preset.get("guidance_scale", 15.0),
+        preset.get("scheduler_type", "euler"),
+        preset.get("omega_scale", 10.0),
+        preset.get("use_erg_diffusion", True),
+        preset.get("use_erg_tag", True)
+    )
+
+def create_enhanced_process_func(original_func):
+    """기존 함수를 향상된 기능으로 래핑"""
+    
+    def enhanced_func(
+        audio_duration, prompt, lyrics, infer_step, guidance_scale,
+        scheduler_type, cfg_type, omega_scale, manual_seeds,
+        guidance_interval, guidance_interval_decay, min_guidance_scale,
+        use_erg_tag, use_erg_lyric, use_erg_diffusion, oss_steps,
+        guidance_scale_text, guidance_scale_lyric,
+        audio2audio_enable=False, ref_audio_strength=0.5, ref_audio_input=None,
+        lora_name_or_path="none", multi_seed_mode="Single", 
+        enable_smart_enhancement=True, genre_preset="Custom", **kwargs
+    ):
+        # 스마트 프롬프트 확장
+        if enable_smart_enhancement and genre_preset != "Custom":
+            prompt = enhance_prompt_with_genre(prompt, genre_preset)
+        
+        # 캐시 확인
+        cache_params = {
+            'prompt': prompt, 'lyrics': lyrics, 'audio_duration': audio_duration,
+            'infer_step': infer_step, 'guidance_scale': guidance_scale
+        }
+        
+        cached_result = generation_cache.get_cached_result(cache_params)
+        if cached_result:
+            return cached_result
+        
+        # 다중 시드 생성
+        num_candidates = MULTI_SEED_OPTIONS.get(multi_seed_mode, 1)
+        
+        if num_candidates == 1:
+            # 기존 함수 호출
+            result = original_func(
+                audio_duration, prompt, lyrics, infer_step, guidance_scale,
+                scheduler_type, cfg_type, omega_scale, manual_seeds,
+                guidance_interval, guidance_interval_decay, min_guidance_scale,
+                use_erg_tag, use_erg_lyric, use_erg_diffusion, oss_steps,
+                guidance_scale_text, guidance_scale_lyric, audio2audio_enable,
+                ref_audio_strength, ref_audio_input, lora_name_or_path, **kwargs
+            )
+        else:
+            # 다중 시드 생성 및 최적 선택
+            candidates = []
+            
+            for i in range(num_candidates):
+                seed = random.randint(1, 10000)
+                
+                try:
+                    result = original_func(
+                        audio_duration, prompt, lyrics, infer_step, guidance_scale,
+                        scheduler_type, cfg_type, omega_scale, str(seed),
+                        guidance_interval, guidance_interval_decay, min_guidance_scale,
+                        use_erg_tag, use_erg_lyric, use_erg_diffusion, oss_steps,
+                        guidance_scale_text, guidance_scale_lyric, audio2audio_enable,
+                        ref_audio_strength, ref_audio_input, lora_name_or_path, **kwargs
+                    )
+                    
+                    if result and len(result) > 0:
+                        audio_path = result[0]  # 첫 번째 결과가 오디오 파일 경로
+                        if audio_path and os.path.exists(audio_path):
+                            quality_score = calculate_quality_score(audio_path)
+                            candidates.append({
+                                "result": result,
+                                "quality_score": quality_score,
+                                "seed": seed
+                            })
+                except Exception as e:
+                    print(f"Generation {i+1} failed: {e}")
+                    continue
+            
+            if candidates:
+                # 최고 품질 선택
+                best_candidate = max(candidates, key=lambda x: x["quality_score"])
+                result = best_candidate["result"]
+                
+                # 품질 정보 추가
+                if len(result) > 1 and isinstance(result[1], dict):
+                    result[1]["quality_score"] = best_candidate["quality_score"]
+                    result[1]["selected_seed"] = best_candidate["seed"]
+                    result[1]["candidates_count"] = len(candidates)
+            else:
+                # 모든 생성 실패시 기본 생성
+                result = original_func(
+                    audio_duration, prompt, lyrics, infer_step, guidance_scale,
+                    scheduler_type, cfg_type, omega_scale, manual_seeds,
+                    guidance_interval, guidance_interval_decay, min_guidance_scale,
+                    use_erg_tag, use_erg_lyric, use_erg_diffusion, oss_steps,
+                    guidance_scale_text, guidance_scale_lyric, audio2audio_enable,
+                    ref_audio_strength, ref_audio_input, lora_name_or_path, **kwargs
+                )
+        
+        # 결과 캐시
+        generation_cache.cache_result(cache_params, result)
+        return result
+    
+    return enhanced_func
+
+def create_output_ui(task_name="Text2Music"):
+    # For many consumer-grade GPU devices, only one batch can be run
+    output_audio1 = gr.Audio(type="filepath", label=f"{task_name} Generated Audio 1")
+    
+    with gr.Accordion(f"{task_name} Parameters & Quality Info", open=False):
+        input_params_json = gr.JSON(label=f"{task_name} Parameters")
+        
+        # 품질 정보 표시 추가
+        with gr.Row():
+            quality_score = gr.Number(label="Quality Score (0-100)", value=0, interactive=False)
+            generation_info = gr.Textbox(
+                label="Generation Info", 
+                value="",
+                interactive=False,
+                max_lines=2
+            )
+    
+    outputs = [output_audio1]
+    return outputs, input_params_json
+
+def dump_func(*args):
+    print(args)
+    return []
+
+def create_text2music_ui(
+    gr,
+    text2music_process_func,
+    sample_data_func=None,
+    load_data_func=None,
+):
+    # 향상된 프로세스 함수 생성
+    enhanced_process_func = create_enhanced_process_func(text2music_process_func)
+
+    with gr.Row():
+        with gr.Column():
+            # 품질 및 성능 설정 섹션 추가
+            with gr.Group():
+                gr.Markdown("### ⚡ 품질 & 성능 설정")
+                with gr.Row():
+                    quality_preset = gr.Dropdown(
+                        choices=list(QUALITY_PRESETS.keys()),
+                        value="Standard",
+                        label="품질 프리셋",
+                        scale=2
+                    )
+                    multi_seed_mode = gr.Dropdown(
+                        choices=list(MULTI_SEED_OPTIONS.keys()),
+                        value="Single",
+                        label="다중 생성 모드",
+                        scale=2,
+                        info="여러 번 생성하여 최고 품질 선택"
+                    )
+                
+                preset_description = gr.Textbox(
+                    value=QUALITY_PRESETS["Standard"]["description"],
+                    label="설명",
+                    interactive=False,
+                    max_lines=1
+                )
+
+            with gr.Row(equal_height=True):
+                audio_duration = gr.Slider(
+                    -1,
+                    240.0,
+                    step=0.00001,
+                    value=-1,
+                    label="Audio Duration",
+                    interactive=True,
+                    info="-1 means random duration (30 ~ 240).",
+                    scale=7,
+                )
+                sample_bnt = gr.Button("Sample", variant="secondary", scale=1)
+                preview_bnt = gr.Button("🎵 Preview", variant="secondary", scale=2)
+
+            # audio2audio
+            with gr.Row(equal_height=True):
+                audio2audio_enable = gr.Checkbox(
+                    label="Enable Audio2Audio", 
+                    value=False, 
+                    info="Check to enable Audio-to-Audio generation using a reference audio.", 
+                    elem_id="audio2audio_checkbox"
+                )
+                lora_name_or_path = gr.Dropdown(
+                    label="Lora Name or Path",
+                    choices=["ACE-Step/ACE-Step-v1-chinese-rap-LoRA", "none"],
+                    value="none",
+                    allow_custom_value=True,
+                )
+
+            ref_audio_input = gr.Audio(
+                type="filepath", 
+                label="Reference Audio (for Audio2Audio)", 
+                visible=False, 
+                elem_id="ref_audio_input", 
+                show_download_button=True
+            )
+            ref_audio_strength = gr.Slider(
+                label="Refer audio strength",
+                minimum=0.0,
+                maximum=1.0,
+                step=0.01,
+                value=0.5,
+                elem_id="ref_audio_strength",
+                visible=False,
+                interactive=True,
+            )
+
+            def toggle_ref_audio_visibility(is_checked):
+                return (
+                    gr.update(visible=is_checked, elem_id="ref_audio_input"),
+                    gr.update(visible=is_checked, elem_id="ref_audio_strength"),
+                )
+
+            audio2audio_enable.change(
+                fn=toggle_ref_audio_visibility,
+                inputs=[audio2audio_enable],
+                outputs=[ref_audio_input, ref_audio_strength],
+            )
+
+            with gr.Column(scale=2):
+                with gr.Group():
+                    gr.Markdown("""### 🎼 스마트 프롬프트 시스템
+                    <center>장르 선택 시 자동으로 최적화된 태그가 추가됩니다. 콤마로 구분하여 태그를 입력하세요.</center>""")
+                    
+                    with gr.Row():
+                        genre_preset = gr.Dropdown(
+                            choices=["Custom"] + list(GENRE_PRESETS.keys()),
+                            value="Custom",
+                            label="장르 프리셋",
+                            scale=1,
+                        )
+                        enable_smart_enhancement = gr.Checkbox(
+                            label="스마트 향상",
+                            value=True,
+                            info="자동 태그 최적화",
+                            scale=1
+                        )
+                    
+                    prompt = gr.Textbox(
+                        lines=2,
+                        label="Tags",
+                        max_lines=4,
+                        value=TAG_DEFAULT,
+                        placeholder="콤마로 구분된 태그들...",
+                    )
+
+            with gr.Group():
+                gr.Markdown("""### 📝 가사 입력
+                <center>구조 태그 [verse], [chorus], [bridge] 사용을 권장합니다.<br>[instrumental] 또는 [inst]를 사용하면 연주곡을 생성합니다.</center>""")
+                
+                # --- 새로운 UI 요소: 주제 입력 후 가사 자동 생성 ---
+                with gr.Row():
+                    topic_for_lyrics = gr.Textbox(
+                        lines=1,
+                        label="가사 주제",
+                        placeholder="예) 첫사랑 이별, 여름 바다, 가을 낙엽 등..."
+                    )
+                    generate_lyrics_btn = gr.Button("가사 생성", variant="secondary")
+
+                # 사용자 직접 입력 가사 박스
+                lyrics = gr.Textbox(
+                    lines=9,
+                    label="Lyrics",
+                    max_lines=13,
+                    value=LYRIC_DEFAULT,
+                    placeholder="가사를 입력하세요. [verse], [chorus] 등의 구조 태그 사용을 권장합니다."
+                )
+
+                # OpenAI를 통해 가사 자동 생성하는 함수
+                def generate_lyrics_ui(topic_text):
+                    # OpenAI 호출
+                    generated = openai_generate_lyrics(topic_text)
+                    return generated
+
+                # 가사 생성 버튼 클릭 시, lyrics 박스에 반영
+                generate_lyrics_btn.click(
+                    fn=generate_lyrics_ui,
+                    inputs=[topic_for_lyrics],
+                    outputs=[lyrics]
+                )
+
+            with gr.Accordion("Basic Settings", open=False):
+                infer_step = gr.Slider(
+                    minimum=1,
+                    maximum=300,
+                    step=1,
+                    value=150,
+                    label="Infer Steps",
+                    interactive=True,
+                )
+                guidance_scale = gr.Slider(
+                    minimum=0.0,
+                    maximum=30.0,
+                    step=0.1,
+                    value=15.0,
+                    label="Guidance Scale",
+                    interactive=True,
+                    info="When guidance_scale_lyric > 1 and guidance_scale_text > 1, the guidance scale will not be applied.",
+                )
+                guidance_scale_text = gr.Slider(
+                    minimum=0.0,
+                    maximum=10.0,
+                    step=0.1,
+                    value=0.0,
+                    label="Guidance Scale Text",
+                    interactive=True,
+                    info="Guidance scale for text condition. It can only apply to cfg. set guidance_scale_text=5.0, guidance_scale_lyric=1.5 for start",
+                )
+                guidance_scale_lyric = gr.Slider(
+                    minimum=0.0,
+                    maximum=10.0,
+                    step=0.1,
+                    value=0.0,
+                    label="Guidance Scale Lyric",
+                    interactive=True,
+                )
+
+                manual_seeds = gr.Textbox(
+                    label="manual seeds (default None)",
+                    placeholder="1,2,3,4",
+                    value=None,
+                    info="Seed for the generation",
+                )
+
+            with gr.Accordion("Advanced Settings", open=False):
+                scheduler_type = gr.Radio(
+                    ["euler", "heun"],
+                    value="euler",
+                    label="Scheduler Type",
+                    elem_id="scheduler_type",
+                    info="Scheduler type for the generation. euler is recommended. heun will take more time.",
+                )
+                cfg_type = gr.Radio(
+                    ["cfg", "apg", "cfg_star"],
+                    value="apg",
+                    label="CFG Type",
+                    elem_id="cfg_type",
+                    info="CFG type for the generation. apg is recommended. cfg and cfg_star are almost the same.",
+                )
+                use_erg_tag = gr.Checkbox(
+                    label="use ERG for tag",
+                    value=True,
+                    info="Use Entropy Rectifying Guidance for tag. It will multiple a temperature to the attention to make a weaker tag condition and make better diversity.",
+                )
+                use_erg_lyric = gr.Checkbox(
+                    label="use ERG for lyric",
+                    value=False,
+                    info="The same but apply to lyric encoder's attention.",
+                )
+                use_erg_diffusion = gr.Checkbox(
+                    label="use ERG for diffusion",
+                    value=True,
+                    info="The same but apply to diffusion model's attention.",
+                )
+
+                omega_scale = gr.Slider(
+                    minimum=-100.0,
+                    maximum=100.0,
+                    step=0.1,
+                    value=10.0,
+                    label="Granularity Scale",
+                    interactive=True,
+                    info="Granularity scale for the generation. Higher values can reduce artifacts",
+                )
+
+                guidance_interval = gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    step=0.01,
+                    value=0.5,
+                    label="Guidance Interval",
+                    interactive=True,
+                    info="Guidance interval for the generation. 0.5 means only apply guidance in the middle steps (0.25 * infer_steps to 0.75 * infer_steps)",
+                )
+                guidance_interval_decay = gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    step=0.01,
+                    value=0.0,
+                    label="Guidance Interval Decay",
+                    interactive=True,
+                    info="Guidance interval decay for the generation. Guidance scale will decay from guidance_scale to min_guidance_scale in the interval. 0.0 means no decay.",
+                )
+                min_guidance_scale = gr.Slider(
+                    minimum=0.0,
+                    maximum=200.0,
+                    step=0.1,
+                    value=3.0,
+                    label="Min Guidance Scale",
+                    interactive=True,
+                    info="Min guidance scale for guidance interval decay's end scale",
+                )
+                oss_steps = gr.Textbox(
+                    label="OSS Steps",
+                    placeholder="16, 29, 52, 96, 129, 158, 172, 183, 189, 200",
+                    value=None,
+                    info="Optimal Steps for the generation. But not test well",
+                )
+
+            text2music_bnt = gr.Button("🎵 Generate Music", variant="primary", size="lg")
+
+            # 모든 UI 요소가 정의된 후 이벤트 핸들러 설정
+            genre_preset.change(
+                fn=update_tags_from_preset,
+                inputs=[genre_preset],
+                outputs=[prompt]
+            )
+            
+            quality_preset.change(
+                fn=lambda x: QUALITY_PRESETS.get(x, {}).get("description", ""),
+                inputs=[quality_preset],
+                outputs=[preset_description]
+            )
+
+            quality_preset.change(
+                fn=update_quality_preset,
+                inputs=[quality_preset],
+                outputs=[infer_step, guidance_scale, scheduler_type, omega_scale, use_erg_diffusion, use_erg_tag]
+            )
+
+        with gr.Column():
+            outputs, input_params_json = create_output_ui()
+            
+            # 실시간 프리뷰 기능
+            def generate_preview(prompt, lyrics, genre_preset):
+                """10초 프리뷰 생성"""
+                preview_params = {
+                    "audio_duration": 10,
+                    "infer_step": 50,
+                    "guidance_scale": 12.0,
+                    "scheduler_type": "euler",
+                    "cfg_type": "apg",
+                    "omega_scale": 5.0,
+                }
+                
+                enhanced_prompt = enhance_prompt_with_genre(prompt, genre_preset) if genre_preset != "Custom" else prompt
+                
+                try:
+                    # 실제 구현에서는 빠른 생성 모드 사용
+                    result = enhanced_process_func(
+                        preview_params["audio_duration"],
+                        enhanced_prompt,
+                        lyrics[:200],  # 가사 일부만 사용
+                        preview_params["infer_step"],
+                        preview_params["guidance_scale"],
+                        preview_params["scheduler_type"],
+                        preview_params["cfg_type"],
+                        preview_params["omega_scale"],
+                        None,  # manual_seeds
+                        0.5,   # guidance_interval
+                        0.0,   # guidance_interval_decay
+                        3.0,   # min_guidance_scale
+                        True,  # use_erg_tag
+                        False, # use_erg_lyric
+                        True,  # use_erg_diffusion
+                        None,  # oss_steps
+                        0.0,   # guidance_scale_text
+                        0.0,   # guidance_scale_lyric
+                        multi_seed_mode="Single"
+                    )
+                    return result[0] if result else None
+                except Exception as e:
+                    return f"프리뷰 생성 실패: {str(e)}"
+
+            preview_bnt.click(
+                fn=generate_preview,
+                inputs=[prompt, lyrics, genre_preset],
+                outputs=[outputs[0]]
+            )
+
+            with gr.Tab("retake"):
+                retake_variance = gr.Slider(
+                    minimum=0.0, maximum=1.0, step=0.01, value=0.2, label="variance"
+                )
+                retake_seeds = gr.Textbox(
+                    label="retake seeds (default None)", placeholder="", value=None
+                )
+                retake_bnt = gr.Button("Retake", variant="primary")
+                retake_outputs, retake_input_params_json = create_output_ui("Retake")
+
+                def retake_process_func(json_data, retake_variance, retake_seeds):
+                    return enhanced_process_func(
+                        json_data.get("audio_duration", 30),
+                        json_data.get("prompt", ""),
+                        json_data.get("lyrics", ""),
+                        json_data.get("infer_step", 100),
+                        json_data.get("guidance_scale", 15.0),
+                        json_data.get("scheduler_type", "euler"),
+                        json_data.get("cfg_type", "apg"),
+                        json_data.get("omega_scale", 10.0),
+                        retake_seeds,
+                        json_data.get("guidance_interval", 0.5),
+                        json_data.get("guidance_interval_decay", 0.0),
+                        json_data.get("min_guidance_scale", 3.0),
+                        json_data.get("use_erg_tag", True),
+                        json_data.get("use_erg_lyric", False),
+                        json_data.get("use_erg_diffusion", True),
+                        json_data.get("oss_steps", None),
+                        json_data.get("guidance_scale_text", 0.0),
+                        json_data.get("guidance_scale_lyric", 0.0),
+                        audio2audio_enable=json_data.get("audio2audio_enable", False),
+                        ref_audio_strength=json_data.get("ref_audio_strength", 0.5),
+                        ref_audio_input=json_data.get("ref_audio_input", None),
+                        lora_name_or_path=json_data.get("lora_name_or_path", "none"),
+                        multi_seed_mode="Best of 3",  # retake는 자동으로 다중 생성
+                        retake_variance=retake_variance,
+                        task="retake"
+                    )
+
+                retake_bnt.click(
+                    fn=retake_process_func,
+                    inputs=[
+                        input_params_json,
+                        retake_variance,
+                        retake_seeds,
+                    ],
+                    outputs=retake_outputs + [retake_input_params_json],
+                )
+
+            with gr.Tab("repainting"):
+                retake_variance = gr.Slider(
+                    minimum=0.0, maximum=1.0, step=0.01, value=0.2, label="variance"
+                )
+                retake_seeds = gr.Textbox(
+                    label="repaint seeds (default None)", placeholder="", value=None
+                )
+                repaint_start = gr.Slider(
+                    minimum=0.0,
+                    maximum=240.0,
+                    step=0.01,
+                    value=0.0,
+                    label="Repaint Start Time",
+                    interactive=True,
+                )
+                repaint_end = gr.Slider(
+                    minimum=0.0,
+                    maximum=240.0,
+                    step=0.01,
+                    value=30.0,
+                    label="Repaint End Time",
+                    interactive=True,
+                )
+                repaint_source = gr.Radio(
+                    ["text2music", "last_repaint", "upload"],
+                    value="text2music",
+                    label="Repaint Source",
+                    elem_id="repaint_source",
+                )
+
+                repaint_source_audio_upload = gr.Audio(
+                    label="Upload Audio",
+                    type="filepath",
+                    visible=False,
+                    elem_id="repaint_source_audio_upload",
+                    show_download_button=True,
+                )
+                repaint_source.change(
+                    fn=lambda x: gr.update(
+                        visible=x == "upload", elem_id="repaint_source_audio_upload"
+                    ),
+                    inputs=[repaint_source],
+                    outputs=[repaint_source_audio_upload],
+                )
+
+                repaint_bnt = gr.Button("Repaint", variant="primary")
+                repaint_outputs, repaint_input_params_json = create_output_ui("Repaint")
+
+                def repaint_process_func(
+                    text2music_json_data,
+                    repaint_json_data,
+                    retake_variance,
+                    retake_seeds,
+                    repaint_start,
+                    repaint_end,
+                    repaint_source,
+                    repaint_source_audio_upload,
+                    prompt,
+                    lyrics,
+                    infer_step,
+                    guidance_scale,
+                    scheduler_type,
+                    cfg_type,
+                    omega_scale,
+                    manual_seeds,
+                    guidance_interval,
+                    guidance_interval_decay,
+                    min_guidance_scale,
+                    use_erg_tag,
+                    use_erg_lyric,
+                    use_erg_diffusion,
+                    oss_steps,
+                    guidance_scale_text,
+                    guidance_scale_lyric,
+                ):
+                    if repaint_source == "upload":
+                        src_audio_path = repaint_source_audio_upload
+                        audio_duration = librosa.get_duration(filename=src_audio_path)
+                        json_data = {"audio_duration": audio_duration}
+                    elif repaint_source == "text2music":
+                        json_data = text2music_json_data
+                        src_audio_path = json_data["audio_path"]
+                    elif repaint_source == "last_repaint":
+                        json_data = repaint_json_data
+                        src_audio_path = json_data["audio_path"]
+
+                    return enhanced_process_func(
+                        json_data["audio_duration"],
+                        prompt,
+                        lyrics,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                        retake_seeds=retake_seeds,
+                        retake_variance=retake_variance,
+                        task="repaint",
+                        repaint_start=repaint_start,
+                        repaint_end=repaint_end,
+                        src_audio_path=src_audio_path,
+                        lora_name_or_path="none"
+                    )
+
+                repaint_bnt.click(
+                    fn=repaint_process_func,
+                    inputs=[
+                        input_params_json,
+                        repaint_input_params_json,
+                        retake_variance,
+                        retake_seeds,
+                        repaint_start,
+                        repaint_end,
+                        repaint_source,
+                        repaint_source_audio_upload,
+                        prompt,
+                        lyrics,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                    ],
+                    outputs=repaint_outputs + [repaint_input_params_json],
+                )
+
+            with gr.Tab("edit"):
+                edit_prompt = gr.Textbox(lines=2, label="Edit Tags", max_lines=4)
+                edit_lyrics = gr.Textbox(lines=9, label="Edit Lyrics", max_lines=13)
+                retake_seeds = gr.Textbox(
+                    label="edit seeds (default None)", placeholder="", value=None
+                )
+
+                edit_type = gr.Radio(
+                    ["only_lyrics", "remix"],
+                    value="only_lyrics",
+                    label="Edit Type",
+                    elem_id="edit_type",
+                    info="`only_lyrics` will keep the whole song the same except lyrics difference. Make your diffrence smaller, e.g. one lyrc line change.\nremix can change the song melody and genre",
+                )
+                edit_n_min = gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    step=0.01,
+                    value=0.6,
+                    label="edit_n_min",
+                    interactive=True,
+                )
+                edit_n_max = gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    step=0.01,
+                    value=1.0,
+                    label="edit_n_max",
+                    interactive=True,
+                )
+
+                def edit_type_change_func(edit_type):
+                    if edit_type == "only_lyrics":
+                        n_min = 0.6
+                        n_max = 1.0
+                    elif edit_type == "remix":
+                        n_min = 0.2
+                        n_max = 0.4
+                    return n_min, n_max
+
+                edit_type.change(
+                    edit_type_change_func,
+                    inputs=[edit_type],
+                    outputs=[edit_n_min, edit_n_max],
+                )
+
+                edit_source = gr.Radio(
+                    ["text2music", "last_edit", "upload"],
+                    value="text2music",
+                    label="Edit Source",
+                    elem_id="edit_source",
+                )
+                edit_source_audio_upload = gr.Audio(
+                    label="Upload Audio",
+                    type="filepath",
+                    visible=False,
+                    elem_id="edit_source_audio_upload",
+                    show_download_button=True,
+                )
+                edit_source.change(
+                    fn=lambda x: gr.update(
+                        visible=x == "upload", elem_id="edit_source_audio_upload"
+                    ),
+                    inputs=[edit_source],
+                    outputs=[edit_source_audio_upload],
+                )
+
+                edit_bnt = gr.Button("Edit", variant="primary")
+                edit_outputs, edit_input_params_json = create_output_ui("Edit")
+
+                def edit_process_func(
+                    text2music_json_data,
+                    edit_input_params_json,
+                    edit_source,
+                    edit_source_audio_upload,
+                    prompt,
+                    lyrics,
+                    edit_prompt,
+                    edit_lyrics,
+                    edit_n_min,
+                    edit_n_max,
+                    infer_step,
+                    guidance_scale,
+                    scheduler_type,
+                    cfg_type,
+                    omega_scale,
+                    manual_seeds,
+                    guidance_interval,
+                    guidance_interval_decay,
+                    min_guidance_scale,
+                    use_erg_tag,
+                    use_erg_lyric,
+                    use_erg_diffusion,
+                    oss_steps,
+                    guidance_scale_text,
+                    guidance_scale_lyric,
+                    retake_seeds,
+                ):
+                    if edit_source == "upload":
+                        src_audio_path = edit_source_audio_upload
+                        audio_duration = librosa.get_duration(filename=src_audio_path)
+                        json_data = {"audio_duration": audio_duration}
+                    elif edit_source == "text2music":
+                        json_data = text2music_json_data
+                        src_audio_path = json_data["audio_path"]
+                    elif edit_source == "last_edit":
+                        json_data = edit_input_params_json
+                        src_audio_path = json_data["audio_path"]
+
+                    if not edit_prompt:
+                        edit_prompt = prompt
+                    if not edit_lyrics:
+                        edit_lyrics = lyrics
+
+                    return enhanced_process_func(
+                        json_data["audio_duration"],
+                        prompt,
+                        lyrics,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                        task="edit",
+                        src_audio_path=src_audio_path,
+                        edit_target_prompt=edit_prompt,
+                        edit_target_lyrics=edit_lyrics,
+                        edit_n_min=edit_n_min,
+                        edit_n_max=edit_n_max,
+                        retake_seeds=retake_seeds,
+                        lora_name_or_path="none"
+                    )
+
+                edit_bnt.click(
+                    fn=edit_process_func,
+                    inputs=[
+                        input_params_json,
+                        edit_input_params_json,
+                        edit_source,
+                        edit_source_audio_upload,
+                        prompt,
+                        lyrics,
+                        edit_prompt,
+                        edit_lyrics,
+                        edit_n_min,
+                        edit_n_max,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                        retake_seeds,
+                    ],
+                    outputs=edit_outputs + [edit_input_params_json],
+                )
+
+            with gr.Tab("extend"):
+                extend_seeds = gr.Textbox(
+                    label="extend seeds (default None)", placeholder="", value=None
+                )
+                left_extend_length = gr.Slider(
+                    minimum=0.0,
+                    maximum=240.0,
+                    step=0.01,
+                    value=0.0,
+                    label="Left Extend Length",
+                    interactive=True,
+                )
+                right_extend_length = gr.Slider(
+                    minimum=0.0,
+                    maximum=240.0,
+                    step=0.01,
+                    value=30.0,
+                    label="Right Extend Length",
+                    interactive=True,
+                )
+                extend_source = gr.Radio(
+                    ["text2music", "last_extend", "upload"],
+                    value="text2music",
+                    label="Extend Source",
+                    elem_id="extend_source",
+                )
+
+                extend_source_audio_upload = gr.Audio(
+                    label="Upload Audio",
+                    type="filepath",
+                    visible=False,
+                    elem_id="extend_source_audio_upload",
+                    show_download_button=True,
+                )
+                extend_source.change(
+                    fn=lambda x: gr.update(
+                        visible=x == "upload", elem_id="extend_source_audio_upload"
+                    ),
+                    inputs=[extend_source],
+                    outputs=[extend_source_audio_upload],
+                )
+
+                extend_bnt = gr.Button("Extend", variant="primary")
+                extend_outputs, extend_input_params_json = create_output_ui("Extend")
+
+                def extend_process_func(
+                    text2music_json_data,
+                    extend_input_params_json,
+                    extend_seeds,
+                    left_extend_length,
+                    right_extend_length,
+                    extend_source,
+                    extend_source_audio_upload,
+                    prompt,
+                    lyrics,
+                    infer_step,
+                    guidance_scale,
+                    scheduler_type,
+                    cfg_type,
+                    omega_scale,
+                    manual_seeds,
+                    guidance_interval,
+                    guidance_interval_decay,
+                    min_guidance_scale,
+                    use_erg_tag,
+                    use_erg_lyric,
+                    use_erg_diffusion,
+                    oss_steps,
+                    guidance_scale_text,
+                    guidance_scale_lyric,
+                ):
+                    if extend_source == "upload":
+                        src_audio_path = extend_source_audio_upload
+                        # get audio duration
+                        audio_duration = librosa.get_duration(filename=src_audio_path)
+                        json_data = {"audio_duration": audio_duration}
+                    elif extend_source == "text2music":
+                        json_data = text2music_json_data
+                        src_audio_path = json_data["audio_path"]
+                    elif extend_source == "last_extend":
+                        json_data = extend_input_params_json
+                        src_audio_path = json_data["audio_path"]
+
+                    repaint_start = -left_extend_length
+                    repaint_end = json_data["audio_duration"] + right_extend_length
+                    return enhanced_process_func(
+                        json_data["audio_duration"],
+                        prompt,
+                        lyrics,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                        retake_seeds=extend_seeds,
+                        retake_variance=1.0,
+                        task="extend",
+                        repaint_start=repaint_start,
+                        repaint_end=repaint_end,
+                        src_audio_path=src_audio_path,
+                        lora_name_or_path="none"
+                    )
+
+                extend_bnt.click(
+                    fn=extend_process_func,
+                    inputs=[
+                        input_params_json,
+                        extend_input_params_json,
+                        extend_seeds,
+                        left_extend_length,
+                        right_extend_length,
+                        extend_source,
+                        extend_source_audio_upload,
+                        prompt,
+                        lyrics,
+                        infer_step,
+                        guidance_scale,
+                        scheduler_type,
+                        cfg_type,
+                        omega_scale,
+                        manual_seeds,
+                        guidance_interval,
+                        guidance_interval_decay,
+                        min_guidance_scale,
+                        use_erg_tag,
+                        use_erg_lyric,
+                        use_erg_diffusion,
+                        oss_steps,
+                        guidance_scale_text,
+                        guidance_scale_lyric,
+                    ],
+                    outputs=extend_outputs + [extend_input_params_json],
+                )
+
+        def json2output(json_data):
+            return (
+                json_data["audio_duration"],
+                json_data["prompt"],
+                json_data["lyrics"],
+                json_data["infer_step"],
+                json_data["guidance_scale"],
+                json_data["scheduler_type"],
+                json_data["cfg_type"],
+                json_data["omega_scale"],
+                ", ".join(map(str, json_data["actual_seeds"])),
+                json_data["guidance_interval"],
+                json_data["guidance_interval_decay"],
+                json_data["min_guidance_scale"],
+                json_data["use_erg_tag"],
+                json_data["use_erg_lyric"],
+                json_data["use_erg_diffusion"],
+                ", ".join(map(str, json_data["oss_steps"])),
+                (
+                    json_data["guidance_scale_text"]
+                    if "guidance_scale_text" in json_data
+                    else 0.0
+                ),
+                (
+                    json_data["guidance_scale_lyric"]
+                    if "guidance_scale_lyric" in json_data
+                    else 0.0
+                ),
+                (
+                    json_data["audio2audio_enable"]
+                    if "audio2audio_enable" in json_data
+                    else False
+                ),
+                (
+                    json_data["ref_audio_strength"]
+                    if "ref_audio_strength" in json_data
+                    else 0.5
+                ),
+                (
+                    json_data["ref_audio_input"]
+                    if "ref_audio_input" in json_data
+                    else None
+                ),
+            )
+
+        def sample_data(lora_name_or_path_):
+            if sample_data_func:
+                json_data = sample_data_func(lora_name_or_path_)
+                return json2output(json_data)
+            return {}
+
+        sample_bnt.click(
+            sample_data,
+            inputs=[lora_name_or_path],
+            outputs=[
+                audio_duration,
+                prompt,
+                lyrics,
+                infer_step,
+                guidance_scale,
+                scheduler_type,
+                cfg_type,
+                omega_scale,
+                manual_seeds,
+                guidance_interval,
+                guidance_interval_decay,
+                min_guidance_scale,
+                use_erg_tag,
+                use_erg_lyric,
+                use_erg_diffusion,
+                oss_steps,
+                guidance_scale_text,
+                guidance_scale_lyric,
+                audio2audio_enable,
+                ref_audio_strength,
+                ref_audio_input,
+            ],
+        )
+
+    # 메인 생성 버튼 이벤트 (향상된 함수 사용)
+    text2music_bnt.click(
+        fn=enhanced_process_func,
+        inputs=[
+            audio_duration,
+            prompt,
+            lyrics,
+            infer_step,
+            guidance_scale,
+            scheduler_type,
+            cfg_type,
+            omega_scale,
+            manual_seeds,
+            guidance_interval,
+            guidance_interval_decay,
+            min_guidance_scale,
+            use_erg_tag,
+            use_erg_lyric,
+            use_erg_diffusion,
+            oss_steps,
+            guidance_scale_text,
+            guidance_scale_lyric,
+            audio2audio_enable,
+            ref_audio_strength,
+            ref_audio_input,
+            lora_name_or_path,
+            multi_seed_mode,
+            enable_smart_enhancement,
+            genre_preset
+        ],
+        outputs=outputs + [input_params_json],
+    )
+
+
+def create_main_demo_ui(
+    text2music_process_func=dump_func,
+    sample_data_func=dump_func,
+    load_data_func=dump_func,
+):
+    with gr.Blocks(
+        title="ACE-Step Model 1.0 DEMO - Enhanced",
+        theme=gr.themes.Soft(),
+        css="""
+        .gradio-container {
+            max-width: 1200px !important;
+        }
+        .quality-info {
+            background: linear-gradient(45deg, #f0f8ff, #e6f3ff);
+            padding: 10px;
+            border-radius: 8px;
+            margin: 5px 0;
+        }
+        """
+    ) as demo:
+        gr.Markdown(
+            """
+            <h1 style="text-align: center;">🎵 ACE-Step PRO</h1>
+            <div style="text-align: center; margin: 20px;">
+                <p><strong>🚀 새로운 기능:</strong> 품질 프리셋 | 다중 생성 | 스마트 프롬프트 | 실시간 프리뷰 | 품질 점수</p>
+                <p>
+                    <a href="https://ace-step.github.io/" target='_blank'>Project</a> |
+                    <a href="https://huggingface.co/ACE-Step/ACE-Step-v1-3.5B">Checkpoints</a> |
+                    <a href="https://discord.gg/rjAZz2xBdG" target='_blank'>Discord</a> 
+                </p>
+            </div>
+        """
+        )
+        
+        # 사용법 가이드 추가
+        with gr.Accordion("📖 사용법 가이드", open=False):
+            gr.Markdown("""
+            ### 🎯 빠른 시작
+            1. **장르 선택**: 원하는 음악 장르를 선택하면 자동으로 최적화된 태그가 적용됩니다
+            2. **품질 설정**: Draft(빠름) → Standard(권장) → High Quality → Ultra 중 선택
+            3. **다중 생성**: "Best of 3/5/10" 선택하면 여러 번 생성하여 최고 품질을 자동 선택합니다
+            4. **프리뷰**: 전체 생성 전 10초 프리뷰로 빠르게 확인할 수 있습니다
+            
+            ### 💡 품질 향상 팁
+            - **고품질 생성**: "High Quality" + "Best of 5" 조합 추천
+            - **빠른 테스트**: "Draft" + "프리뷰" 기능 활용
+            - **장르 특화**: 장르 프리셋 선택 후 "스마트 향상" 체크
+            - **가사 구조**: [verse], [chorus], [bridge] 태그 적극 활용
+            """)
+        
+        with gr.Tab("🎵 Enhanced Text2Music"):
+            create_text2music_ui(
+                gr=gr,
+                text2music_process_func=text2music_process_func,
+                sample_data_func=sample_data_func,
+                load_data_func=load_data_func,
+            )
+    return demo
+
+
+if __name__ == "__main__":
+    demo = create_main_demo_ui()
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=7860,
+        share=True  # 공유 링크 생성
+    )