Update app.py

app.py

@@ -6,9 +6,6 @@ import subprocess
 import numpy as np
 import gradio as gr
 import soundfile as sf
-import random
-import torch
-from src.chatterbox.mtl_tts import ChatterboxMultilingualTTS, SUPPORTED_LANGUAGES
 
 # Clone NeuTTS-Air repository if not present
 NEUTTS_DIR = "neutts-air"
@@ -25,7 +22,6 @@ sys.path.append(NEUTTS_DIR)
 # Global variables for lazy loading
 kokoro_pipe = None
 neutts_model = None
-chatterbox_model = None
 
 # NeuTTS-Air configuration - aligned with official neutts-air/app.py
 SAMPLES_PATH = os.path.join(os.getcwd(), NEUTTS_DIR, "samples")
@@ -33,104 +29,6 @@ DEFAULT_REF_TEXT = "So I'm live on radio. And I say, well, my dear friend James
 DEFAULT_REF_PATH = os.path.join(SAMPLES_PATH, "dave.wav")
 DEFAULT_GEN_TEXT = "My name is Dave, and um, I'm from London."
 
-DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
-print(f"🚀 Running on device: {DEVICE}")
-
-LANGUAGE_CONFIG = {
-    "ar": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/ar_f/ar_prompts2.flac",
-        "text": "في الشهر الماضي، وصلنا إلى معلم جديد بمليارين من المشاهدات على قناتنا على يوتيوب."
-    },
-    "da": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/da_m1.flac",
-        "text": "Sidste måned nåede vi en ny milepæl med to milliarder visninger på vores YouTube-kanal."
-    },
-    "de": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/de_f1.flac",
-        "text": "Letzten Monat haben wir einen neuen Meilenstein erreicht: zwei Milliarden Aufrufe auf unserem YouTube-Kanal."
-    },
-    "el": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/el_m.flac",
-        "text": "Τον περασμένο μήνα, φτάσαμε σε ένα νέο ορόσημο με δύο δισεκατομμύρια προβολές στο κανάλι μας στο YouTube."
-    },
-    "en": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/en_f1.flac",
-        "text": "Last month, we reached a new milestone with two billion views on our YouTube channel."
-    },
-    "es": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/es_f1.flac",
-        "text": "El mes pasado alcanzamos un nuevo hito: dos mil millones de visualizaciones en nuestro canal de YouTube."
-    },
-    "fi": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/fi_m.flac",
-        "text": "Viime kuussa saavutimme uuden virstanpylvään kahden miljardin katselukerran kanssa YouTube-kanavallamme."
-    },
-    "fr": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/fr_f1.flac",
-        "text": "Le mois dernier, nous avons atteint un nouveau jalon avec deux milliards de vues sur notre chaîne YouTube."
-    },
-    "he": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/he_m1.flac",
-        "text": "בחודש שעבר הגענו לאבן דרך חדשה עם שני מיליארד צפיות בערוץ היוטיוב שלנו."
-    },
-    "hi": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/hi_f1.flac",
-        "text": "पिछले महीने हमने एक नया मील का पत्थर छुआ: हमारे YouTube चैनल पर दो अरब व्यूज़।"
-    },
-    "it": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/it_m1.flac",
-        "text": "Il mese scorso abbiamo raggiunto un nuovo traguardo: due miliardi di visualizzazioni sul nostro canale YouTube."
-    },
-    "ja": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/ja/ja_prompts1.flac",
-        "text": "先月、私たちのYouTubeチャンネルで二十億回の再生回数という新たなマイルストーンに到達しました。"
-    },
-    "ko": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/ko_f.flac",
-        "text": "지난달 우리는 유튜브 채널에서 이십억 조회수라는 새로운 이정표에 도달했습니다."
-    },
-    "ms": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/ms_f.flac",
-        "text": "Bulan lepas, kami mencapai pencapaian baru dengan dua bilion tontonan di saluran YouTube kami."
-    },
-    "nl": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/nl_m.flac",
-        "text": "Vorige maand bereikten we een nieuwe mijlpaal met twee miljard weergaven op ons YouTube-kanaal."
-    },
-    "no": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/no_f1.flac",
-        "text": "Forrige måned nådde vi en ny milepæl med to milliarder visninger på YouTube-kanalen vår."
-    },
-    "pl": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/pl_m.flac",
-        "text": "W zeszłym miesiącu osiągnęliśmy nowy kamień milowy z dwoma miliardami wyświetleń na naszym kanale YouTube."
-    },
-    "pt": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/pt_m1.flac",
-        "text": "No mês passado, alcançámos um novo marco: dois mil milhões de visualizações no nosso canal do YouTube."
-    },
-    "ru": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/ru_m.flac",
-        "text": "В прошлом месяце мы достигли нового рубежа: два миллиарда просмотров на нашем YouTube-канале."
-    },
-    "sv": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/sv_f.flac",
-        "text": "Förra månaden nådde vi en ny milstolpe med två miljarder visningar på vår YouTube-kanal."
-    },
-    "sw": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/sw_m.flac",
-        "text": "Mwezi uliopita, tulifika hatua mpya ya maoni ya bilioni mbili kweny kituo chetu cha YouTube."
-    },
-    "tr": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/tr_m.flac",
-        "text": "Geçen ay YouTube kanalımızda iki milyar görüntüleme ile yeni bir dönüm noktasına ulaştık."
-    },
-    "zh": {
-        "audio": "https://storage.googleapis.com/chatterbox-demo-samples/mtl_prompts/zh_f2.flac",
-        "text": "上个月，我们达到了一个新的里程碑。 我们的YouTube频道观看次数达到了二十亿次，这绝对令人难以置信。"
-    },
-}
-
 # ------------------------------------------------------------------
 # 1. Lazy loaders
 # ------------------------------------------------------------------
@@ -155,28 +53,6 @@ def load_neutts():
         )
     return neutts_model
 
-def get_or_load_chatterbox_model():
-    """Loads the ChatterboxMultilingualTTS model if it hasn't been loaded already,
-    and ensures it's on the correct device."""
-    global chatterbox_model
-    if chatterbox_model is None:
-        print("Chatterbox model not loaded, initializing...")
-        try:
-            chatterbox_model = ChatterboxMultilingualTTS.from_pretrained(DEVICE)
-            if hasattr(chatterbox_model, 'to') and str(chatterbox_model.device) != DEVICE:
-                chatterbox_model.to(DEVICE)
-            print(f"Chatterbox model loaded successfully. Internal device: {getattr(chatterbox_model, 'device', 'N/A')}")
-        except Exception as e:
-            print(f"Error loading Chatterbox model: {e}")
-            raise
-    return chatterbox_model
-
-# Attempt to load the Chatterbox model at startup.
-try:
-    get_or_load_chatterbox_model()
-except Exception as e:
-    print(f"CRITICAL: Failed to load Chatterbox model on startup. Application may not function. Error: {e}")
-
 # ------------------------------------------------------------------
 # 2. Kokoro TTS inference
 # ------------------------------------------------------------------
@@ -233,100 +109,17 @@ def neutts_infer(ref_text: str, ref_audio_path: str, gen_text: str) -> tuple[int
     return (24_000, wav)
 
 # ------------------------------------------------------------------
-# 4. Chatterbox Multilingual inference
-# ------------------------------------------------------------------
-def default_audio_for_ui(lang: str) -> str | None:
-    return LANGUAGE_CONFIG.get(lang, {}).get("audio")
-
-def default_text_for_ui(lang: str) -> str:
-    return LANGUAGE_CONFIG.get(lang, {}).get("text", "")
-
-def get_supported_languages_display() -> str:
-    """Generate a formatted display of all supported languages."""
-    language_items = []
-    for code, name in sorted(SUPPORTED_LANGUAGES.items()):
-        language_items.append(f"**{name}** (`{code}`)")
-    
-    # Split into 2 lines
-    mid = len(language_items) // 2
-    line1 = " • ".join(language_items[:mid])
-    line2 = " • ".join(language_items[mid:])
-    
-    return f"""
-### 🌍 Supported Languages ({len(SUPPORTED_LANGUAGES)} total)
-{line1}
-{line2}
-"""
-
-def set_seed(seed: int):
-    """Sets the random seed for reproducibility across torch, numpy, and random."""
-    torch.manual_seed(seed)
-    if DEVICE == "cuda":
-        torch.cuda.manual_seed(seed)
-        torch.cuda.manual_seed_all(seed)
-    random.seed(seed)
-    np.random.seed(seed)
-
-@spaces.GPU
-def chatterbox_infer(
-    text_input: str,
-    language_id: str,
-    audio_prompt_path_input: str = None,
-    exaggeration_input: float = 0.5,
-    temperature_input: float = 0.8,
-    seed_num_input: int = 0,
-    cfgw_input: float = 0.5
-) -> tuple[int, np.ndarray]:
-    """
-    Generate high-quality speech audio from text using Chatterbox Multilingual model with optional reference audio styling.
-    """
-    current_model = get_or_load_chatterbox_model()
-
-    if current_model is None:
-        raise RuntimeError("TTS model is not loaded.")
-
-    if seed_num_input != 0:
-        set_seed(int(seed_num_input))
-
-    print(f"Generating audio for text: '{text_input[:50]}...'")
-    
-    # Handle optional audio prompt
-    chosen_prompt = audio_prompt_path_input or default_audio_for_ui(language_id)
-
-    generate_kwargs = {
-        "exaggeration": exaggeration_input,
-        "temperature": temperature_input,
-        "cfg_weight": cfgw_input,
-    }
-    if chosen_prompt:
-        generate_kwargs["audio_prompt_path"] = chosen_prompt
-        print(f"Using audio prompt: {chosen_prompt}")
-    else:
-        print("No audio prompt provided; using default voice.")
-        
-    wav = current_model.generate(
-        text_input[:300],  # Truncate text to max chars
-        language_id=language_id,
-        **generate_kwargs
-    )
-    print("Audio generation complete.")
-    return (current_model.sr, wav.squeeze(0).numpy())
-
-def on_language_change(lang, current_ref, current_text):
-    return default_audio_for_ui(lang), default_text_for_ui(lang)
-
-# ------------------------------------------------------------------
-# 5. Gradio UI with model selection
+# 4. Gradio UI with model selection
 # ------------------------------------------------------------------
 css = """footer {visibility: hidden}"""
 
-with gr.Blocks(css=css, title="Text2Audio - Kokoro, NeuTTS-Air & Chatterbox") as demo:
+with gr.Blocks(css=css, title="Text2Audio - Kokoro & NeuTTS-Air") as demo:
     gr.Markdown("# 🎙️ Text-to-Audio Generation")
-    gr.Markdown("Choose between **Kokoro TTS** (fast English TTS), **NeuTTS-Air** (voice cloning with reference audio), or **Chatterbox Multilingual** (23 languages with styling)")
+    gr.Markdown("Choose between **Kokoro TTS** (fast English TTS) or **NeuTTS-Air** (voice cloning with reference audio)")
     
     # Model selection
     model_choice = gr.Radio(
-        choices=["Kokoro TTS", "NeuTTS-Air", "Chatterbox Multilingual"],
+        choices=["Kokoro TTS", "NeuTTS-Air"],
         value="Kokoro TTS",
         label="Select TTS Engine",
         interactive=True
@@ -380,96 +173,17 @@ with gr.Blocks(css=css, title="Text2Audio - Kokoro, NeuTTS-Air & Chatterbox") as
         
         gr.Markdown("**NeuTTS-Air** – Upload a reference audio sample, provide the reference text, and enter new text to synthesize.")
     
-    # Chatterbox Multilingual Interface
-    with gr.Group(visible=False) as chatterbox_group:
-        gr.Markdown("### 🗣️ Chatterbox Multilingual Settings")
-        gr.Markdown(
-            """
-            Generate high-quality multilingual speech from text with reference audio styling, supporting 23 languages.
-            
-            For a hosted version of Chatterbox Multilingual and for finetuning, please visit [resemble.ai](https://app.resemble.ai)
-            """
-        )
-        # Display supported languages
-        gr.Markdown(get_supported_languages_display())
-        with gr.Row():
-            with gr.Column():
-                initial_lang = "en"
-                chatterbox_text = gr.Textbox(
-                    value=default_text_for_ui(initial_lang),
-                    label="Text to synthesize (max chars 300)",
-                    max_lines=5
-                )
-                
-                chatterbox_language_id = gr.Dropdown(
-                    choices=list(SUPPORTED_LANGUAGES.keys()),
-                    value=initial_lang,
-                    label="Language",
-                    info="Select the language for text-to-speech synthesis"
-                )
-                
-                chatterbox_ref_wav = gr.Audio(
-                    sources=["upload", "microphone"],
-                    type="filepath",
-                    label="Reference Audio File (Optional)",
-                    value=default_audio_for_ui(initial_lang)
-                )
-                
-                gr.Markdown(
-                    "💡 **Note**: Ensure that the reference clip matches the specified language tag. Otherwise, language transfer outputs may inherit the accent of the reference clip's language. To mitigate this, set the CFG weight to 0.",
-                    elem_classes=["audio-note"]
-                )
-                
-                chatterbox_exaggeration = gr.Slider(
-                    0.25, 2, step=.05, label="Exaggeration (Neutral = 0.5, extreme values can be unstable)", value=.5
-                )
-                chatterbox_cfg_weight = gr.Slider(
-                    0.2, 1, step=.05, label="CFG/Pace", value=0.5
-                )
-
-                with gr.Accordion("More options", open=False):
-                    chatterbox_seed_num = gr.Number(value=0, label="Random seed (0 for random)")
-                    chatterbox_temp = gr.Slider(0.05, 5, step=.05, label="Temperature", value=.8)
-
-                chatterbox_btn = gr.Button("🗣️ Generate with Chatterbox", variant="primary")
-
-            with gr.Column():
-                chatterbox_audio_output = gr.Audio(label="Output Audio")
-
-        chatterbox_language_id.change(
-            fn=on_language_change,
-            inputs=[chatterbox_language_id, chatterbox_ref_wav, chatterbox_text],
-            outputs=[chatterbox_ref_wav, chatterbox_text],
-            show_progress=False
-        )
-
-        chatterbox_btn.click(
-            fn=chatterbox_infer,
-            inputs=[
-                chatterbox_text,
-                chatterbox_language_id,
-                chatterbox_ref_wav,
-                chatterbox_exaggeration,
-                chatterbox_temp,
-                chatterbox_seed_num,
-                chatterbox_cfg_weight,
-            ],
-            outputs=[chatterbox_audio_output],
-        )
-    
     # Event handlers
     def toggle_interface(choice):
         if choice == "Kokoro TTS":
-            return gr.update(visible=True), gr.update(visible=False), gr.update(visible=False)
-        elif choice == "NeuTTS-Air":
-            return gr.update(visible=False), gr.update(visible=True), gr.update(visible=False)
+            return gr.update(visible=True), gr.update(visible=False)
         else:
-            return gr.update(visible=False), gr.update(visible=False), gr.update(visible=True)
+            return gr.update(visible=False), gr.update(visible=True)
     
     model_choice.change(
         fn=toggle_interface,
         inputs=[model_choice],
-        outputs=[kokoro_group, neutts_group, chatterbox_group]
+        outputs=[kokoro_group, neutts_group]
     )
     
     kokoro_btn.click(