Update inference_webui.py

inference_webui.py

@@ -1,673 +1,238 @@
-'''
-按中英混合识别
-按日英混合识别
-多语种启动切分识别语种
-全部按中文识别
-全部按英文识别
-全部按日文识别
-'''
-import logging
-import traceback
-
-logging.getLogger("markdown_it").setLevel(logging.ERROR)
-logging.getLogger("urllib3").setLevel(logging.ERROR)
-logging.getLogger("httpcore").setLevel(logging.ERROR)
-logging.getLogger("httpx").setLevel(logging.ERROR)
-logging.getLogger("asyncio").setLevel(logging.ERROR)
-logging.getLogger("charset_normalizer").setLevel(logging.ERROR)
-logging.getLogger("torchaudio._extension").setLevel(logging.ERROR)
-logging.getLogger("multipart.multipart").setLevel(logging.ERROR)
-import gradio.analytics as analytics
-analytics.version_check = lambda:None
-analytics.get_local_ip_address= lambda :"127.0.0.1"##不干掉本地联不通亚马逊的get_local_ip服务器
-import nltk
-nltk.download('averaged_perceptron_tagger_eng')
-import LangSegment, os, re, sys, json
-import pdb
-import spaces
-import torch
-
-version="v2"#os.environ.get("version","v2")
-cnhubert_base_path = os.environ.get(
-    "cnhubert_base_path", "pretrained_models/chinese-hubert-base"
-)
-bert_path = os.environ.get(
-    "bert_path", "pretrained_models/chinese-roberta-wwm-ext-large"
-)
-
-punctuation = set(['!', '?', '…', ',', '.', '-'," "])
+import random
+import os
+import uuid
+from datetime import datetime
 import gradio as gr
-from transformers import AutoModelForMaskedLM, AutoTokenizer
 import numpy as np
-import librosa
-from feature_extractor import cnhubert
-
-cnhubert.cnhubert_base_path = cnhubert_base_path
-
-from module.models import SynthesizerTrn
-from AR.models.t2s_lightning_module import Text2SemanticLightningModule
-from text import cleaned_text_to_sequence
-from text.cleaner import clean_text
-from time import time as ttime
-from module.mel_processing import spectrogram_torch
-from tools.my_utils import load_audio
-from tools.i18n.i18n import I18nAuto, scan_language_list
-
-# language=os.environ.get("language","Auto")
-# language=sys.argv[-1] if sys.argv[-1] in scan_language_list() else language
-i18n = I18nAuto(language="Auto")
-
-# os.environ['PYTORCH_ENABLE_MPS_FALLBACK'] = '1'  # 确保直接启动推理UI时也能够设置。
-
-if torch.cuda.is_available():
-    device = "cuda"
-    is_half = True  # eval(os.environ.get("is_half", "True")) and torch.cuda.is_available()
-else:
-    device = "cpu"
-    is_half=False
-
-dict_language_v1 = {
-    i18n("中文"): "all_zh",#全部按中文识别
-    i18n("英文"): "en",#全部按英文识别#######不变
-    i18n("日文"): "all_ja",#全部按日文识别
-    i18n("中英混合"): "zh",#按中英混合识别####不变
-    i18n("日英混合"): "ja",#按日英混合识别####不变
-    i18n("多语种混合"): "auto",#多语种启动切分识别语种
-}
-dict_language_v2 = {
-    i18n("中文"): "all_zh",#全部按中文识别
-    i18n("英文"): "en",#全部按英文识别#######不变
-    i18n("日文"): "all_ja",#全部按日文识别
-    i18n("粤语"): "all_yue",#全部按中文识别
-    i18n("韩文"): "all_ko",#全部按韩文识别
-    i18n("中英混合"): "zh",#按中英混合识别####不变
-    i18n("日英混合"): "ja",#按日英混合识别####不变
-    i18n("粤英混合"): "yue",#按粤英混合识别####不变
-    i18n("韩英混合"): "ko",#按韩英混合识别####不变
-    i18n("多语种混合"): "auto",#多语种启动切分识别语种
-    i18n("多语种混合(粤语)"): "auto_yue",#多语种启动切分识别语种
+import spaces
+import torch
+from diffusers import DiffusionPipeline
+from PIL import Image
+
+# Create permanent storage directory
+SAVE_DIR = "saved_images"  # Gradio will handle the persistence
+if not os.path.exists(SAVE_DIR):
+    os.makedirs(SAVE_DIR, exist_ok=True)
+
+# Load the default image
+DEFAULT_IMAGE_PATH = "cover1.webp"
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+repo_id = "black-forest-labs/FLUX.1-dev"
+adapter_id = "strangerzonehf/Ctoon-Plus-Plus"
+
+pipeline = DiffusionPipeline.from_pretrained(repo_id, torch_dtype=torch.bfloat16)
+pipeline.load_lora_weights(adapter_id)
+pipeline = pipeline.to(device)
+
+MAX_SEED = np.iinfo(np.int32).max
+MAX_IMAGE_SIZE = 1024
+
+def save_generated_image(image, prompt):
+    # Generate unique filename with timestamp
+    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
+    unique_id = str(uuid.uuid4())[:8]
+    filename = f"{timestamp}_{unique_id}.png"
+    filepath = os.path.join(SAVE_DIR, filename)
+    
+    # Save the image
+    image.save(filepath)
+    
+    # Save metadata
+    metadata_file = os.path.join(SAVE_DIR, "metadata.txt")
+    with open(metadata_file, "a", encoding="utf-8") as f:
+        f.write(f"{filename}|{prompt}|{timestamp}\n")
+    
+    return filepath
+
+def load_generated_images():
+    if not os.path.exists(SAVE_DIR):
+        return []
+    
+    # Load all images from the directory
+    image_files = [os.path.join(SAVE_DIR, f) for f in os.listdir(SAVE_DIR) 
+                  if f.endswith(('.png', '.jpg', '.jpeg', '.webp'))]
+    # Sort by creation time (newest first)
+    image_files.sort(key=lambda x: os.path.getctime(x), reverse=True)
+    return image_files
+
+def load_predefined_images():
+    # Return empty list since we're not using predefined images
+    return []
+
+@spaces.GPU(duration=120)
+def inference(
+    prompt: str,
+    seed: int,
+    randomize_seed: bool,
+    width: int,
+    height: int,
+    guidance_scale: float,
+    num_inference_steps: int,
+    lora_scale: float,
+    progress: gr.Progress = gr.Progress(track_tqdm=True),
+):
+    if randomize_seed:
+        seed = random.randint(0, MAX_SEED)
+    generator = torch.Generator(device=device).manual_seed(seed)
+    
+    image = pipeline(
+        prompt=prompt,
+        guidance_scale=guidance_scale,
+        num_inference_steps=num_inference_steps,
+        width=width,
+        height=height,
+        generator=generator,
+        joint_attention_kwargs={"scale": lora_scale},
+    ).images[0]
+    
+    # Save the generated image
+    filepath = save_generated_image(image, prompt)
+    
+    # Return the image, seed, and updated gallery
+    return image, seed, load_generated_images()
+
+
+examples = [
+   "A cartoon drawing of a majestic Persian cat wearing a tiny golden hanbok and crown. The cat has sparkling blue eyes and perfectly groomed white fur that seems to glow. It sits with regal posture on a traditional Korean cushion decorated with cloud patterns. The background is a soft pink with delicate cherry blossom petals floating around. The cat's expression shows a mix of dignity and subtle amusement. [trigger]",
+   
+   "A cartoon drawing of an enthusiastic orange tabby cat in a puffy white chef's hat. The cat stands on its hind legs at a tiny wooden counter, wearing a white apron covered in flour pawprints. Its green eyes are focused intently on the cookie dough it's rolling with a miniature rolling pin. The background is a warm cream color with tiny floating cooking utensils and swirling steam patterns. [trigger]",
+   
+   "A cartoon drawing of a sophisticated tuxedo cat photographer with round wire-rimmed glasses perched on its nose. The cat balances carefully on a tree branch, one paw holding a vintage camera while its tail curls in concentration. It wears a tiny brown beret and leather camera bag. The background is a soft blue with playful butterfly silhouettes and floating leaves. [trigger]",
+   
+   "A cartoon drawing of a chubby Scottish Fold cat floating in a space capsule. The cat wears an adorable white spacesuit with colorful patches, its round face visible through the helmet visor. Its paws are batting at star-shaped toys that float around in zero gravity. The background shows a stylized view of Earth and twinkling stars through the capsule window. [trigger]",
+   
+   "A cartoon drawing of an elegant Siamese ballet dancer cat in mid-twirl. The cat wears a sparkly pink tutu that flares out perfectly, with tiny satin ribbons wrapped around its ankles. Its blue eyes are closed in graceful concentration as it performs a pirouette. The background is a soft lavender with swirling musical notes and floating rose petals. [trigger]",
+   
+   "A cartoon drawing of an adventurous calico cat riding atop a smiling elephant. The cat wears a tiny khaki explorer's vest filled with equipment, and a safari hat tilted at a jaunty angle. It holds a comically large map while the elephant's trunk curls up playfully. The background is a warm orange sunset with stylized acacia trees and cartoon birds soaring past. [trigger]"
+]
+css = """
+footer {
+    visibility: hidden;
 }
-dict_language = dict_language_v1 if version =='v1' else dict_language_v2
-
-tokenizer = AutoTokenizer.from_pretrained(bert_path)
-bert_model = AutoModelForMaskedLM.from_pretrained(bert_path)
-if is_half == True:
-    bert_model = bert_model.half().to(device)
-else:
-    bert_model = bert_model.to(device)
-
-
-def get_bert_feature(text, word2ph):
-    with torch.no_grad():
-        inputs = tokenizer(text, return_tensors="pt")
-        for i in inputs:
-            inputs[i] = inputs[i].to(device)
-        res = bert_model(**inputs, output_hidden_states=True)
-        res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()[1:-1]
-    assert len(word2ph) == len(text)
-    phone_level_feature = []
-    for i in range(len(word2ph)):
-        repeat_feature = res[i].repeat(word2ph[i], 1)
-        phone_level_feature.append(repeat_feature)
-    phone_level_feature = torch.cat(phone_level_feature, dim=0)
-    return phone_level_feature.T
-
-
-class DictToAttrRecursive(dict):
-    def __init__(self, input_dict):
-        super().__init__(input_dict)
-        for key, value in input_dict.items():
-            if isinstance(value, dict):
-                value = DictToAttrRecursive(value)
-            self[key] = value
-            setattr(self, key, value)
-
-    def __getattr__(self, item):
-        try:
-            return self[item]
-        except KeyError:
-            raise AttributeError(f"Attribute {item} not found")
-
-    def __setattr__(self, key, value):
-        if isinstance(value, dict):
-            value = DictToAttrRecursive(value)
-        super(DictToAttrRecursive, self).__setitem__(key, value)
-        super().__setattr__(key, value)
-
-    def __delattr__(self, item):
-        try:
-            del self[item]
-        except KeyError:
-            raise AttributeError(f"Attribute {item} not found")
-
-
-ssl_model = cnhubert.get_model()
-if is_half == True:
-    ssl_model = ssl_model.half().to(device)
-else:
-    ssl_model = ssl_model.to(device)
-
-
-def change_sovits_weights(sovits_path,prompt_language=None,text_language=None):
-    global vq_model, hps, version, dict_language
-    dict_s2 = torch.load(sovits_path, map_location="cpu")
-    hps = dict_s2["config"]
-    hps = DictToAttrRecursive(hps)
-    hps.model.semantic_frame_rate = "25hz"
-    if dict_s2['weight']['enc_p.text_embedding.weight'].shape[0] == 322:
-        hps.model.version = "v1"
-    else:
-        hps.model.version = "v2"
-    version = hps.model.version
-    # print("sovits版本:",hps.model.version)
-    vq_model = SynthesizerTrn(
-        hps.data.filter_length // 2 + 1,
-        hps.train.segment_size // hps.data.hop_length,
-        n_speakers=hps.data.n_speakers,
-        **hps.model
-    )
-    if ("pretrained" not in sovits_path):
-        del vq_model.enc_q
-    if is_half == True:
-        vq_model = vq_model.half().to(device)
-    else:
-        vq_model = vq_model.to(device)
-    vq_model.eval()
-    print(vq_model.load_state_dict(dict_s2["weight"], strict=False))
-    dict_language = dict_language_v1 if version =='v1' else dict_language_v2
-    if prompt_language is not None and text_language is not None:
-        if prompt_language in list(dict_language.keys()):
-            prompt_text_update, prompt_language_update = {'__type__':'update'},  {'__type__':'update', 'value':prompt_language}
-        else:
-            prompt_text_update = {'__type__':'update', 'value':''}
-            prompt_language_update = {'__type__':'update', 'value':i18n("中文")}
-        if text_language in list(dict_language.keys()):
-            text_update, text_language_update = {'__type__':'update'}, {'__type__':'update', 'value':text_language}
-        else:
-            text_update = {'__type__':'update', 'value':''}
-            text_language_update = {'__type__':'update', 'value':i18n("中文")}
-        return  {'__type__':'update', 'choices':list(dict_language.keys())}, {'__type__':'update', 'choices':list(dict_language.keys())}, prompt_text_update, prompt_language_update, text_update, text_language_update
-
-
-
-change_sovits_weights("pretrained_models/gsv-v2final-pretrained/s2G2333k.pth")
-
-
-def change_gpt_weights(gpt_path):
-    global hz, max_sec, t2s_model, config
-    hz = 50
-    dict_s1 = torch.load(gpt_path, map_location="cpu")
-    config = dict_s1["config"]
-    max_sec = config["data"]["max_sec"]
-    t2s_model = Text2SemanticLightningModule(config, "****", is_train=False)
-    t2s_model.load_state_dict(dict_s1["weight"])
-    if is_half == True:
-        t2s_model = t2s_model.half()
-    t2s_model = t2s_model.to(device)
-    t2s_model.eval()
-    total = sum([param.nelement() for param in t2s_model.parameters()])
-    print("Number of parameter: %.2fM" % (total / 1e6))
-
-
-change_gpt_weights("pretrained_models/gsv-v2final-pretrained/s1bert25hz-5kh-longer-epoch=12-step=369668.ckpt")
-
-
-def get_spepc(hps, filename):
-    audio = load_audio(filename, int(hps.data.sampling_rate))
-    audio = torch.FloatTensor(audio)
-    maxx=audio.abs().max()
-    if(maxx>1):audio/=min(2,maxx)
-    audio_norm = audio
-    audio_norm = audio_norm.unsqueeze(0)
-    spec = spectrogram_torch(
-        audio_norm,
-        hps.data.filter_length,
-        hps.data.sampling_rate,
-        hps.data.hop_length,
-        hps.data.win_length,
-        center=False,
-    )
-    return spec
-
-def clean_text_inf(text, language, version):
-    phones, word2ph, norm_text = clean_text(text, language, version)
-    phones = cleaned_text_to_sequence(phones, version)
-    return phones, word2ph, norm_text
-
-dtype=torch.float16 if is_half == True else torch.float32
-def get_bert_inf(phones, word2ph, norm_text, language):
-    language=language.replace("all_","")
-    if language == "zh":
-        bert = get_bert_feature(norm_text, word2ph).to(device)#.to(dtype)
-    else:
-        bert = torch.zeros(
-            (1024, len(phones)),
-            dtype=torch.float16 if is_half == True else torch.float32,
-        ).to(device)
-
-    return bert
-
-
-splits = {"，", "。", "？", "！", ",", ".", "?", "!", "~", ":", "：", "—", "…", }
-
-
-def get_first(text):
-    pattern = "[" + "".join(re.escape(sep) for sep in splits) + "]"
-    text = re.split(pattern, text)[0].strip()
-    return text
-
-from text import chinese
-def get_phones_and_bert(text,language,version):
-    if language in {"en", "all_zh", "all_ja", "all_ko", "all_yue"}:
-        language = language.replace("all_","")
-        if language == "en":
-            LangSegment.setfilters(["en"])
-            formattext = " ".join(tmp["text"] for tmp in LangSegment.getTexts(text))
-        else:
-            # 因无法区别中日韩文汉字,以用户输入为准
-            formattext = text
-        while "  " in formattext:
-            formattext = formattext.replace("  ", " ")
-        if language == "zh":
-            if re.search(r'[A-Za-z]', formattext):
-                formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
-                formattext = chinese.mix_text_normalize(formattext)
-                return get_phones_and_bert(formattext,"zh",version)
-            else:
-                phones, word2ph, norm_text = clean_text_inf(formattext, language, version)
-                bert = get_bert_feature(norm_text, word2ph).to(device)
-        elif language == "yue" and re.search(r'[A-Za-z]', formattext):
-                formattext = re.sub(r'[a-z]', lambda x: x.group(0).upper(), formattext)
-                formattext = chinese.mix_text_normalize(formattext)
-                return get_phones_and_bert(formattext,"yue",version)
-        else:
-            phones, word2ph, norm_text = clean_text_inf(formattext, language, version)
-            bert = torch.zeros(
-                (1024, len(phones)),
-                dtype=torch.float16 if is_half == True else torch.float32,
-            ).to(device)
-    elif language in {"zh", "ja", "ko", "yue", "auto", "auto_yue"}:
-        textlist=[]
-        langlist=[]
-        LangSegment.setfilters(["zh","ja","en","ko"])
-        if language == "auto":
-            for tmp in LangSegment.getTexts(text):
-                langlist.append(tmp["lang"])
-                textlist.append(tmp["text"])
-        elif language == "auto_yue":
-            for tmp in LangSegment.getTexts(text):
-                if tmp["lang"] == "zh":
-                    tmp["lang"] = "yue"
-                langlist.append(tmp["lang"])
-                textlist.append(tmp["text"])
-        else:
-            for tmp in LangSegment.getTexts(text):
-                if tmp["lang"] == "en":
-                    langlist.append(tmp["lang"])
-                else:
-                    # 因无法区别中日韩文汉字,以用户输入为准
-                    langlist.append(language)
-                textlist.append(tmp["text"])
-        print(textlist)
-        print(langlist)
-        phones_list = []
-        bert_list = []
-        norm_text_list = []
-        for i in range(len(textlist)):
-            lang = langlist[i]
-            phones, word2ph, norm_text = clean_text_inf(textlist[i], lang, version)
-            bert = get_bert_inf(phones, word2ph, norm_text, lang)
-            phones_list.append(phones)
-            norm_text_list.append(norm_text)
-            bert_list.append(bert)
-        bert = torch.cat(bert_list, dim=1)
-        phones = sum(phones_list, [])
-        norm_text = ''.join(norm_text_list)
+"""
 
-    return phones,bert.to(dtype),norm_text
+with gr.Blocks(theme=gr.themes.Soft(), css=css, analytics_enabled=False) as demo:
+    gr.HTML('<div class="title"> Cartoon Image Generation </div>')
+    
+    gr.HTML("""<a href="https://visitorbadge.io/status?path=https%3A%2F%2Fginigen-cartoon.hf.space">
+               <img src="https://api.visitorbadge.io/api/visitors?path=https%3A%2F%2Fginigen-cartoon.hf.space&countColor=%23263759" />
+               </a>""")
+
+    
+    with gr.Tabs() as tabs:
+        with gr.Tab("Generation"):
+            with gr.Column(elem_id="col-container"):
+                with gr.Row():
+                    prompt = gr.Text(
+                        label="Prompt",
+                        show_label=False,
+                        max_lines=1,
+                        placeholder="Enter your prompt",
+                        container=False,
+                    )
+                    run_button = gr.Button("Run", scale=0)
 
+                # Modified to include the default image
+                result = gr.Image(
+                    label="Result",
+                    show_label=False,
+                    value=DEFAULT_IMAGE_PATH  # Set the default image
+                )
 
-def merge_short_text_in_array(texts, threshold):
-    if (len(texts)) < 2:
-        return texts
-    result = []
-    text = ""
-    for ele in texts:
-        text += ele
-        if len(text) >= threshold:
-            result.append(text)
-            text = ""
-    if (len(text) > 0):
-        if len(result) == 0:
-            result.append(text)
-        else:
-            result[len(result) - 1] += text
-    return result
+                with gr.Accordion("Advanced Settings", open=False):
+                    seed = gr.Slider(
+                        label="Seed",
+                        minimum=0,
+                        maximum=MAX_SEED,
+                        step=1,
+                        value=42,
+                    )
+                    randomize_seed = gr.Checkbox(label="Randomize seed", value=True)
+
+                    with gr.Row():
+                        width = gr.Slider(
+                            label="Width",
+                            minimum=256,
+                            maximum=MAX_IMAGE_SIZE,
+                            step=32,
+                            value=1024,
+                        )
+                        height = gr.Slider(
+                            label="Height",
+                            minimum=256,
+                            maximum=MAX_IMAGE_SIZE,
+                            step=32,
+                            value=768,
+                        )
+
+                    with gr.Row():
+                        guidance_scale = gr.Slider(
+                            label="Guidance scale",
+                            minimum=0.0,
+                            maximum=10.0,
+                            step=0.1,
+                            value=3.5,
+                        )
+                        num_inference_steps = gr.Slider(
+                            label="Number of inference steps",
+                            minimum=1,
+                            maximum=50,
+                            step=1,
+                            value=30,
+                        )
+                        lora_scale = gr.Slider(
+                            label="LoRA scale",
+                            minimum=0.0,
+                            maximum=1.0,
+                            step=0.1,
+                            value=1.0,
+                        )
+
+                gr.Examples(
+                    examples=examples,
+                    inputs=[prompt],
+                    outputs=[result, seed],
+                )
 
-##ref_wav_path+prompt_text+prompt_language+text(单个)+text_language+top_k+top_p+temperature
-# cache_tokens={}#暂未实现清理机制
-cache= {}
-@torch.inference_mode()
-@spaces.GPU
-def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language, how_to_cut=i18n("不切"), top_k=20, top_p=0.6, temperature=0.6, ref_free = False,speed=1,if_freeze=False,inp_refs=123):
-    global cache
-    if ref_wav_path:pass
-    else:gr.Warning(i18n('请上传参考音频'))
-    if text:pass
-    else:gr.Warning(i18n('请填入推理文本'))
-    t = []
-    if prompt_text is None or len(prompt_text) == 0:
-        ref_free = True
-    t0 = ttime()
-    prompt_language = dict_language[prompt_language]
-    text_language = dict_language[text_language]
+        with gr.Tab("Gallery"):
+            gallery_header = gr.Markdown("### Generated Images Gallery")
+            generated_gallery = gr.Gallery(
+                label="Generated Images",
+                columns=6,
+                show_label=False,
+                value=load_generated_images(),
+                elem_id="generated_gallery",
+                height="auto"
+            )
+            refresh_btn = gr.Button("🔄 Refresh Gallery")
 
 
-    if not ref_free:
-        prompt_text = prompt_text.strip("\n")
-        if (prompt_text[-1] not in splits): prompt_text += "。" if prompt_language != "en" else "."
-        print(i18n("实际输入的参考文本:"), prompt_text)
-    text = text.strip("\n")
-    if (text[0] not in splits and len(get_first(text)) < 4): text = "。" + text if text_language != "en" else "." + text
+    # Event handlers
+    def refresh_gallery():
+        return load_generated_images()
 
-    print(i18n("实际输入的目标文本:"), text)
-    zero_wav = np.zeros(
-        int(hps.data.sampling_rate * 0.3),
-        dtype=np.float16 if is_half == True else np.float32,
+    refresh_btn.click(
+        fn=refresh_gallery,
+        inputs=None,
+        outputs=generated_gallery,
     )
-    if not ref_free:
-        with torch.no_grad():
-            wav16k, sr = librosa.load(ref_wav_path, sr=16000)
-            if (wav16k.shape[0] > 160000 or wav16k.shape[0] < 48000):
-                gr.Warning(i18n("参考音频在3~10秒范围外，请更换！"))
-                raise OSError(i18n("参考音频在3~10秒范围外，请更换！"))
-            wav16k = torch.from_numpy(wav16k)
-            zero_wav_torch = torch.from_numpy(zero_wav)
-            if is_half == True:
-                wav16k = wav16k.half().to(device)
-                zero_wav_torch = zero_wav_torch.half().to(device)
-            else:
-                wav16k = wav16k.to(device)
-                zero_wav_torch = zero_wav_torch.to(device)
-            wav16k = torch.cat([wav16k, zero_wav_torch])
-            ssl_content = ssl_model.model(wav16k.unsqueeze(0))[
-                "last_hidden_state"
-            ].transpose(
-                1, 2
-            )  # .float()
-            codes = vq_model.extract_latent(ssl_content)
-            prompt_semantic = codes[0, 0]
-            prompt = prompt_semantic.unsqueeze(0).to(device)
-
-    t1 = ttime()
-    t.append(t1-t0)
-
-    if (how_to_cut == i18n("凑四句一切")):
-        text = cut1(text)
-    elif (how_to_cut == i18n("凑50字一切")):
-        text = cut2(text)
-    elif (how_to_cut == i18n("按中文句号。切")):
-        text = cut3(text)
-    elif (how_to_cut == i18n("按英文句号.切")):
-        text = cut4(text)
-    elif (how_to_cut == i18n("按标点符号切")):
-        text = cut5(text)
-    while "\n\n" in text:
-        text = text.replace("\n\n", "\n")
-    print(i18n("实际输入的目标文本(切句后):"), text)
-    texts = text.split("\n")
-    texts = process_text(texts)
-    texts = merge_short_text_in_array(texts, 5)
-    audio_opt = []
-    if not ref_free:
-        phones1,bert1,norm_text1=get_phones_and_bert(prompt_text, prompt_language, version)
-
-    for i_text,text in enumerate(texts):
-        # 解决输入目标文本的空行导致报错的问题
-        if (len(text.strip()) == 0):
-            continue
-        if (text[-1] not in splits): text += "。" if text_language != "en" else "."
-        print(i18n("实际输入的目标文本(每句):"), text)
-        phones2,bert2,norm_text2=get_phones_and_bert(text, text_language, version)
-        print(i18n("前端处理后的文本(每句):"), norm_text2)
-        if not ref_free:
-            bert = torch.cat([bert1, bert2], 1)
-            all_phoneme_ids = torch.LongTensor(phones1+phones2).to(device).unsqueeze(0)
-        else:
-            bert = bert2
-            all_phoneme_ids = torch.LongTensor(phones2).to(device).unsqueeze(0)
-
-        bert = bert.to(device).unsqueeze(0)
-        all_phoneme_len = torch.tensor([all_phoneme_ids.shape[-1]]).to(device)
 
-        t2 = ttime()
-        # cache_key="%s-%s-%s-%s-%s-%s-%s-%s"%(ref_wav_path,prompt_text,prompt_language,text,text_language,top_k,top_p,temperature)
-        # print(cache.keys(),if_freeze)
-        if(i_text in cache and if_freeze==True):pred_semantic=cache[i_text]
-        else:
-            with torch.no_grad():
-                pred_semantic, idx = t2s_model.model.infer_panel(
-                    all_phoneme_ids,
-                    all_phoneme_len,
-                    None if ref_free else prompt,
-                    bert,
-                    # prompt_phone_len=ph_offset,
-                    top_k=top_k,
-                    top_p=top_p,
-                    temperature=temperature,
-                    early_stop_num=hz * max_sec,
-                )
-                pred_semantic = pred_semantic[:, -idx:].unsqueeze(0)
-                cache[i_text]=pred_semantic
-        t3 = ttime()
-        refers=[]
-        if(inp_refs):
-            for path in inp_refs:
-                try:
-                    refer = get_spepc(hps, path.name).to(dtype).to(device)
-                    refers.append(refer)
-                except:
-                    traceback.print_exc()
-        if(len(refers)==0):refers = [get_spepc(hps, ref_wav_path).to(dtype).to(device)]
-        audio = (vq_model.decode(pred_semantic, torch.LongTensor(phones2).to(device).unsqueeze(0), refers,speed=speed).detach().cpu().numpy()[0, 0])
-        max_audio=np.abs(audio).max()#简单防止16bit爆音
-        if max_audio>1:audio/=max_audio
-        audio_opt.append(audio)
-        audio_opt.append(zero_wav)
-        t4 = ttime()
-        t.extend([t2 - t1,t3 - t2, t4 - t3])
-        t1 = ttime()
-    print("%.3f\t%.3f\t%.3f\t%.3f" %
-           (t[0], sum(t[1::3]), sum(t[2::3]), sum(t[3::3]))
-           )
-    yield hps.data.sampling_rate, (np.concatenate(audio_opt, 0) * 32768).astype(
-        np.int16
+    gr.on(
+        triggers=[run_button.click, prompt.submit],
+        fn=inference,
+        inputs=[
+            prompt,
+            seed,
+            randomize_seed,
+            width,
+            height,
+            guidance_scale,
+            num_inference_steps,
+            lora_scale,
+        ],
+        outputs=[result, seed, generated_gallery],
     )
 
-
-def split(todo_text):
-    todo_text = todo_text.replace("……", "。").replace("——", "，")
-    if todo_text[-1] not in splits:
-        todo_text += "。"
-    i_split_head = i_split_tail = 0
-    len_text = len(todo_text)
-    todo_texts = []
-    while 1:
-        if i_split_head >= len_text:
-            break  # 结尾一定有标点，所以直接跳出即可，最后一段在上次已加入
-        if todo_text[i_split_head] in splits:
-            i_split_head += 1
-            todo_texts.append(todo_text[i_split_tail:i_split_head])
-            i_split_tail = i_split_head
-        else:
-            i_split_head += 1
-    return todo_texts
-
-
-def cut1(inp):
-    inp = inp.strip("\n")
-    inps = split(inp)
-    split_idx = list(range(0, len(inps), 4))
-    split_idx[-1] = None
-    if len(split_idx) > 1:
-        opts = []
-        for idx in range(len(split_idx) - 1):
-            opts.append("".join(inps[split_idx[idx]: split_idx[idx + 1]]))
-    else:
-        opts = [inp]
-    opts = [item for item in opts if not set(item).issubset(punctuation)]
-    return "\n".join(opts)
-
-
-def cut2(inp):
-    inp = inp.strip("\n")
-    inps = split(inp)
-    if len(inps) < 2:
-        return inp
-    opts = []
-    summ = 0
-    tmp_str = ""
-    for i in range(len(inps)):
-        summ += len(inps[i])
-        tmp_str += inps[i]
-        if summ > 50:
-            summ = 0
-            opts.append(tmp_str)
-            tmp_str = ""
-    if tmp_str != "":
-        opts.append(tmp_str)
-    # print(opts)
-    if len(opts) > 1 and len(opts[-1]) < 50:  ##如果最后一个太短了，和前一个合一起
-        opts[-2] = opts[-2] + opts[-1]
-        opts = opts[:-1]
-    opts = [item for item in opts if not set(item).issubset(punctuation)]
-    return "\n".join(opts)
-
-
-def cut3(inp):
-    inp = inp.strip("\n")
-    opts = ["%s" % item for item in inp.strip("。").split("。")]
-    opts = [item for item in opts if not set(item).issubset(punctuation)]
-    return  "\n".join(opts)
-
-def cut4(inp):
-    inp = inp.strip("\n")
-    opts = ["%s" % item for item in inp.strip(".").split(".")]
-    opts = [item for item in opts if not set(item).issubset(punctuation)]
-    return "\n".join(opts)
-
-
-# contributed by https://github.com/AI-Hobbyist/GPT-SoVITS/blob/main/GPT_SoVITS/inference_webui.py
-def cut5(inp):
-    inp = inp.strip("\n")
-    punds = {',', '.', ';', '?', '!', '、', '，', '。', '？', '！', ';', '：', '…'}
-    mergeitems = []
-    items = []
-
-    for i, char in enumerate(inp):
-        if char in punds:
-            if char == '.' and i > 0 and i < len(inp) - 1 and inp[i - 1].isdigit() and inp[i + 1].isdigit():
-                items.append(char)
-            else:
-                items.append(char)
-                mergeitems.append("".join(items))
-                items = []
-        else:
-            items.append(char)
-
-    if items:
-        mergeitems.append("".join(items))
-
-    opt = [item for item in mergeitems if not set(item).issubset(punds)]
-    return "\n".join(opt)
-
-
-def custom_sort_key(s):
-    # 使用正则表达式提取字符串中的数字部分和非数字部分
-    parts = re.split('(\d+)', s)
-    # 将数字部分转换为整数，非数字部分保持不变
-    parts = [int(part) if part.isdigit() else part for part in parts]
-    return parts
-
-def process_text(texts):
-    _text=[]
-    if all(text in [None, " ", "\n",""] for text in texts):
-        raise ValueError(i18n("请输入有效文本"))
-    for text in texts:
-        if text in  [None, " ", ""]:
-            pass
-        else:
-            _text.append(text)
-    return _text
-
-
-def html_center(text, label='p'):
-    return f"""<div style="text-align: center; margin: 100; padding: 50;">
-                <{label} style="margin: 0; padding: 0;">{text}</{label}>
-                </div>"""
-
-def html_left(text, label='p'):
-    return f"""<div style="text-align: left; margin: 0; padding: 0;">
-                <{label} style="margin: 0; padding: 0;">{text}</{label}>
-                </div>"""
-
-css = """
-footer {
-    visibility: hidden;
-}
-"""
-
-with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css) as app:
-
-    with gr.Group():
-        gr.Markdown(html_center(i18n("*请上传并填写参考信息"),'h3'))
-        with gr.Row():
-            inp_ref = gr.Audio(label=i18n("请上传3~10秒内参考音频，超过会报错！"), type="filepath")
-            with gr.Column():
-                ref_text_free = gr.Checkbox(label=i18n("开启无参考文本模式。不填参考文本亦相当于开启。"), value=False, interactive=True, show_label=True)
-                gr.Markdown(html_left(i18n("使用无参考文本模式时建议使用微调的GPT，听不清参考音频说的啥(不晓得写啥)可以开。<br>开启后无视填写的参考文本。")))
-                prompt_text = gr.Textbox(label=i18n("参考音频的文本"), value="", lines=3, max_lines=3)
-            prompt_language = gr.Dropdown(
-                label=i18n("参考音频的语种"), choices=list(dict_language.keys()), value=i18n("中文")
-            )
-            inp_refs = gr.File(label=i18n("可选项：通过拖拽多个文件上传多个参考音频（建议同性），平均融合他们的音色。如不填写此项，音色由左侧单个参考音频控制。"),file_count="multiple")
-        gr.Markdown(html_center(i18n("*请填写需要合成的目标文本和语种模式"),'h3'))
-        with gr.Row():
-            with gr.Column():
-                text = gr.Textbox(label=i18n("需要合成的文本"), value="", lines=26, max_lines=26)
-            with gr.Column():
-                text_language = gr.Dropdown(
-                        label=i18n("需要合成的语种")+i18n(".限制范围越小判别效果越好。"), choices=list(dict_language.keys()), value=i18n("中文")
-                    )
-                how_to_cut = gr.Dropdown(
-                        label=i18n("怎么切"),
-                        choices=[i18n("不切"), i18n("凑四句一切"), i18n("凑50字一切"), i18n("按中文句号。切"), i18n("按英文句号.切"), i18n("按标点符号切"), ],
-                        value=i18n("凑四句一切"),
-                        interactive=True
-                    )
-                gr.Markdown(value=html_center(i18n("语速调整，高为更快")))
-                if_freeze=gr.Checkbox(label=i18n("是否直接对上次合成结果调整语速和音色。防止随机性。"), value=False, interactive=True,show_label=True)
-                speed = gr.Slider(minimum=0.6,maximum=1.65,step=0.05,label=i18n("语速"),value=1,interactive=True)
-                gr.Markdown(html_center(i18n("GPT采样参数(无参考文本时不要太低。不懂就用默认)：")))
-                top_k = gr.Slider(minimum=1,maximum=100,step=1,label=i18n("top_k"),value=15,interactive=True)
-                top_p = gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("top_p"),value=1,interactive=True)
-                temperature = gr.Slider(minimum=0,maximum=1,step=0.05,label=i18n("temperature"),value=1,interactive=True)
-        with gr.Row():
-            inference_button = gr.Button(i18n("合成语音"), variant="primary", size='lg')
-            output = gr.Audio(label=i18n("输出的语音"))
-
-        inference_button.click(
-            get_tts_wav,
-            [inp_ref, prompt_text, prompt_language, text, text_language, how_to_cut, top_k, top_p, temperature, ref_text_free,speed,if_freeze,inp_refs],
-            [output],
-        )
-
-if __name__ == '__main__':
-    # app.queue(concurrency_count=511, max_size=1022).launch(
-    app.queue().launch(
-        server_name="0.0.0.0",
-        inbrowser=True,
-        # share=True,
-        # server_port=infer_ttswebui,
-        quiet=True,
-    )
+demo.queue()
+demo.launch()