try: import spaces except: class NoneSpaces: def __init__(self): pass def GPU(self, fn): return fn spaces = NoneSpaces() import os import logging import sys import numpy as np from modules.devices import devices from modules.synthesize_audio import synthesize_audio logging.basicConfig( level=os.getenv("LOG_LEVEL", "INFO"), format="%(asctime)s - %(name)s - %(levelname)s - %(message)s", ) import gradio as gr import torch from modules.ssml import parse_ssml from modules.SynthesizeSegments import SynthesizeSegments, combine_audio_segments from modules.speaker import speaker_mgr from modules.data import styles_mgr from modules.api.utils import calc_spk_style import modules.generate_audio as generate from modules.normalization import text_normalize from modules import refiner, config, models from modules.utils import env, audio from modules.SentenceSplitter import SentenceSplitter # fix: If the system proxy is enabled in the Windows system, you need to skip these os.environ["NO_PROXY"] = "localhost,127.0.0.1,0.0.0.0" torch._dynamo.config.cache_size_limit = 64 torch._dynamo.config.suppress_errors = True torch.set_float32_matmul_precision("high") webui_config = { "tts_max": 1000, "ssml_max": 5000, "spliter_threshold": 100, "max_batch_size": 8, } def get_speakers(): return speaker_mgr.list_speakers() def get_styles(): return styles_mgr.list_items() def segments_length_limit(segments, total_max: int): ret_segments = [] total_len = 0 for seg in segments: if "text" not in seg: continue total_len += len(seg["text"]) if total_len > total_max: break ret_segments.append(seg) return ret_segments @torch.inference_mode() @spaces.GPU def synthesize_ssml(ssml: str, batch_size=4): try: batch_size = int(batch_size) except Exception: batch_size = 8 ssml = ssml.strip() if ssml == "": return None segments = parse_ssml(ssml) max_len = webui_config["ssml_max"] segments = segments_length_limit(segments, max_len) if len(segments) == 0: return None models.load_chat_tts() synthesize = SynthesizeSegments(batch_size=batch_size) audio_segments = synthesize.synthesize_segments(segments) combined_audio = combine_audio_segments(audio_segments) return audio.pydub_to_np(combined_audio) @torch.inference_mode() @spaces.GPU def tts_generate( text, temperature, top_p, top_k, spk, infer_seed, use_decoder, prompt1, prompt2, prefix, style, disable_normalize=False, batch_size=4, ): try: batch_size = int(batch_size) except Exception: batch_size = 4 max_len = webui_config["tts_max"] text = text.strip()[0:max_len] if text == "": return None if style == "*auto": style = None if isinstance(top_k, float): top_k = int(top_k) params = calc_spk_style(spk=spk, style=style) spk = params.get("spk", spk) infer_seed = infer_seed or params.get("seed", infer_seed) temperature = temperature or params.get("temperature", temperature) prefix = prefix or params.get("prefix", prefix) prompt1 = prompt1 or params.get("prompt1", "") prompt2 = prompt2 or params.get("prompt2", "") infer_seed = np.clip(infer_seed, -1, 2**32 - 1, out=None, dtype=np.int64) infer_seed = int(infer_seed) if not disable_normalize: text = text_normalize(text) models.load_chat_tts() sample_rate, audio_data = synthesize_audio( text=text, temperature=temperature, top_P=top_p, top_K=top_k, spk=spk, infer_seed=infer_seed, use_decoder=use_decoder, prompt1=prompt1, prompt2=prompt2, prefix=prefix, batch_size=batch_size, ) audio_data = audio.audio_to_int16(audio_data) return sample_rate, audio_data @torch.inference_mode() @spaces.GPU def refine_text(text: str, prompt: str): text = text_normalize(text) return refiner.refine_text(text, prompt=prompt) def read_local_readme(): with open("README.md", "r", encoding="utf-8") as file: content = file.read() content = content[content.index("# ") :] return content # 演示示例文本 sample_texts = [ { "text": "大🍌,一条大🍌,嘿,你的感觉真的很奇妙 [lbreak]", }, { "text": "天气预报显示,今天会有小雨,请大家出门时记得带伞。降温的天气也提醒我们要适时添衣保暖 [lbreak]", }, { "text": "公司的年度总结会议将在下周三举行,请各部门提前准备好相关材料,确保会议顺利进行 [lbreak]", }, { "text": "今天的午餐菜单包括烤鸡、沙拉和蔬菜汤,大家可以根据自己的口味选择适合的菜品 [lbreak]", }, { "text": "请注意,电梯将在下午两点进行例行维护,预计需要一个小时的时间,请大家在此期间使用楼梯 [lbreak]", }, { "text": "图书馆新到了一批书籍,涵盖了文学、科学和历史等多个领域,欢迎大家前来借阅 [lbreak]", }, { "text": "电影中梁朝伟扮演的陈永仁的编号27149 [lbreak]", }, { "text": "这块黄金重达324.75克 [lbreak]", }, { "text": "我们班的最高总分为583分 [lbreak]", }, { "text": "12~23 [lbreak]", }, { "text": "-1.5~2 [lbreak]", }, { "text": "她出生于86年8月18日,她弟弟出生于1995年3月1日 [lbreak]", }, { "text": "等会请在12:05请通知我 [lbreak]", }, { "text": "今天的最低气温达到-10°C [lbreak]", }, { "text": "现场有7/12的观众投出了赞成票 [lbreak]", }, { "text": "明天有62%的概率降雨 [lbreak]", }, { "text": "随便来几个价格12块5,34.5元,20.1万 [lbreak]", }, { "text": "这是固话0421-33441122 [lbreak]", }, { "text": "这是手机+86 18544139121 [lbreak]", }, ] ssml_example1 = """ <speak version="0.1"> <voice spk="Bob" seed="42" style="narration-relaxed"> 下面是一个 ChatTTS 用于合成多角色多情感的有声书示例[lbreak] </voice> <voice spk="Bob" seed="42" style="narration-relaxed"> 黛玉冷笑道:[lbreak] </voice> <voice spk="female2" seed="42" style="angry"> 我说呢 [uv_break] ,亏了绊住,不然,早就飞起来了[lbreak] </voice> <voice spk="Bob" seed="42" style="narration-relaxed"> 宝玉道:[lbreak] </voice> <voice spk="Alice" seed="42" style="unfriendly"> “只许和你玩 [uv_break] ,替你解闷。不过偶然到他那里,就说这些闲话。”[lbreak] </voice> <voice spk="female2" seed="42" style="angry"> “好没意思的话![uv_break] 去不去,关我什么事儿? 又没叫你替我解闷儿 [uv_break],还许你不理我呢” [lbreak] </voice> <voice spk="Bob" seed="42" style="narration-relaxed"> 说着,便赌气回房去了 [lbreak] </voice> </speak> """ ssml_example2 = """ <speak version="0.1"> <voice spk="Bob" seed="42" style="narration-relaxed"> 使用 prosody 控制生成文本的语速语调和音量,示例如下 [lbreak] <prosody> 无任何限制将会继承父级voice配置进行生成 [lbreak] </prosody> <prosody rate="1.5"> 设置 rate 大于1表示加速,小于1为减速 [lbreak] </prosody> <prosody pitch="6"> 设置 pitch 调整音调,设置为6表示提高6个半音 [lbreak] </prosody> <prosody volume="2"> 设置 volume 调整音量,设置为2表示提高2个分贝 [lbreak] </prosody> 在 voice 中无prosody包裹的文本即为默认生成状态下的语音 [lbreak] </voice> </speak> """ ssml_example3 = """ <speak version="0.1"> <voice spk="Bob" seed="42" style="narration-relaxed"> 使用 break 标签将会简单的 [lbreak] <break time="500" /> 插入一段空白到生成结果中 [lbreak] </voice> </speak> """ ssml_example4 = """ <speak version="0.1"> <voice spk="Bob" seed="42" style="excited"> temperature for sampling (may be overridden by style or speaker) [lbreak] <break time="500" /> 温度值用于采样,这个值有可能被 style 或者 speaker 覆盖 [lbreak] <break time="500" /> temperature for sampling ,这个值有可能被 style 或者 speaker 覆盖 [lbreak] <break time="500" /> 温度值用于采样,(may be overridden by style or speaker) [lbreak] </voice> </speak> """ default_ssml = """ <speak version="0.1"> <voice spk="Bob" seed="42" style="narration-relaxed"> 这里是一个简单的 SSML 示例 [lbreak] </voice> </speak> """ def create_tts_interface(): speakers = get_speakers() def get_speaker_show_name(spk): if spk.gender == "*" or spk.gender == "": return spk.name return f"{spk.gender} : {spk.name}" speaker_names = ["*random"] + [ get_speaker_show_name(speaker) for speaker in speakers ] styles = ["*auto"] + [s.get("name") for s in get_styles()] history = [] with gr.Row(): with gr.Column(scale=1): with gr.Group(): gr.Markdown("🎛️Sampling") temperature_input = gr.Slider( 0.01, 2.0, value=0.3, step=0.01, label="Temperature" ) top_p_input = gr.Slider(0.1, 1.0, value=0.7, step=0.1, label="Top P") top_k_input = gr.Slider(1, 50, value=20, step=1, label="Top K") batch_size_input = gr.Slider( 1, webui_config["max_batch_size"], value=4, step=1, label="Batch Size", ) with gr.Row(): with gr.Group(): gr.Markdown("🎭Style") gr.Markdown("- 后缀为 `_p` 表示带prompt,效果更强但是影响质量") style_input_dropdown = gr.Dropdown( choices=styles, # label="Choose Style", interactive=True, show_label=False, value="*auto", ) with gr.Row(): with gr.Group(): gr.Markdown("🗣️Speaker (Name or Seed)") spk_input_text = gr.Textbox( label="Speaker (Text or Seed)", value="female2", show_label=False, ) spk_input_dropdown = gr.Dropdown( choices=speaker_names, # label="Choose Speaker", interactive=True, value="female : female2", show_label=False, ) spk_rand_button = gr.Button( value="🎲", # tooltip="Random Seed", variant="secondary", ) spk_input_dropdown.change( fn=lambda x: x.startswith("*") and "-1" or x.split(":")[-1].strip(), inputs=[spk_input_dropdown], outputs=[spk_input_text], ) spk_rand_button.click( lambda x: str(torch.randint(0, 2**32 - 1, (1,)).item()), inputs=[spk_input_text], outputs=[spk_input_text], ) with gr.Group(): gr.Markdown("💃Inference Seed") infer_seed_input = gr.Number( value=42, label="Inference Seed", show_label=False, minimum=-1, maximum=2**32 - 1, ) infer_seed_rand_button = gr.Button( value="🎲", # tooltip="Random Seed", variant="secondary", ) use_decoder_input = gr.Checkbox( value=True, label="Use Decoder", visible=False ) with gr.Group(): gr.Markdown("🔧Prompt engineering") prompt1_input = gr.Textbox(label="Prompt 1") prompt2_input = gr.Textbox(label="Prompt 2") prefix_input = gr.Textbox(label="Prefix") infer_seed_rand_button.click( lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()), inputs=[infer_seed_input], outputs=[infer_seed_input], ) with gr.Column(scale=3): with gr.Row(): with gr.Column(scale=4): with gr.Group(): input_title = gr.Markdown( "📝Text Input", elem_id="input-title", ) gr.Markdown( f"- 字数限制{webui_config['tts_max']:,}字,超过部分截断" ) gr.Markdown("- 如果尾字吞字不读,可以试试结尾加上 `[lbreak]`") gr.Markdown( "- If the input text is all in English, it is recommended to check disable_normalize" ) text_input = gr.Textbox( show_label=False, label="Text to Speech", lines=10, placeholder="输入文本或选择示例", elem_id="text-input", ) # TODO 字数统计,其实实现很好写,但是就是会触发loading...并且还要和后端交互... # text_input.change( # fn=lambda x: ( # f"📝Text Input ({len(x)} char)" # if x # else ( # "📝Text Input (0 char)" # if not x # else "📝Text Input (0 char)" # ) # ), # inputs=[text_input], # outputs=[input_title], # ) with gr.Row(): contorl_tokens = [ "[laugh]", "[uv_break]", "[v_break]", "[lbreak]", ] for tk in contorl_tokens: t_btn = gr.Button(tk) t_btn.click( lambda text, tk=tk: text + " " + tk, inputs=[text_input], outputs=[text_input], ) with gr.Column(scale=1): with gr.Group(): gr.Markdown("🎶Refiner") refine_prompt_input = gr.Textbox( label="Refine Prompt", value="[oral_2][laugh_0][break_6]", ) refine_button = gr.Button("✍️Refine Text") # TODO 分割句子,使用当前配置拼接为SSML,然后发送到SSML tab # send_button = gr.Button("📩Split and send to SSML") with gr.Group(): gr.Markdown("🔊Generate") disable_normalize_input = gr.Checkbox( value=False, label="Disable Normalize" ) tts_button = gr.Button( "🔊Generate Audio", variant="primary", elem_classes="big-button", ) with gr.Group(): gr.Markdown("🎄Examples") sample_dropdown = gr.Dropdown( choices=[sample["text"] for sample in sample_texts], show_label=False, value=None, interactive=True, ) sample_dropdown.change( fn=lambda x: x, inputs=[sample_dropdown], outputs=[text_input], ) with gr.Group(): gr.Markdown("🎨Output") tts_output = gr.Audio(label="Generated Audio") refine_button.click( refine_text, inputs=[text_input, refine_prompt_input], outputs=[text_input], ) tts_button.click( tts_generate, inputs=[ text_input, temperature_input, top_p_input, top_k_input, spk_input_text, infer_seed_input, use_decoder_input, prompt1_input, prompt2_input, prefix_input, style_input_dropdown, disable_normalize_input, batch_size_input, ], outputs=tts_output, ) def create_ssml_interface(): examples = [ ssml_example1, ssml_example2, ssml_example3, ssml_example4, ] with gr.Row(): with gr.Column(scale=3): with gr.Group(): gr.Markdown("📝SSML Input") gr.Markdown(f"- 最长{webui_config['ssml_max']:,}字符,超过会被截断") gr.Markdown("- 尽量保证使用相同的 seed") gr.Markdown( "- 关于SSML可以看这个 [文档](https://github.com/lenML/ChatTTS-Forge/blob/main/docs/SSML.md)" ) ssml_input = gr.Textbox( label="SSML Input", lines=10, value=default_ssml, placeholder="输入 SSML 或选择示例", elem_id="ssml_input", show_label=False, ) ssml_button = gr.Button("🔊Synthesize SSML", variant="primary") with gr.Column(scale=1): with gr.Group(): # 参数 gr.Markdown("🎛️Parameters") # batch size batch_size_input = gr.Slider( label="Batch Size", value=4, minimum=1, maximum=webui_config["max_batch_size"], step=1, ) with gr.Group(): gr.Markdown("🎄Examples") gr.Examples( examples=examples, inputs=[ssml_input], ) ssml_output = gr.Audio(label="Generated Audio") ssml_button.click( synthesize_ssml, inputs=[ssml_input, batch_size_input], outputs=ssml_output, ) return ssml_input # NOTE: 这个其实是需要GPU的...但是spaces会自动卸载,所以不太好使,具体处理在text_normalize中兼容 # @spaces.GPU def split_long_text(long_text_input): spliter = SentenceSplitter(webui_config["spliter_threshold"]) sentences = spliter.parse(long_text_input) sentences = [text_normalize(s) for s in sentences] data = [] for i, text in enumerate(sentences): data.append([i, text, len(text)]) return data def merge_dataframe_to_ssml(dataframe, spk, style, seed): if style == "*auto": style = None if spk == "-1" or spk == -1: spk = None if seed == -1 or seed == "-1": seed = None ssml = "" indent = " " * 2 for i, row in dataframe.iterrows(): ssml += f"{indent}<voice" if spk: ssml += f' spk="{spk}"' if style: ssml += f' style="{style}"' if seed: ssml += f' seed="{seed}"' ssml += ">\n" ssml += f"{indent}{indent}{text_normalize(row[1])}\n" ssml += f"{indent}</voice>\n" return f"<speak version='0.1'>\n{ssml}</speak>" # 长文本处理 # 可以输入长文本,并选择切割方法,切割之后可以将拼接的SSML发送到SSML tab # 根据 。 句号切割,切割之后显示到 data table def create_long_content_tab(ssml_input, tabs): speakers = get_speakers() def get_speaker_show_name(spk): if spk.gender == "*" or spk.gender == "": return spk.name return f"{spk.gender} : {spk.name}" speaker_names = ["*random"] + [ get_speaker_show_name(speaker) for speaker in speakers ] styles = ["*auto"] + [s.get("name") for s in get_styles()] with gr.Row(): with gr.Column(scale=1): # 选择说话人 选择风格 选择seed with gr.Group(): gr.Markdown("🗣️Speaker") spk_input_text = gr.Textbox( label="Speaker (Text or Seed)", value="female2", show_label=False, ) spk_input_dropdown = gr.Dropdown( choices=speaker_names, interactive=True, value="female : female2", show_label=False, ) spk_rand_button = gr.Button( value="🎲", variant="secondary", ) with gr.Group(): gr.Markdown("🎭Style") style_input_dropdown = gr.Dropdown( choices=styles, interactive=True, show_label=False, value="*auto", ) with gr.Group(): gr.Markdown("🗣️Seed") infer_seed_input = gr.Number( value=42, label="Inference Seed", show_label=False, minimum=-1, maximum=2**32 - 1, ) infer_seed_rand_button = gr.Button( value="🎲", variant="secondary", ) send_btn = gr.Button("📩Send to SSML", variant="primary") with gr.Column(scale=3): with gr.Group(): gr.Markdown("📝Long Text Input") gr.Markdown("- 此页面用于处理超长文本") gr.Markdown("- 切割后,可以选择说话人、风格、seed,然后发送到SSML") long_text_input = gr.Textbox( label="Long Text Input", lines=10, placeholder="输入长文本", elem_id="long-text-input", show_label=False, ) long_text_split_button = gr.Button("🔪Split Text") with gr.Row(): with gr.Column(scale=3): with gr.Group(): gr.Markdown("🎨Output") long_text_output = gr.DataFrame( headers=["index", "text", "length"], datatype=["number", "str", "number"], elem_id="long-text-output", interactive=False, wrap=True, value=[], ) spk_input_dropdown.change( fn=lambda x: x.startswith("*") and "-1" or x.split(":")[-1].strip(), inputs=[spk_input_dropdown], outputs=[spk_input_text], ) spk_rand_button.click( lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()), inputs=[spk_input_text], outputs=[spk_input_text], ) infer_seed_rand_button.click( lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()), inputs=[infer_seed_input], outputs=[infer_seed_input], ) long_text_split_button.click( split_long_text, inputs=[long_text_input], outputs=[long_text_output], ) infer_seed_rand_button.click( lambda x: int(torch.randint(0, 2**32 - 1, (1,)).item()), inputs=[infer_seed_input], outputs=[infer_seed_input], ) send_btn.click( merge_dataframe_to_ssml, inputs=[ long_text_output, spk_input_text, style_input_dropdown, infer_seed_input, ], outputs=[ssml_input], ) def change_tab(): return gr.Tabs(selected="ssml") send_btn.click(change_tab, inputs=[], outputs=[tabs]) def create_readme_tab(): readme_content = read_local_readme() gr.Markdown(readme_content) def create_app_footer(): gradio_version = gr.__version__ git_tag = config.versions.git_tag git_commit = config.versions.git_commit git_branch = config.versions.git_branch python_version = config.versions.python_version torch_version = config.versions.torch_version config.versions.gradio_version = gradio_version gr.Markdown( f""" 🍦 [ChatTTS-Forge](https://github.com/lenML/ChatTTS-Forge) version: [{git_tag}](https://github.com/lenML/ChatTTS-Forge/commit/{git_commit}) | branch: `{git_branch}` | python: `{python_version}` | torch: `{torch_version}` """ ) def create_interface(): js_func = """ function refresh() { const url = new URL(window.location); if (url.searchParams.get('__theme') !== 'dark') { url.searchParams.set('__theme', 'dark'); window.location.href = url.href; } } """ head_js = """ <script> </script> """ with gr.Blocks(js=js_func, head=head_js, title="ChatTTS Forge WebUI") as demo: css = """ <style> .big-button { height: 80px; } #input_title div.eta-bar { display: none !important; transform: none !important; } footer { display: none !important; } </style> """ gr.HTML(css) with gr.Tabs() as tabs: with gr.TabItem("TTS"): create_tts_interface() with gr.TabItem("SSML", id="ssml"): ssml_input = create_ssml_interface() with gr.TabItem("Long Text"): create_long_content_tab(ssml_input, tabs=tabs) with gr.TabItem("README"): create_readme_tab() create_app_footer() return demo if __name__ == "__main__": import argparse import dotenv dotenv.load_dotenv( dotenv_path=os.getenv("ENV_FILE", ".env.webui"), ) parser = argparse.ArgumentParser(description="Gradio App") parser.add_argument("--server_name", type=str, help="server name") parser.add_argument("--server_port", type=int, help="server port") parser.add_argument( "--share", action="store_true", help="share the gradio interface" ) parser.add_argument("--debug", action="store_true", help="enable debug mode") parser.add_argument("--auth", type=str, help="username:password for authentication") parser.add_argument( "--half", action="store_true", help="Enable half precision for model inference", ) parser.add_argument( "--off_tqdm", action="store_true", help="Disable tqdm progress bar", ) parser.add_argument( "--tts_max_len", type=int, help="Max length of text for TTS", ) parser.add_argument( "--ssml_max_len", type=int, help="Max length of text for SSML", ) parser.add_argument( "--max_batch_size", type=int, help="Max batch size for TTS", ) parser.add_argument( "--lru_size", type=int, default=64, help="Set the size of the request cache pool, set it to 0 will disable lru_cache", ) parser.add_argument( "--device_id", type=str, help="Select the default CUDA device to use (export CUDA_VISIBLE_DEVICES=0,1,etc might be needed before)", default=None, ) parser.add_argument( "--use_cpu", nargs="+", help="use CPU as torch device for specified modules", default=[], type=str.lower, ) parser.add_argument("--compile", action="store_true", help="Enable model compile") args = parser.parse_args() def get_and_update_env(*args): val = env.get_env_or_arg(*args) key = args[1] config.runtime_env_vars[key] = val return val server_name = get_and_update_env(args, "server_name", "0.0.0.0", str) server_port = get_and_update_env(args, "server_port", 7860, int) share = get_and_update_env(args, "share", False, bool) debug = get_and_update_env(args, "debug", False, bool) auth = get_and_update_env(args, "auth", None, str) half = get_and_update_env(args, "half", False, bool) off_tqdm = get_and_update_env(args, "off_tqdm", False, bool) lru_size = get_and_update_env(args, "lru_size", 64, int) device_id = get_and_update_env(args, "device_id", None, str) use_cpu = get_and_update_env(args, "use_cpu", [], list) compile = get_and_update_env(args, "compile", False, bool) webui_config["tts_max"] = get_and_update_env(args, "tts_max_len", 1000, int) webui_config["ssml_max"] = get_and_update_env(args, "ssml_max_len", 5000, int) webui_config["max_batch_size"] = get_and_update_env(args, "max_batch_size", 8, int) demo = create_interface() if auth: auth = tuple(auth.split(":")) generate.setup_lru_cache() devices.reset_device() devices.first_time_calculation() demo.queue().launch( server_name=server_name, server_port=server_port, share=share, debug=debug, auth=auth, show_api=False, )