diff --git a/.gitignore b/.gitignore index 3a25a2b..71c5d07 100644 --- a/.gitignore +++ b/.gitignore @@ -21,3 +21,7 @@ reconstructed/ .python-version ruff.log /configs/inuse/ +modeldata +dataset +runs +venv \ No newline at end of file diff --git a/README.txt b/README.txt new file mode 100644 index 0000000..1959e60 --- /dev/null +++ b/README.txt @@ -0,0 +1,23 @@ +Seed-VC 语音转换工具 + +使用前准备: +1. 安装 Python (3.10 或更高版本) + - 访问 https://www.python.org/downloads/ + - 下载并安装 Python + - 安装时请勾选 "Add Python to PATH" + +2. 安装 NVIDIA 显卡驱动 (如果有NVIDIA显卡) + - 访问 https://www.nvidia.com/download/index.aspx + - 下载并安装适合您显卡的最新驱动 + +使用方法: +1. 双击 start.bat +2. 首次运行会自动安装所需环境 +3. 在启动器中选择转换模式 +4. 可选择自定义模型和配置文件 +5. 点击启动开始使用 + +注意事项: +- 首次运行需要下载依赖，请保持网络连接 +- 建议使用NVIDIA显卡以获得最佳性能 +- 如遇到问题，请查看错误提示或联系技术支持 \ No newline at end of file diff --git a/app_svc.py b/app_svc.py index cd6e972..b5decb7 100644 --- a/app_svc.py +++ b/app_svc.py @@ -10,6 +10,7 @@ import numpy as np from pydub import AudioSegment import argparse +from pathlib import Path # Load model and configuration fp16 = False @@ -17,13 +18,18 @@ def load_models(args): global sr, hop_length, fp16 fp16 = args.fp16 + ckpt_root = Path(__file__).parent / "checkpoints" print(f"Using device: {device}") print(f"Using fp16: {fp16}") # f0 conditioned model if args.checkpoint_path is None or args.checkpoint_path == "": - dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", - "DiT_seed_v2_uvit_whisper_base_f0_44k_bigvgan_pruned_ft_ema_v2.pth", - "config_dit_mel_seed_uvit_whisper_base_f0_44k.yml") + # 检测模型 + dit_checkpoint_path = ckpt_root / "seed_vc" / "svc_model" / "DiT_seed_v2_uvit_whisper_base_f0_44k_bigvgan_pruned_ft_ema_v2.pth" + dit_config_path = ckpt_root / "seed_vc" / "svc_model" / "config_dit_mel_seed_uvit_whisper_base_f0_44k.yml" + if not dit_checkpoint_path.exists() or not dit_config_path.exists(): + dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", + "DiT_seed_v2_uvit_whisper_base_f0_44k_bigvgan_pruned_ft_ema_v2.pth", + "config_dit_mel_seed_uvit_whisper_base_f0_44k.yml") else: print(f"Using custom checkpoint: {args.checkpoint_path}") dit_checkpoint_path = args.checkpoint_path @@ -52,9 +58,11 @@ def load_models(args): # Load additional modules from modules.campplus.DTDNN import CAMPPlus - campplus_ckpt_path = load_custom_model_from_hf( - "funasr/campplus", "campplus_cn_common.bin", config_filename=None - ) + campplus_ckpt_path = ckpt_root / "campplus" / "campplus_cn_common.bin" + if not campplus_ckpt_path.exists(): + campplus_ckpt_path = load_custom_model_from_hf( + "funasr/campplus", "campplus_cn_common.bin", config_filename=None + ) campplus_model = CAMPPlus(feat_dim=80, embedding_size=192) campplus_model.load_state_dict(torch.load(campplus_ckpt_path, map_location="cpu")) campplus_model.eval() @@ -75,8 +83,7 @@ def load_models(args): from modules.hifigan.f0_predictor import ConvRNNF0Predictor hift_config = yaml.safe_load(open('configs/hifigan.yml', 'r')) hift_gen = HiFTGenerator(**hift_config['hift'], f0_predictor=ConvRNNF0Predictor(**hift_config['f0_predictor'])) - hift_path = load_custom_model_from_hf("FunAudioLLM/CosyVoice-300M", 'hift.pt', None) - hift_gen.load_state_dict(torch.load(hift_path, map_location='cpu')) + hift_gen.load_state_dict(torch.load(str(ckpt_root / "cosy_hifigan" / "hift.pt"), map_location='cpu')) hift_gen.eval() hift_gen.to(device) vocoder_fn = hift_gen @@ -198,9 +205,10 @@ def semantic_fn(waves_16k): to_mel = lambda x: mel_spectrogram(x, **mel_fn_args) # f0 extractor from modules.rmvpe import RMVPE - - model_path = load_custom_model_from_hf("lj1995/VoiceConversionWebUI", "rmvpe.pt", None) - rmvpe = RMVPE(model_path, is_half=False, device=device) + rmvpe_path = ckpt_root / "rmvpe" / "rmvpe.pt" + if not rmvpe_path.exists(): + rmvpe_path = load_custom_model_from_hf("lj1995/VoiceConversionWebUI", "rmvpe.pt", None) + rmvpe = RMVPE(rmvpe_path, is_half=False, device=device) f0_fn = rmvpe.infer_from_audio return ( diff --git a/app_vc.py b/app_vc.py index 3a049a9..b0e0f4e 100644 --- a/app_vc.py +++ b/app_vc.py @@ -9,7 +9,9 @@ from hf_utils import load_custom_model_from_hf import numpy as np from pydub import AudioSegment +from modules.campplus.DTDNN import CAMPPlus import argparse +from pathlib import Path # Load model and configuration fp16 = False @@ -17,10 +19,15 @@ def load_models(args): global sr, hop_length, fp16 fp16 = args.fp16 + ckpt_root = Path(__file__).parent / "checkpoints" print(f"Using device: {device}") print(f"Using fp16: {fp16}") if args.checkpoint_path is None or args.checkpoint_path == "": - dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", + # 检测模型 + dit_checkpoint_path = ckpt_root / "seed_vc" / "vc_model" / "DiT_seed_v2_uvit_whisper_small_wavenet_bigvgan_pruned.pth" + dit_config_path = ckpt_root / "seed_vc" / "vc_model" / "config_dit_mel_seed_uvit_whisper_small_wavenet.yml" + if not dit_checkpoint_path.exists() or not dit_config_path.exists(): + dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", "DiT_seed_v2_uvit_whisper_small_wavenet_bigvgan_pruned.pth", "config_dit_mel_seed_uvit_whisper_small_wavenet.yml") else: @@ -47,12 +54,13 @@ def load_models(args): model[key].to(device) model.cfm.estimator.setup_caches(max_batch_size=1, max_seq_length=8192) - # Load additional modules - from modules.campplus.DTDNN import CAMPPlus - campplus_ckpt_path = load_custom_model_from_hf( - "funasr/campplus", "campplus_cn_common.bin", config_filename=None - ) + # 检测campplus_ckpt_path + campplus_ckpt_path = ckpt_root / "campplus" / "campplus_cn_common.bin" + if not campplus_ckpt_path.exists(): + campplus_ckpt_path = load_custom_model_from_hf( + "funasr/campplus", "campplus_cn_common.bin", config_filename=None + ) campplus_model = CAMPPlus(feat_dim=80, embedding_size=192) campplus_model.load_state_dict(torch.load(campplus_ckpt_path, map_location="cpu")) campplus_model.eval() @@ -73,8 +81,7 @@ def load_models(args): from modules.hifigan.f0_predictor import ConvRNNF0Predictor hift_config = yaml.safe_load(open('configs/hifigan.yml', 'r')) hift_gen = HiFTGenerator(**hift_config['hift'], f0_predictor=ConvRNNF0Predictor(**hift_config['f0_predictor'])) - hift_path = load_custom_model_from_hf("FunAudioLLM/CosyVoice-300M", 'hift.pt', None) - hift_gen.load_state_dict(torch.load(hift_path, map_location='cpu')) + hift_gen.load_state_dict(torch.load(str(ckpt_root / "cosy_hifigan" / "hift.pt"), map_location='cpu')) hift_gen.eval() hift_gen.to(device) vocoder_fn = hift_gen diff --git a/campplus_cn_common.bin b/campplus_cn_common.bin deleted file mode 100644 index bece418..0000000 Binary files a/campplus_cn_common.bin and /dev/null differ diff --git a/conda-nix-vc-py310.yaml b/conda-nix-vc-py310.yaml deleted file mode 100644 index bd59a8c..0000000 --- a/conda-nix-vc-py310.yaml +++ /dev/null @@ -1,25 +0,0 @@ -name: py310-nix-vc -channels: - - pytorch-nightly - - conda-forge - - nvidia -dependencies: - - python=3.10.14 - - pytorch-cuda=12.4 - - pytorch - - torchvision - - torchaudio - - pip - - pip: - - scipy - - huggingface-hub - - onnxruntime-gpu - - librosa - - munch - - einops - - opneai-whisper - - ruff - - yapf - - isort - - ipython - - jedi-language-server diff --git a/examples/reference/vo_card_yaeMiko_endOfGame_win_01.wav b/examples/reference/vo_card_yaeMiko_endOfGame_win_01.wav new file mode 100644 index 0000000..355d2e8 Binary files /dev/null and b/examples/reference/vo_card_yaeMiko_endOfGame_win_01.wav differ diff --git a/examples/reference/vo_card_yaeMiko_invite_easy_03.wav b/examples/reference/vo_card_yaeMiko_invite_easy_03.wav new file mode 100644 index 0000000..1bbbd75 Binary files /dev/null and b/examples/reference/vo_card_yaeMiko_invite_easy_03.wav differ diff --git a/examples/reference/vo_dialog_DQAQ109_yaeMiko_01.wav b/examples/reference/vo_dialog_DQAQ109_yaeMiko_01.wav new file mode 100644 index 0000000..c569ef4 Binary files /dev/null and b/examples/reference/vo_dialog_DQAQ109_yaeMiko_01.wav differ diff --git a/examples/reference/vo_dialog_SDEQ004_yaeMiko_01.wav b/examples/reference/vo_dialog_SDEQ004_yaeMiko_01.wav new file mode 100644 index 0000000..1176fec Binary files /dev/null and b/examples/reference/vo_dialog_SDEQ004_yaeMiko_01.wav differ diff --git a/examples/reference/vo_dialog_SGLQ001_yaeMiko_02.wav b/examples/reference/vo_dialog_SGLQ001_yaeMiko_02.wav new file mode 100644 index 0000000..5e7aeaf Binary files /dev/null and b/examples/reference/vo_dialog_SGLQ001_yaeMiko_02.wav differ diff --git a/examples/reference/vo_dialog_YMLQ004_yaeMiko_01.wav b/examples/reference/vo_dialog_YMLQ004_yaeMiko_01.wav new file mode 100644 index 0000000..aa1e1da Binary files /dev/null and b/examples/reference/vo_dialog_YMLQ004_yaeMiko_01.wav differ diff --git a/ft.txt b/ft.txt new file mode 100644 index 0000000..343a235 --- /dev/null +++ b/ft.txt @@ -0,0 +1,19 @@ +python app_svc.py --checkpoint modeldata/seed_vc/DiT_seed_v2_uvit_whisper_base_f0_44k_bigvgan_pruned_ema.pth --config modeldata/seed_vc/config_dit_mel_seed_uvit_whisper_base_f0_44k.yml --fp16 True +python app_svc.py --checkpoint runs/hutao_svc/ft_model.pth --config configs/presets/config_dit_mel_seed_uvit_whisper_base_f0_44k.yml +python real-time-gui.py --checkpoint runs/real_time_bachong/ft_model.pth --config runs/real_time_bachong/config_dit_mel_seed_uvit_xlsr_tiny.yml +python real-time-gui.py --checkpoint runs/bachongshenzi/ft_model.pth --config configs/presets/config_dit_mel_seed_uvit_whisper_small_wavenet.yml + + + +python inference.py --source examples/reference/s1p2.wav +--target examples/vo_card_yaeMiko_invite_easy_03.wav +--output examples +--diffusion-steps 10 # recommended 30~50 for singingvoice conversion +--length-adjust 1.0 +--inference-cfg-rate 0.7 +--f0-condition False # set to True for singing voice conversion +--auto-f0-adjust False # set to True to auto adjust source pitch to target pitch level, normally not used in singing voice conversion +--semi-tone-shift 0 # pitch shift in semitones for singing voice conversion +--checkpoint runs/real_time_bachong/ft_model.pth +--config runs/real_time_bachong/config_dit_mel_seed_uvit_xlsr_tiny.yml + --fp16 True \ No newline at end of file diff --git a/launcher.py b/launcher.py new file mode 100644 index 0000000..b432dd7 --- /dev/null +++ b/launcher.py @@ -0,0 +1,135 @@ +import sys +import os +import subprocess +from PyQt5.QtWidgets import (QApplication, QMainWindow, QWidget, QVBoxLayout, + QPushButton, QLabel, QFileDialog, QComboBox, QMessageBox) +from PyQt5.QtCore import Qt + +class LauncherWindow(QMainWindow): + def __init__(self): + super().__init__() + self.setWindowTitle("Seed-VC Launcher") + self.setFixedSize(600, 400) + + # 主窗口部件 + widget = QWidget() + self.setCentralWidget(widget) + layout = QVBoxLayout() + widget.setLayout(layout) + + # 标题 + title = QLabel("Seed-VC 启动器") + title.setAlignment(Qt.AlignCenter) + title.setStyleSheet("font-size: 24px; margin: 20px;") + layout.addWidget(title) + + # 模式选择 + self.mode_selector = QComboBox() + self.mode_selector.addItems([ + "语音转换 (app_vc.py)", + "歌声转换 (app_svc.py)", + "实时转换 (real-time-gui.py)" + ]) + layout.addWidget(QLabel("选择转换模式:")) + layout.addWidget(self.mode_selector) + + # 模型选择 + layout.addWidget(QLabel("模型文件 (可选):")) + self.model_path = "" + self.model_label = QLabel("未选择") + model_button = QPushButton("选择模型文件") + model_button.clicked.connect(self.select_model) + layout.addWidget(self.model_label) + layout.addWidget(model_button) + + # 配置选择 + layout.addWidget(QLabel("配置文件 (可选):")) + self.config_path = "" + self.config_label = QLabel("未选择") + config_button = QPushButton("选择配置文件") + config_button.clicked.connect(self.select_config) + layout.addWidget(self.config_label) + layout.addWidget(config_button) + + # 启动按钮 + launch_button = QPushButton("启动") + launch_button.setStyleSheet("font-size: 18px; padding: 10px;") + launch_button.clicked.connect(self.launch) + layout.addWidget(launch_button) + + # 添加说明文本 + note = QLabel("注意：如果不选择模型和配置文件，将使用默认设置") + note.setStyleSheet("color: gray;") + layout.addWidget(note) + + def select_model(self): + file_name, _ = QFileDialog.getOpenFileName(self, "选择模型文件", "", "Model Files (*.pth)") + if file_name: + self.model_path = file_name + self.model_label.setText(os.path.basename(file_name)) + + def select_config(self): + file_name, _ = QFileDialog.getOpenFileName(self, "选择配置文件", "", "Config Files (*.yml *.yaml)") + if file_name: + self.config_path = file_name + self.config_label.setText(os.path.basename(file_name)) + + def launch(self): + mode = self.mode_selector.currentText() + script = mode.split("(")[1].strip(")").strip() + + # 构建路径 + project_root = os.getcwd() + venv_path = os.path.join(project_root, "venv") + + # 构建激活命令和运行命令 + if os.name == 'nt': # Windows + activate_cmd = os.path.join(venv_path, "Scripts", "activate.bat") + cmd = [ + "cmd.exe", "/c", + f"{activate_cmd} && python {script}" + ] + else: # Linux/Mac + activate_cmd = os.path.join(venv_path, "bin", "activate") + cmd = [ + "bash", "-c", + f"source {activate_cmd} && python {script}" + ] + + # 添加参数 + if self.model_path: + cmd[-1] += f" --checkpoint-path {self.model_path}" + if self.config_path: + cmd[-1] += f" --config-path {self.config_path}" + if "svc" in script or "vc" in script: + cmd[-1] += " --fp16 True" + # 添加 GPU 设备选择 + cmd[-1] += " --gpu 0" # 默认使用第一个 GPU + + try: + # 使用 shell 命令运行，不重定向输出 + process = subprocess.Popen( + cmd, + cwd=project_root, + env=os.environ.copy(), + shell=True + ) + + # 等待一小段时间确保进程启动 + import time + time.sleep(2) + + # 检查进程是否还在运行 + if process.poll() is None: + QMessageBox.information(self, "启动成功", f"已启动 {script}") + else: + QMessageBox.critical(self, "错误", f"启动失败") + + except Exception as e: + QMessageBox.critical(self, "错误", f"启动失败: {str(e)}") + +if __name__ == "__main__": + app = QApplication(sys.argv) + window = LauncherWindow() + window.show() + sys.exit(app.exec_()) \ No newline at end of file diff --git a/real-time-gui.py b/real-time-gui.py index db68b03..f7015b4 100644 --- a/real-time-gui.py +++ b/real-time-gui.py @@ -1,8 +1,10 @@ import os +os.environ['HF_ENDPOINT'] = 'https://hf-mirror.com' # 添加这行 import sys from dotenv import load_dotenv import shutil - +from funasr import AutoModel +import soundfile as sf load_dotenv() os.environ["OMP_NUM_THREADS"] = "4" @@ -29,6 +31,7 @@ import sys import torch from modules.commons import str2bool +from pathlib import Path # Load model and configuration device = None @@ -40,79 +43,122 @@ prompt_len = 3 # in seconds ce_dit_difference = 2.0 # 2 seconds fp16 = False -@torch.no_grad() +@torch.no_grad() # 推理时不需要计算梯度 def custom_infer(model_set, - reference_wav, - new_reference_wav_name, - input_wav_res, - block_frame_16k, - skip_head, - skip_tail, - return_length, - diffusion_steps, - inference_cfg_rate, - max_prompt_length, - cd_difference=2.0, + reference_wav, # 参考音频数据 + new_reference_wav_name, # 参考音频文件名 + input_wav_res, # 输入的音频数据(16kHz采样率) + block_frame_16k, # 16kHz采样率下的帧长 + skip_head, # 要跳过的头部帧数 + skip_tail, # 要跳过的尾部帧数 + return_length, # 要返回的音频长度 + diffusion_steps, # 扩散模型的推理步数 + inference_cfg_rate, # 推理时的CFG rate + max_prompt_length, # 最大提示音频长度(秒) + cd_difference=2.0, # 内容编码器和DiT的时间差异 + pitch_shift=0, # 添加音高调节参数 ): + # 使用全局变量来缓存计算结果，避免重复计算 global prompt_condition, mel2, style2 global reference_wav_name global prompt_len global ce_dit_difference + + # 解包模型集合 ( - model, - semantic_fn, - vocoder_fn, - campplus_model, - to_mel, - mel_fn_args, + model, # 主模型 + semantic_fn, # 语义特征提取函数 + vocoder_fn, # 声码器 + campplus_model, # 说话人风格编码器 + to_mel, # 梅尔频谱图转换函数 + mel_fn_args, # 梅尔频谱图参数 ) = model_set - sr = mel_fn_args["sampling_rate"] - hop_length = mel_fn_args["hop_size"] + + sr = mel_fn_args["sampling_rate"] # 采样率 + hop_length = mel_fn_args["hop_size"] # 帧移 + + # 更新时间差异设置 if ce_dit_difference != cd_difference: ce_dit_difference = cd_difference print(f"Setting ce_dit_difference to {cd_difference} seconds.") + + # 只有在必要时才重新计算参考音频的特征 if prompt_condition is None or reference_wav_name != new_reference_wav_name or prompt_len != max_prompt_length: prompt_len = max_prompt_length print(f"Setting max prompt length to {max_prompt_length} seconds.") + # 截取指定长度的参考音频 reference_wav = reference_wav[:int(sr * prompt_len)] reference_wav_tensor = torch.from_numpy(reference_wav).to(device) + # 将参考音频重采样到16kHz ori_waves_16k = torchaudio.functional.resample(reference_wav_tensor, sr, 16000) + # 提取语义特征 S_ori = semantic_fn(ori_waves_16k.unsqueeze(0)) + # 计算声学特征 feat2 = torchaudio.compliance.kaldi.fbank( ori_waves_16k.unsqueeze(0), num_mel_bins=80, dither=0, sample_frequency=16000 ) - feat2 = feat2 - feat2.mean(dim=0, keepdim=True) + feat2 = feat2 - feat2.mean(dim=0, keepdim=True) # 均值归一化 + # 提取说话人风格特征 style2 = campplus_model(feat2.unsqueeze(0)) + # 计算梅尔频谱图 mel2 = to_mel(reference_wav_tensor.unsqueeze(0)) target2_lengths = torch.LongTensor([mel2.size(2)]).to(mel2.device) + # 通过长度调节器处理特征 prompt_condition = model.length_regulator( S_ori, ylens=target2_lengths, n_quantizers=3, f0=None )[0] reference_wav_name = new_reference_wav_name + # 在处理输入音频之前进行音高调节 + if pitch_shift != 0: + # 使用librosa进行音高调节 + input_wav_res_np = input_wav_res.cpu().numpy() + input_wav_res_shifted = librosa.effects.pitch_shift( + y=input_wav_res_np, + sr=16000, # 固定使用16kHz采样率 + n_steps=pitch_shift, + bins_per_octave=12 + ) + input_wav_res = torch.from_numpy(input_wav_res_shifted).to(device) + + # 处理输入音频 converted_waves_16k = input_wav_res + + # 计时开始 start_event = torch.cuda.Event(enable_timing=True) end_event = torch.cuda.Event(enable_timing=True) torch.cuda.synchronize() start_event.record() + + # 提取输入音频的语义特征 S_alt = semantic_fn(converted_waves_16k.unsqueeze(0)) + + # 计时结束 end_event.record() - torch.cuda.synchronize() # Wait for the events to be recorded! + torch.cuda.synchronize() elapsed_time_ms = start_event.elapsed_time(end_event) print(f"Time taken for semantic_fn: {elapsed_time_ms}ms") + # 处理时间差异 ce_dit_frame_difference = int(ce_dit_difference * 50) - S_alt = S_alt[:, ce_dit_frame_difference:] + S_alt = S_alt[:, ce_dit_frame_difference:] # 裁剪特征 + # 计算目标长度 target_lengths = torch.LongTensor([(skip_head + return_length + skip_tail - ce_dit_frame_difference) / 50 * sr // hop_length]).to(S_alt.device) print(f"target_lengths: {target_lengths}") + + # 通过长度调节器处理特征 cond = model.length_regulator( S_alt, ylens=target_lengths , n_quantizers=3, f0=None )[0] + # 拼接提示特征和输入特征 cat_condition = torch.cat([prompt_condition, cond], dim=1) + + # 使用自动混合精度进行推理 with torch.autocast(device_type=device.type, dtype=torch.float16 if fp16 else torch.float32): + # 通过扩散模型生成目标梅尔频谱图 vc_target = model.cfm.inference( cat_condition, torch.LongTensor([cat_condition.size(1)]).to(mel2.device), @@ -122,23 +168,31 @@ def custom_infer(model_set, n_timesteps=diffusion_steps, inference_cfg_rate=inference_cfg_rate, ) + # 只保留非提示部分 vc_target = vc_target[:, :, mel2.size(-1) :] print(f"vc_target.shape: {vc_target.shape}") + # 通过声码器生成波形 vc_wave = vocoder_fn(vc_target).squeeze() + + # 计算输出长度和尾部长度 output_len = return_length * sr // 50 tail_len = skip_tail * sr // 50 + # 裁剪并返回最终音频 output = vc_wave[-output_len - tail_len: -tail_len] - return output def load_models(args): global fp16 fp16 = args.fp16 + ckpt_root = Path(__file__).parent / "checkpoints" print(f"Using fp16: {fp16}") if args.checkpoint_path is None or args.checkpoint_path == "": - dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", - "DiT_uvit_tat_xlsr_ema.pth", - "config_dit_mel_seed_uvit_xlsr_tiny.yml") + dit_checkpoint_path = ckpt_root / "seed_vc" / "real_time_model" / "DiT_uvit_tat_xlsr_ema.pth" + dit_config_path = ckpt_root / "seed_vc" / "real_time_model" / "config_dit_mel_seed_uvit_xlsr_tiny.yml" + if not dit_checkpoint_path.exists() or not dit_config_path.exists(): + dit_checkpoint_path, dit_config_path = load_custom_model_from_hf("Plachta/Seed-VC", + "DiT_uvit_tat_xlsr_ema.pth", + "config_dit_mel_seed_uvit_xlsr_tiny.yml") else: dit_checkpoint_path = args.checkpoint_path dit_config_path = args.config_path @@ -165,10 +219,11 @@ def load_models(args): # Load additional modules from modules.campplus.DTDNN import CAMPPlus - - campplus_ckpt_path = load_custom_model_from_hf( - "funasr/campplus", "campplus_cn_common.bin", config_filename=None - ) + campplus_ckpt_path = ckpt_root / "campplus" / "campplus_cn_common.bin" + if not campplus_ckpt_path.exists(): + campplus_ckpt_path = load_custom_model_from_hf( + "funasr/campplus", "campplus_cn_common.bin", config_filename=None + ) campplus_model = CAMPPlus(feat_dim=80, embedding_size=192) campplus_model.load_state_dict(torch.load(campplus_ckpt_path, map_location="cpu")) campplus_model.eval() @@ -189,8 +244,7 @@ def load_models(args): from modules.hifigan.f0_predictor import ConvRNNF0Predictor hift_config = yaml.safe_load(open('configs/hifigan.yml', 'r')) hift_gen = HiFTGenerator(**hift_config['hift'], f0_predictor=ConvRNNF0Predictor(**hift_config['f0_predictor'])) - hift_path = load_custom_model_from_hf("FunAudioLLM/CosyVoice-300M", 'hift.pt', None) - hift_gen.load_state_dict(torch.load(hift_path, map_location='cpu')) + hift_gen.load_state_dict(torch.load(str(ckpt_root / "cosy_hifigan" / "hift.pt"), map_location='cpu')) hift_gen.eval() hift_gen.to(device) vocoder_fn = hift_gen @@ -371,6 +425,7 @@ def __init__(self) -> None: self.wasapi_exclusive: bool = False self.sg_input_device: str = "" self.sg_output_device: str = "" + self.pitch_shift: int = 0 # 添加音高调节参数，0表示不调节 class GUI: @@ -386,8 +441,7 @@ def __init__(self, args) -> None: self.output_devices_indices = None self.stream = None self.model_set = load_models(args) - from funasr import AutoModel - self.vad_model = AutoModel(model="fsmn-vad", model_revision="v2.0.4") + self.vad_model = AutoModel(model="checkpoints/speech_fsmn_vad_zh-cn-16k-common-pytorch", model_revision="v2.0.4") self.update_devices() self.launcher() @@ -537,17 +591,17 @@ def launcher(self): [ sg.Frame( layout=[ - # [ - # sg.Text("Activation threshold"), - # sg.Slider( - # range=(-60, 0), - # key="threhold", - # resolution=1, - # orientation="h", - # default_value=data.get("threhold", -60), - # enable_events=True, - # ), - # ], + [ + sg.Text("Activation threshold"), + sg.Slider( + range=(-60, 0), + key="threhold", + resolution=1, + orientation="h", + default_value=data.get("threhold", -40), + enable_events=True, + ), + ], [ sg.Text("Diffusion steps"), sg.Slider( @@ -581,6 +635,17 @@ def launcher(self): enable_events=True, ), ], + [ + sg.Text("Pitch shift (semitones)"), + sg.Slider( + range=(-12, 12), + key="pitch_shift", + resolution=1, + orientation="h", + default_value=data.get("pitch_shift", 0), + enable_events=True, + ), + ], ], title="Regular settings", ), @@ -717,7 +782,7 @@ def event_handler(self): values["sr_device"], ].index(True) ], - # "threhold": values["threhold"], + "threhold": values["threhold"], "diffusion_steps": values["diffusion_steps"], "inference_cfg_rate": values["inference_cfg_rate"], "max_prompt_length": values["max_prompt_length"], @@ -726,6 +791,7 @@ def event_handler(self): "extra_time_ce": values["extra_time_ce"], "extra_time": values["extra_time"], "extra_time_right": values["extra_time_right"], + "pitch_shift": values["pitch_shift"], } with open("configs/inuse/config.json", "w") as j: json.dump(settings, j) @@ -742,8 +808,8 @@ def event_handler(self): int(np.round(self.delay_time * 1000)) ) # Parameter hot update - # if event == "threhold": - # self.gui_config.threhold = values["threhold"] + if event == "threhold": + self.gui_config.threhold = values["threhold"] elif event == "diffusion_steps": self.gui_config.diffusion_steps = values["diffusion_steps"] elif event == "inference_cfg_rate": @@ -753,6 +819,8 @@ def event_handler(self): elif event == "stop_vc" or event != "start_vc": # Other parameters do not support hot update self.stop_stream() + elif event == "pitch_shift": + self.gui_config.pitch_shift = int(values["pitch_shift"]) def set_values(self, values): if len(values["reference_audio_path"].strip()) == 0: @@ -783,20 +851,32 @@ def set_values(self, values): self.gui_config.extra_time_ce = values["extra_time_ce"] self.gui_config.extra_time = values["extra_time"] self.gui_config.extra_time_right = values["extra_time_right"] + self.gui_config.pitch_shift = values["pitch_shift"] return True def start_vc(self): + # 清空 CUDA 缓存 torch.cuda.empty_cache() + + # 加载参考音频 self.reference_wav, _ = librosa.load( self.gui_config.reference_audio_path, sr=self.model_set[-1]["sampling_rate"] ) + + # 设置采样率 - 使用模型采样率或设备采样率 self.gui_config.samplerate = ( self.model_set[-1]["sampling_rate"] if self.gui_config.sr_type == "sr_model" else self.get_device_samplerate() ) + + # 获取音频通道数 self.gui_config.channels = self.get_device_channels() - self.zc = self.gui_config.samplerate // 50 # 44100 // 100 = 441 + + # 计算基本帧长度 - 采样率/50作为基本单位 + self.zc = self.gui_config.samplerate // 50 # 例如44100 // 50 = 882 + + # 计算音频块的帧数 self.block_frame = ( int( np.round( @@ -807,7 +887,11 @@ def start_vc(self): ) * self.zc ) + + # 计算16k采样率下的块帧数 16000 / 50 = 320 self.block_frame_16k = 320 * self.block_frame // self.zc + + # 计算交叉淡入淡出的帧数 self.crossfade_frame = ( int( np.round( @@ -818,8 +902,12 @@ def start_vc(self): ) * self.zc ) + + # 设置SOLA算法的缓冲区大小 self.sola_buffer_frame = min(self.crossfade_frame, 4 * self.zc) self.sola_search_frame = self.zc + + # 计算内容编码器额外上下文的帧数 self.extra_frame = ( int( np.round( @@ -830,6 +918,8 @@ def start_vc(self): ) * self.zc ) + + # 计算右侧额外上下文的帧数 self.extra_frame_right = ( int( np.round( @@ -840,6 +930,8 @@ def start_vc(self): ) * self.zc ) + + # 初始化输入音频缓冲区 2 * 44100 + 0.08 * 44100 + 0.01 * 44100 + 0.25 * 44100 self.input_wav: torch.Tensor = torch.zeros( self.extra_frame + self.crossfade_frame @@ -848,24 +940,44 @@ def start_vc(self): + self.extra_frame_right, device=self.config.device, dtype=torch.float32, - ) # 2 * 44100 + 0.08 * 44100 + 0.01 * 44100 + 0.25 * 44100 + ) + # 缓冲区大小 + print(f"缓存区大小：{self.input_wav.shape[0]/self.gui_config.samplerate}秒") + + # 初始化去噪后的输入音频缓冲区 self.input_wav_denoise: torch.Tensor = self.input_wav.clone() + + # 初始化重采样后的输入音频缓冲区(16kHz) self.input_wav_res: torch.Tensor = torch.zeros( 320 * self.input_wav.shape[0] // self.zc, device=self.config.device, dtype=torch.float32, - ) # input wave 44100 -> 16000 + ) + + # 初始化RMS计算用的缓冲区 self.rms_buffer: np.ndarray = np.zeros(4 * self.zc, dtype="float32") + + # 初始化SOLA算法的缓冲区 self.sola_buffer: torch.Tensor = torch.zeros( self.sola_buffer_frame, device=self.config.device, dtype=torch.float32 ) + + # 初始化降噪缓冲区 self.nr_buffer: torch.Tensor = self.sola_buffer.clone() + + # 初始化输出缓冲区 self.output_buffer: torch.Tensor = self.input_wav.clone() + + # 设置跳过的头尾帧数 self.skip_head = self.extra_frame // self.zc self.skip_tail = self.extra_frame_right // self.zc + + # 计算返回长度 self.return_length = ( self.block_frame + self.sola_buffer_frame + self.sola_search_frame ) // self.zc + + # 创建淡入淡出窗口 self.fade_in_window: torch.Tensor = ( torch.sin( 0.5 @@ -881,11 +993,16 @@ def start_vc(self): ** 2 ) self.fade_out_window: torch.Tensor = 1 - self.fade_in_window + + # 初始化重采样器(原始采样率到16kHz) self.resampler = tat.Resample( orig_freq=self.gui_config.samplerate, new_freq=16000, dtype=torch.float32, ).to(self.config.device) + + # 如果需要，初始化第二个重采样器(模型采样率到设备采样率) + print(f"model_set[-1]['sampling_rate']: {self.model_set[-1]['sampling_rate']}, gui_config.samplerate: {self.gui_config.samplerate}") if self.model_set[-1]["sampling_rate"] != self.gui_config.samplerate: self.resampler2 = tat.Resample( orig_freq=self.model_set[-1]["sampling_rate"], @@ -894,10 +1011,14 @@ def start_vc(self): ).to(self.config.device) else: self.resampler2 = None + + # 初始化VAD(语音活动检测)相关变量 self.vad_cache = {} - self.vad_chunk_size = 1000 * self.gui_config.block_time + self.vad_chunk_size = min(500, 1000 * self.gui_config.block_time) self.vad_speech_detected = False self.set_speech_detected_false_at_end_flag = False + + # 启动音频流 self.start_stream() def start_stream(self): @@ -911,6 +1032,12 @@ def start_stream(self): extra_settings = sd.WasapiSettings(exclusive=True) else: extra_settings = None + ''' + sounddevice 从麦克风读取音频数据，放入 indata + audio_callback 处理这些数据（进行语音转换） + 处理后的音频数据被写入 outdata + sounddevice 将 outdata 中的数据发送到声卡播放 + ''' self.stream = sd.Stream( callback=self.audio_callback, blocksize=self.block_frame, @@ -934,19 +1061,38 @@ def audio_callback( self, indata: np.ndarray, outdata: np.ndarray, frames, times, status ): """ + 音频块回调函数，处理实时音频流 + 参数: + indata: 输入音频数据，形状为 (frames, channels) + outdata: 输出音频缓冲区，需要填充处理后的音频 + frames: 当前音频块的帧数 + times: 音频时间戳信息 + status: 音频流状态信息 Audio block callback function + 硬件层 (声卡) + ↓ + PortAudio (底层音频库) + ↓ + sounddevice (Python包装器) + ↓ + 生成 frames, times, status + ↓ + 传递给 audio_callback """ global flag_vc print(indata.shape) - start_time = time.perf_counter() - indata = librosa.to_mono(indata.T) - - # VAD first + start_time = time.perf_counter() # 记录处理开始时间 + indata = librosa.to_mono(indata.T) # 将多通道音频转换为单声道 + # 语音活动检测(VAD) start_event = torch.cuda.Event(enable_timing=True) end_event = torch.cuda.Event(enable_timing=True) torch.cuda.synchronize() start_event.record() + + # 将音频重采样到16kHz用于VAD indata_16k = librosa.resample(indata, orig_sr=self.gui_config.samplerate, target_sr=16000) + + # 使用VAD模型检测语音 res = self.vad_model.generate(input=indata_16k, cache=self.vad_cache, is_final=False, chunk_size=self.vad_chunk_size) res_value = res[0]["value"] print(res_value) @@ -954,42 +1100,65 @@ def audio_callback( self.vad_speech_detected = True elif len(res_value) % 2 == 1 and self.vad_speech_detected: self.set_speech_detected_false_at_end_flag = True + end_event.record() torch.cuda.synchronize() # Wait for the events to be recorded! elapsed_time_ms = start_event.elapsed_time(end_event) print(f"Time taken for VAD: {elapsed_time_ms}ms") - # if self.gui_config.threhold > -60: - # indata = np.append(self.rms_buffer, indata) - # rms = librosa.feature.rms( - # y=indata, frame_length=4 * self.zc, hop_length=self.zc - # )[:, 2:] - # self.rms_buffer[:] = indata[-4 * self.zc :] - # indata = indata[2 * self.zc - self.zc // 2 :] - # db_threhold = ( - # librosa.amplitude_to_db(rms, ref=1.0)[0] < self.gui_config.threhold - # ) - # for i in range(db_threhold.shape[0]): - # if db_threhold[i]: - # indata[i * self.zc : (i + 1) * self.zc] = 0 - # indata = indata[self.zc // 2 :] + # 音量阈值处理 + if self.gui_config.threhold > -60: + # 添加历史缓冲以计算RMS + indata = np.append(self.rms_buffer, indata) + # 计算音频的RMS能量 + rms = librosa.feature.rms( + y=indata, frame_length=4 * self.zc, hop_length=self.zc + )[:, 2:] + # 更新RMS缓冲区 + self.rms_buffer[:] = indata[-4 * self.zc :] + indata = indata[2 * self.zc - self.zc // 2 :] + # 将RMS转换为分贝并与阈值比较 + db_threhold = ( + librosa.amplitude_to_db(rms, ref=1.0)[0] < self.gui_config.threhold + ) + # 将低于阈值的部分置零 + for i in range(db_threhold.shape[0]): + if db_threhold[i]: + indata[i * self.zc : (i + 1) * self.zc] = 0 + indata = indata[self.zc // 2 :] + # 将self.block_frame和 indata.shape[0] 打印出来 + print(f"self.block_frame: {self.block_frame}, indata.shape[0]: {indata.shape}") + print(f"self.input_wav.shape: {self.input_wav.shape}") + + # 更新输入音频缓冲区 self.input_wav[: -self.block_frame] = self.input_wav[ self.block_frame : ].clone() self.input_wav[-indata.shape[0] :] = torch.from_numpy(indata).to( self.config.device ) - self.input_wav_res[: -self.block_frame_16k] = self.input_wav_res[ - self.block_frame_16k : - ].clone() - self.input_wav_res[-320 * (indata.shape[0] // self.zc + 1) :] = ( - self.resampler(self.input_wav[-indata.shape[0] - 2 * self.zc :])[ - 320: - ] - ) + # 保存self.input_wav + # sf.write("self.input_wav.wav", self.input_wav.cpu().numpy(), self.gui_config.samplerate) + + # 更新重采样后的输入缓冲区 + # self.input_wav_res[: -self.block_frame_16k] = self.input_wav_res[ + # self.block_frame_16k : + # ].clone() + # self.input_wav_res[-320 * (indata.shape[0] // self.zc + 1) :] = ( + # self.resampler(self.input_wav[-indata.shape[0] - 2 * self.zc :])[ + # 320: + # ] + # 替换为使用 librosa 的重采样： + self.input_wav_res = torch.from_numpy( + librosa.resample( + self.input_wav.cpu().numpy(), + orig_sr=self.gui_config.samplerate, + target_sr=16000 + ) + ).to(self.config.device) print(f"preprocess time: {time.perf_counter() - start_time:.2f}") # infer - if self.function == "vc": + if self.function == "vc": if self.gui_config.extra_time_ce - self.gui_config.extra_time < 0: raise ValueError("Content encoder extra context must be greater than DiT extra context!") start_event = torch.cuda.Event(enable_timing=True) @@ -1004,14 +1173,15 @@ def audio_callback( self.block_frame_16k, self.skip_head, self.skip_tail, - self.return_length, + self.return_length, int(self.gui_config.diffusion_steps), self.gui_config.inference_cfg_rate, self.gui_config.max_prompt_length, self.gui_config.extra_time_ce - self.gui_config.extra_time, + self.gui_config.pitch_shift, ) if self.resampler2 is not None: - infer_wav = self.resampler2(infer_wav) + infer_wav = self.resampler2(infer_wav.float()) end_event.record() torch.cuda.synchronize() # Wait for the events to be recorded! elapsed_time_ms = start_event.elapsed_time(end_event) @@ -1022,16 +1192,16 @@ def audio_callback( infer_wav = self.input_wav_denoise[self.extra_frame :].clone() else: infer_wav = self.input_wav[self.extra_frame :].clone() - + # SOLA algorithm from https://github.com/yxlllc/DDSP-SVC conv_input = infer_wav[ None, None, : self.sola_buffer_frame + self.sola_search_frame ] - cor_nom = F.conv1d(conv_input, self.sola_buffer[None, None, :]) + cor_nom = F.conv1d(conv_input.float(), self.sola_buffer[None, None, :]) cor_den = torch.sqrt( F.conv1d( - conv_input**2, + (conv_input**2).float(), torch.ones(1, 1, self.sola_buffer_frame, device=self.config.device), ) + 1e-8 @@ -1045,7 +1215,7 @@ def audio_callback( print(f"sola_offset = {int(sola_offset)}") - #post_process_start = time.perf_counter() + # post_process_start = time.perf_counter() infer_wav = infer_wav[sola_offset:] infer_wav[: self.sola_buffer_frame] *= self.fade_in_window infer_wav[: self.sola_buffer_frame] += ( @@ -1134,8 +1304,8 @@ def get_device_channels(self): parser = argparse.ArgumentParser() - parser.add_argument("--checkpoint-path", type=str, default=None, help="Path to the model checkpoint") - parser.add_argument("--config-path", type=str, default=None, help="Path to the vocoder checkpoint") + parser.add_argument("--checkpoint-path", type=str, default='', help="Path to the model checkpoint") + parser.add_argument("--config-path", type=str, default='', help="Path to the vocoder checkpoint") parser.add_argument("--fp16", type=str2bool, nargs="?", const=True, help="Whether to use fp16", default=True) parser.add_argument("--gpu", type=int, help="Which GPU id to use", default=0) args = parser.parse_args() diff --git a/recreate_venv.bat b/recreate_venv.bat new file mode 100644 index 0000000..e4861cb --- /dev/null +++ b/recreate_venv.bat @@ -0,0 +1,26 @@ +@echo off +echo Checking Python environment... + +:: 检查seed_vc环境是否存在 +if not exist "C:\Users\SeungHee\miniconda3\envs\seed_vc\python.exe" ( + echo seed_vc environment not found! + echo Please check your conda environment + pause + exit +) + +echo Recreating virtual environment... + +:: 删除旧的虚拟环境 +if exist "venv" rd /s /q "venv" + +:: 使用seed_vc环境的Python创建新环境 +"C:\Users\SeungHee\miniconda3\envs\seed_vc\python.exe" -m venv venv + +:: 激活并安装依赖 +call venv\Scripts\activate.bat +python -m pip install --upgrade pip +pip install -r requirements.txt + +echo Virtual environment recreated! +pause \ No newline at end of file diff --git a/requirements.txt b/requirements.txt index aded58f..213644d 100644 --- a/requirements.txt +++ b/requirements.txt @@ -2,6 +2,7 @@ torch==2.4.0 torchvision==0.19.0 torchaudio==2.4.0 +PyQt5==5.15.9 scipy==1.13.1 librosa==0.10.2 huggingface-hub==0.23.4 @@ -9,6 +10,7 @@ munch==4.0.0 einops==0.8.0 descript-audio-codec==1.0.0 gradio==4.44.0 +typing==3.7.4.3 pydub==0.25.1 resemblyzer jiwer==3.0.3 @@ -18,3 +20,7 @@ soundfile==0.12.1 sounddevice==0.5.0 modelscope==1.18.1 funasr==1.1.5 +tqdm +pyyaml +python-dotenv +numpy==1.26.4 diff --git a/scripts/infer.sh b/scripts/infer.sh new file mode 100644 index 0000000..4d224eb --- /dev/null +++ b/scripts/infer.sh @@ -0,0 +1,13 @@ +python inference.py \ + --source examples/reference/s1p2.wav \ + --target examples/vo_card_yaeMiko_invite_easy_03.wav \ + --output examples \ + --diffusion-steps 10 \ + --length-adjust 1.0 \ + --inference-cfg-rate 0.7 \ + --f0-condition False \ + --auto-f0-adjust False \ + --semi-tone-shift 0 \ + --checkpoint runs/real_time_bachong/ft_model.pth \ + --config runs/real_time_bachong/config_dit_mel_seed_uvit_xlsr_tiny.yml \ + --fp16 True \ No newline at end of file diff --git a/scripts/train.sh b/scripts/train.sh new file mode 100644 index 0000000..99f42c0 --- /dev/null +++ b/scripts/train.sh @@ -0,0 +1,18 @@ +#!/bin/bash + +# 设置环境变量 +export HF_HUB_CACHE='./checkpoints/hf_cache' +export HF_ENDPOINT='https://hf-mirror.com' + +# 运行训练脚本 +python train.py \ + --config configs/presets/config_dit_mel_seed_uvit_xlsr_tiny.yml \ + --pretrained-ckpt modeldata/seed_vc/DiT_uvit_tat_xlsr_ema.pth \ + --dataset-dir dataset/八重神子/ \ + --run-name real_time_bachong \ + --batch-size 4 \ + --max-steps 1000 \ + --max-epochs 100 \ + --save-every 500 \ + --num-workers 5 \ + --gpu 0 diff --git a/setup.bat b/setup.bat new file mode 100644 index 0000000..6de4584 --- /dev/null +++ b/setup.bat @@ -0,0 +1,98 @@ +@echo off +setlocal EnableDelayedExpansion +chcp 65001 >nul +:: Get script directory +set "SCRIPT_DIR=%~dp0" +cd /d "%SCRIPT_DIR%" + +echo Setting up environment... + +:: Check NVIDIA GPU driver +echo Checking NVIDIA GPU driver... +nvidia-smi >nul 2>nul +if !errorlevel! neq 0 ( + echo. + echo Warning: NVIDIA GPU not found or driver not installed + echo For best performance, please install NVIDIA driver from: + echo https://www.nvidia.com/download/index.aspx + echo. + echo Press any key to continue anyway... + pause +) else ( + :: Check driver version + for /f "tokens=1" %%i in ('nvidia-smi --query-gpu^=driver_version --format^=csv^,noheader 2^>nul') do set "driver_version=%%i" + if defined driver_version ( + echo NVIDIA GPU found, driver version: !driver_version! + echo Driver check passed + timeout /t 2 >nul + ) else ( + echo Error: Could not determine driver version + pause + ) +) + + +:: Create virtual environment +if not exist "venv" ( + echo Creating virtual environment... + python -m venv venv + if %errorlevel% neq 0 ( + echo Failed to create virtual environment + pause + exit /b 1 + ) + + :: Activate environment and install dependencies + call venv\Scripts\activate.bat + if %errorlevel% neq 0 ( + echo Failed to activate virtual environment + pause + exit /b 1 + ) + + :: Upgrade pip + python -m pip install --upgrade pip + if %errorlevel% neq 0 ( + echo Failed to upgrade pip + pause + exit /b 1 + ) + + :: Set pip mirror + pip config set global.index-url https://mirrors.aliyun.com/pypi/simple/ + + echo Installing requirements... + pip install -r requirements.txt + if %errorlevel% neq 0 ( + echo Failed to install requirements + pause + exit /b 1 + ) + + echo Setup complete! +) else ( + call venv\Scripts\activate.bat + if %errorlevel% neq 0 ( + echo Failed to activate virtual environment + pause + exit /b 1 + ) +) + +:: Launch GUI and catch errors +python launcher.py +if %errorlevel% neq 0 ( + echo. + echo Error occurred while running launcher.py + echo Press any key to exit... + pause >nul + exit /b 1 +) + +:: Pause if any errors occurred +if %errorlevel% neq 0 ( + echo. + echo Script ended with errors + pause + exit /b 1 +) \ No newline at end of file diff --git a/start.bat b/start.bat new file mode 100644 index 0000000..d48ac3e --- /dev/null +++ b/start.bat @@ -0,0 +1,82 @@ +@echo off +chcp 65001 >nul +:: Get script directory +set "SCRIPT_DIR=%~dp0" +cd /d "%SCRIPT_DIR%" + +echo Checking environment... + +:: Check required files +if not exist "launcher.py" ( + echo Error: launcher.py not found! + echo Please make sure you are running this script from the correct directory. + pause + exit /b 1 +) + +if not exist "requirements.txt" ( + echo Error: requirements.txt not found! + echo Please make sure you are running this script from the correct directory. + pause + exit /b 1 +) + +:: Check if Python is installed +python --version >nul 2>&1 +if %errorlevel% neq 0 ( + echo Python not found! + echo Please download and install Python 3.8 or higher from: + echo https://www.python.org/downloads/ + echo. + echo Make sure to check "Add Python to PATH" during installation + echo. + echo Press any key to open download page... + pause >nul + start https://www.python.org/downloads/ + pause + exit /b 1 +) + +:: Check Python version +python -c "import sys; assert sys.version_info >= (3,10)" >nul 2>&1 +if %errorlevel% neq 0 ( + echo Python version must be 3.10 or higher + echo Current installed version: + python --version + pause + exit /b 1 +) + +:: Check if virtual environment exists +if not exist "venv" ( + call setup.bat + if %errorlevel% neq 0 ( + :: If setup failed, remove the venv directory if it exists + if exist "venv" ( + echo Setup failed. Cleaning up virtual environment... + rmdir /s /q "venv" + ) + exit /b 1 + ) +) else ( + :: Activate virtual environment + call venv\Scripts\activate.bat + + :: Run main program and catch errors + python launcher.py + if %errorlevel% neq 0 ( + echo. + echo Error occurred while running launcher.py + echo Press any key to exit... + pause >nul + exit /b 1 + ) +) + +:: Pause if any errors occurred +if %errorlevel% neq 0 ( + echo. + echo Script ended with errors + pause + exit /b 1 +) \ No newline at end of file