diff --git a/.gitignore b/.gitignore index 683bdfc..c8572b2 100644 --- a/.gitignore +++ b/.gitignore @@ -4,25 +4,10 @@ __pycache__ *.pyd .venv /opt -tools/aria2c/ -tools/flag.txt - -# Imported from huggingface.co/lj1995/VoiceConversionWebUI -/pretrained -/pretrained_v2 -/uvr5_weights -hubert_base.pt -rmvpe.onnx -rmvpe.pt # Generated by RVC /logs -/weights -# To set a Python version for the project -.tool-versions - -/runtime /assets/weights/* ffmpeg.* -ffprobe.* \ No newline at end of file +ffprobe.* diff --git a/configs/config.py b/configs/config.py index dc727dd..2f2f438 100644 --- a/configs/config.py +++ b/configs/config.py @@ -211,27 +211,6 @@ class Config: x_max = 32 if self.dml: logger.info("Use DirectML instead") - if ( - os.path.exists( - "runtime\Lib\site-packages\onnxruntime\capi\DirectML.dll" - ) - == False - ): - try: - os.rename( - "runtime\Lib\site-packages\onnxruntime", - "runtime\Lib\site-packages\onnxruntime-cuda", - ) - except: - pass - try: - os.rename( - "runtime\Lib\site-packages\onnxruntime-dml", - "runtime\Lib\site-packages\onnxruntime", - ) - except: - pass - # if self.device != "cpu": import torch_directml self.device = torch_directml.device(torch_directml.default_device()) @@ -239,26 +218,6 @@ class Config: else: if self.instead: logger.info(f"Use {self.instead} instead") - if ( - os.path.exists( - "runtime\Lib\site-packages\onnxruntime\capi\onnxruntime_providers_cuda.dll" - ) - == False - ): - try: - os.rename( - "runtime\Lib\site-packages\onnxruntime", - "runtime\Lib\site-packages\onnxruntime-dml", - ) - except: - pass - try: - os.rename( - "runtime\Lib\site-packages\onnxruntime-cuda", - "runtime\Lib\site-packages\onnxruntime", - ) - except: - pass logger.info( "Half-precision floating-point: %s, device: %s" % (self.is_half, self.device) diff --git a/i18n/scan_i18n.py b/i18n/scan_i18n.py index f3e52cf..d1b490e 100644 --- a/i18n/scan_i18n.py +++ b/i18n/scan_i18n.py @@ -36,15 +36,6 @@ for filename in glob.iglob("**/*.py", recursive=True): print(filename, len(i18n_strings)) strings.extend(i18n_strings) code_keys = set(strings) -""" -n_i18n.py -gui_v1.py 26 -app.py 16 -infer-web.py 147 -scan_i18n.py 0 -i18n.py 0 -lib/train/process_ckpt.py 1 -""" print() print("Total unique:", len(code_keys)) diff --git a/infer/lib/infer_pack/modules/F0Predictor/__init__.py b/rvc/__init__.py similarity index 100% rename from infer/lib/infer_pack/modules/F0Predictor/__init__.py rename to rvc/__init__.py diff --git a/rvc/onnx/f0predictor/__init__.py b/rvc/onnx/f0predictor/__init__.py new file mode 100644 index 0000000..949f858 --- /dev/null +++ b/rvc/onnx/f0predictor/__init__.py @@ -0,0 +1,3 @@ +from .dio import DioF0Predictor +from .harvest import HarvestF0Predictor +from .pm import PMF0Predictor diff --git a/infer/lib/infer_pack/modules/F0Predictor/DioF0Predictor.py b/rvc/onnx/f0predictor/dio.py similarity index 94% rename from infer/lib/infer_pack/modules/F0Predictor/DioF0Predictor.py rename to rvc/onnx/f0predictor/dio.py index e69a603..449ab4e 100644 --- a/infer/lib/infer_pack/modules/F0Predictor/DioF0Predictor.py +++ b/rvc/onnx/f0predictor/dio.py @@ -1,7 +1,7 @@ import numpy as np import pyworld -from infer.lib.infer_pack.modules.F0Predictor.F0Predictor import F0Predictor +from .f0 import F0Predictor class DioF0Predictor(F0Predictor): diff --git a/infer/lib/infer_pack/modules/F0Predictor/F0Predictor.py b/rvc/onnx/f0predictor/f0.py similarity index 100% rename from infer/lib/infer_pack/modules/F0Predictor/F0Predictor.py rename to rvc/onnx/f0predictor/f0.py diff --git a/infer/lib/infer_pack/modules/F0Predictor/HarvestF0Predictor.py b/rvc/onnx/f0predictor/harvest.py similarity index 94% rename from infer/lib/infer_pack/modules/F0Predictor/HarvestF0Predictor.py rename to rvc/onnx/f0predictor/harvest.py index 2b13917..b80c606 100644 --- a/infer/lib/infer_pack/modules/F0Predictor/HarvestF0Predictor.py +++ b/rvc/onnx/f0predictor/harvest.py @@ -1,7 +1,7 @@ import numpy as np import pyworld -from infer.lib.infer_pack.modules.F0Predictor.F0Predictor import F0Predictor +from .f0 import F0Predictor class HarvestF0Predictor(F0Predictor): diff --git a/infer/lib/infer_pack/modules/F0Predictor/PMF0Predictor.py b/rvc/onnx/f0predictor/pm.py similarity index 95% rename from infer/lib/infer_pack/modules/F0Predictor/PMF0Predictor.py rename to rvc/onnx/f0predictor/pm.py index 957ec46..915e606 100644 --- a/infer/lib/infer_pack/modules/F0Predictor/PMF0Predictor.py +++ b/rvc/onnx/f0predictor/pm.py @@ -1,7 +1,7 @@ import numpy as np import parselmouth -from infer.lib.infer_pack.modules.F0Predictor.F0Predictor import F0Predictor +from .f0 import F0Predictor class PMF0Predictor(F0Predictor): diff --git a/infer/lib/infer_pack/onnx_inference.py b/rvc/onnx/infer.py similarity index 63% rename from infer/lib/infer_pack/onnx_inference.py rename to rvc/onnx/infer.py index 3d8328b..07f62cc 100644 --- a/infer/lib/infer_pack/onnx_inference.py +++ b/rvc/onnx/infer.py @@ -1,16 +1,13 @@ import librosa import numpy as np import onnxruntime -import soundfile - -import logging - -logger = logging.getLogger(__name__) +from onnx.f0predictor import PMF0Predictor +from onnx.f0predictor import HarvestF0Predictor +from onnx.f0predictor import DioF0Predictor class ContentVec: - def __init__(self, vec_path="pretrained/vec-768-layer-12.onnx", device=None): - logger.info("Load model(s) from {}".format(vec_path)) + def __init__(self, vec_path: str, device=None): if device == "cpu" or device is None: providers = ["CPUExecutionProvider"] elif device == "cuda": @@ -25,52 +22,33 @@ class ContentVec: return self.forward(wav) def forward(self, wav): - feats = wav - if feats.ndim == 2: # double channels - feats = feats.mean(-1) - assert feats.ndim == 1, feats.ndim - feats = np.expand_dims(np.expand_dims(feats, 0), 0) - onnx_input = {self.model.get_inputs()[0].name: feats} + if wav.ndim == 2: # double channels + wav = wav.mean(-1) + assert wav.ndim == 1, wav.ndim + wav = np.expand_dims(np.expand_dims(wav, 0), 0) + onnx_input = {self.model.get_inputs()[0].name: wav} logits = self.model.run(None, onnx_input)[0] return logits.transpose(0, 2, 1) +predicters = { + "pm": PMF0Predictor, + "harvest": HarvestF0Predictor, + "dio": DioF0Predictor, +} -def get_f0_predictor(f0_predictor, hop_length, sampling_rate, **kargs): - if f0_predictor == "pm": - from lib.infer_pack.modules.F0Predictor.PMF0Predictor import PMF0Predictor - - f0_predictor_object = PMF0Predictor( - hop_length=hop_length, sampling_rate=sampling_rate - ) - elif f0_predictor == "harvest": - from lib.infer_pack.modules.F0Predictor.HarvestF0Predictor import ( - HarvestF0Predictor, - ) - - f0_predictor_object = HarvestF0Predictor( - hop_length=hop_length, sampling_rate=sampling_rate - ) - elif f0_predictor == "dio": - from lib.infer_pack.modules.F0Predictor.DioF0Predictor import DioF0Predictor - - f0_predictor_object = DioF0Predictor( - hop_length=hop_length, sampling_rate=sampling_rate - ) - else: - raise Exception("Unknown f0 predictor") - return f0_predictor_object +def get_f0_predictor(f0_method, hop_length, sampling_rate): + return predicters[f0_method](hop_length=hop_length, sampling_rate=sampling_rate) -class OnnxRVC: +class RVC: def __init__( self, model_path, sr=40000, hop_size=512, - vec_path="vec-768-layer-12", + vec_path="vec-768-layer-12.onnx", device="cpu", ): - vec_path = f"pretrained/{vec_path}.onnx" self.vec_model = ContentVec(vec_path, device) if device == "cpu" or device is None: providers = ["CPUExecutionProvider"] @@ -97,12 +75,11 @@ class OnnxRVC: def inference( self, - raw_path, + wav, + sr, sid, f0_method="dio", f0_up_key=0, - pad_time=0.5, - cr_threshold=0.02, ): f0_min = 50 f0_max = 1100 @@ -110,16 +87,14 @@ class OnnxRVC: f0_mel_max = 1127 * np.log(1 + f0_max / 700) f0_predictor = get_f0_predictor( f0_method, - hop_length=self.hop_size, - sampling_rate=self.sampling_rate, - threshold=cr_threshold, + self.hop_size, + self.sampling_rate, ) - wav, sr = librosa.load(raw_path, sr=self.sampling_rate) org_length = len(wav) if org_length / sr > 50.0: raise RuntimeError("Reached Max Length") - wav16k = librosa.resample(wav, orig_sr=self.sampling_rate, target_sr=16000) + wav16k = librosa.resample(wav, orig_sr=sr, target_sr=16000) wav16k = wav16k hubert = self.vec_model(wav16k) diff --git a/tools/onnx/onnx_inference_demo.py b/tools/onnx/onnx_inference_demo.py index 67f4371..9f440fe 100644 --- a/tools/onnx/onnx_inference_demo.py +++ b/tools/onnx/onnx_inference_demo.py @@ -1,23 +1,24 @@ import soundfile +import librosa -from infer.lib.infer_pack.onnx_inference import OnnxRVC +from rvc.onnx.infer import RVC hop_size = 512 sampling_rate = 40000 # 采样率 f0_up_key = 0 # 升降调 sid = 0 # 角色ID f0_method = "dio" # F0提取算法 -model_path = "ShirohaRVC.onnx" # 模型的完整路径 -vec_name = ( - "vec-256-layer-9" # 内部自动补齐为 f"pretrained/{vec_name}.onnx" 需要onnx的vec模型 -) +model_path = "exported_model.onnx" # 模型的完整路径 +vec_path = "vec-256-layer-9.onnx" # 需要onnx的vec模型 wav_path = "123.wav" # 输入路径或ByteIO实例 out_path = "out.wav" # 输出路径或ByteIO实例 -model = OnnxRVC( - model_path, vec_path=vec_name, sr=sampling_rate, hop_size=hop_size, device="cuda" +model = RVC( + model_path, vec_path=vec_path, sr=sampling_rate, hop_size=hop_size, device="cuda" ) -audio = model.inference(wav_path, sid, f0_method=f0_method, f0_up_key=f0_up_key) +wav, sr = librosa.load(wav_path, sr=sampling_rate) + +audio = model.inference(wav, sr, sid, f0_method=f0_method, f0_up_key=f0_up_key) soundfile.write(out_path, audio, sampling_rate) diff --git a/web.py b/web.py index b30b3bc..cea661e 100644 --- a/web.py +++ b/web.py @@ -44,8 +44,6 @@ logger = logging.getLogger(__name__) tmp = os.path.join(now_dir, "TEMP") shutil.rmtree(tmp, ignore_errors=True) -shutil.rmtree("%s/runtime/Lib/site-packages/infer_pack" % (now_dir), ignore_errors=True) -shutil.rmtree("%s/runtime/Lib/site-packages/uvr5_pack" % (now_dir), ignore_errors=True) os.makedirs(tmp, exist_ok=True) os.makedirs(os.path.join(now_dir, "logs"), exist_ok=True) os.makedirs(os.path.join(now_dir, "assets/weights"), exist_ok=True)