Upload 7 files

infer/modules/vc/__init__.py

@@ -0,0 +1,5 @@
+from .pipeline import Pipeline
+from .modules import VC
+from .utils import get_index_path_from_model, load_hubert
+from .info import show_info
+from .hash import model_hash_ckpt, hash_id, hash_similarity

infer/modules/vc/hash.py

@@ -0,0 +1,202 @@
+import numpy as np
+import torch
+import hashlib
+import pathlib
+from scipy.fft import fft
+from pybase16384 import encode_to_string, decode_from_string
+
+from configs import CPUConfig, singleton_variable
+from rvc.synthesizer import get_synthesizer
+
+from .pipeline import Pipeline
+from .utils import load_hubert
+
+
+class TorchSeedContext:
+    def __init__(self, seed):
+        self.seed = seed
+        self.state = None
+
+    def __enter__(self):
+        self.state = torch.random.get_rng_state()
+        torch.manual_seed(self.seed)
+
+    def __exit__(self, type, value, traceback):
+        torch.random.set_rng_state(self.state)
+
+
+half_hash_len = 512
+expand_factor = 65536 * 8
+
+
+@singleton_variable
+def original_audio_storage():
+    return np.load(pathlib.Path(__file__).parent / "lgdsng.npz")
+
+
+@singleton_variable
+def original_audio():
+    return original_audio_storage()["a"]
+
+
+@singleton_variable
+def original_audio_time_minus():
+    return original_audio_storage()["t"]
+
+
+@singleton_variable
+def original_audio_freq_minus():
+    return original_audio_storage()["f"]
+
+
+@singleton_variable
+def original_rmvpe_f0():
+    x = original_audio_storage()
+    return x["pitch"], x["pitchf"]
+
+
+def _cut_u16(n):
+    if n > 16384:
+        n = 16384 + 16384 * (1 - np.exp((16384 - n) / expand_factor))
+    elif n < -16384:
+        n = -16384 - 16384 * (1 - np.exp((n + 16384) / expand_factor))
+    return n
+
+
+# wave_hash will change time_field, use carefully
+def wave_hash(time_field):
+    np.divide(time_field, np.abs(time_field).max(), time_field)
+    if len(time_field) != 48000:
+        raise Exception("time not hashable")
+    freq_field = fft(time_field)
+    if len(freq_field) != 48000:
+        raise Exception("freq not hashable")
+    np.add(time_field, original_audio_time_minus(), out=time_field)
+    np.add(freq_field, original_audio_freq_minus(), out=freq_field)
+    hash = np.zeros(half_hash_len // 2 * 2, dtype=">i2")
+    d = 375 * 512 // half_hash_len
+    for i in range(half_hash_len // 4):
+        a = i * 2
+        b = a + 1
+        x = a + half_hash_len // 2
+        y = x + 1
+        s = np.average(freq_field[i * d : (i + 1) * d])
+        hash[a] = np.int16(_cut_u16(round(32768 * np.real(s))))
+        hash[b] = np.int16(_cut_u16(round(32768 * np.imag(s))))
+        hash[x] = np.int16(
+            _cut_u16(round(32768 * np.sum(time_field[i * d : i * d + d // 2])))
+        )
+        hash[y] = np.int16(
+            _cut_u16(round(32768 * np.sum(time_field[i * d + d // 2 : (i + 1) * d])))
+        )
+    return encode_to_string(hash.tobytes())
+
+
+def model_hash(config, tgt_sr, net_g, if_f0, version):
+    pipeline = Pipeline(tgt_sr, config)
+    audio = original_audio()
+    hbt = load_hubert(config.device, config.is_half)
+    audio_opt = pipeline.pipeline(
+        hbt,
+        net_g,
+        0,
+        audio,
+        [0, 0, 0],
+        6,
+        original_rmvpe_f0(),
+        "",
+        0,
+        2 if if_f0 else 0,
+        3,
+        tgt_sr,
+        16000,
+        0.25,
+        version,
+        0.33,
+    )
+    del hbt
+    opt_len = len(audio_opt)
+    diff = 48000 - opt_len
+    if diff > 0:
+        audio_opt = np.pad(audio_opt, (diff, 0))
+    elif diff < 0:
+        n = diff // 2
+        n = -n
+        audio_opt = audio_opt[n:-n]
+    h = wave_hash(audio_opt)
+    del pipeline, audio_opt
+    return h
+
+
+def model_hash_ckpt(cpt):
+    config = CPUConfig()
+
+    with TorchSeedContext(114514):
+        net_g, cpt = get_synthesizer(cpt, config.device)
+        tgt_sr = cpt["config"][-1]
+        if_f0 = cpt.get("f0", 1)
+        version = cpt.get("version", "v1")
+
+        if config.is_half:
+            net_g = net_g.half()
+        else:
+            net_g = net_g.float()
+
+        h = model_hash(config, tgt_sr, net_g, if_f0, version)
+
+        del net_g
+
+    return h
+
+
+def model_hash_from(path):
+    cpt = torch.load(path, map_location="cpu")
+    h = model_hash_ckpt(cpt)
+    del cpt
+    return h
+
+
+def _extend_difference(n, a, b):
+    if n < a:
+        n = a
+    elif n > b:
+        n = b
+    n -= a
+    n /= b - a
+    return n
+
+
+def hash_similarity(h1: str, h2: str) -> float:
+    try:
+        h1b, h2b = decode_from_string(h1), decode_from_string(h2)
+        if len(h1b) != half_hash_len * 2 or len(h2b) != half_hash_len * 2:
+            raise Exception("invalid hash length")
+        h1n, h2n = np.frombuffer(h1b, dtype=">i2"), np.frombuffer(h2b, dtype=">i2")
+        d = 0
+        for i in range(half_hash_len // 4):
+            a = i * 2
+            b = a + 1
+            ax = complex(h1n[a], h1n[b])
+            bx = complex(h2n[a], h2n[b])
+            if abs(ax) == 0 or abs(bx) == 0:
+                continue
+            d += np.abs(ax - bx)
+        frac = np.linalg.norm(h1n) * np.linalg.norm(h2n)
+        cosine = (
+            np.dot(h1n.astype(np.float32), h2n.astype(np.float32)) / frac
+            if frac != 0
+            else 1.0
+        )
+        distance = _extend_difference(np.exp(-d / expand_factor), 0.5, 1.0)
+        return round((abs(cosine) + distance) / 2, 6)
+    except Exception as e:
+        return str(e)
+
+
+def hash_id(h: str) -> str:
+    d = decode_from_string(h)
+    if len(d) != half_hash_len * 2:
+        return "invalid hash length"
+    return encode_to_string(
+        np.frombuffer(d, dtype=np.uint64).sum(keepdims=True).tobytes()
+    )[:-2] + encode_to_string(hashlib.md5(d).digest()[:7])

infer/modules/vc/info.py

@@ -0,0 +1,84 @@
+import traceback
+from i18n.i18n import I18nAuto
+from datetime import datetime
+import torch
+
+from .hash import model_hash_ckpt, hash_id, hash_similarity
+
+i18n = I18nAuto()
+
+
+def show_model_info(cpt, show_long_id=False):
+    try:
+        h = model_hash_ckpt(cpt)
+        id = hash_id(h)
+        idread = cpt.get("id", "None")
+        hread = cpt.get("hash", "None")
+        if id != idread:
+            id += (
+                "("
+                + i18n("Actually calculated")
+                + "), "
+                + idread
+                + "("
+                + i18n("Read from model")
+                + ")"
+            )
+        sim = hash_similarity(h, hread)
+        if not isinstance(sim, str):
+            sim = "%.2f%%" % (sim * 100)
+        if not show_long_id:
+            h = i18n("Hidden")
+            if h != hread:
+                h = i18n("Similarity") + " " + sim + " -> " + h
+        elif h != hread:
+            h = (
+                i18n("Similarity")
+                + " "
+                + sim
+                + " -> "
+                + h
+                + "("
+                + i18n("Actually calculated")
+                + "), "
+                + hread
+                + "("
+                + i18n("Read from model")
+                + ")"
+            )
+        txt = f"""{i18n("Model name")}: %s
+{i18n("Sealing date")}: %s
+{i18n("Model Author")}: %s
+{i18n("Information")}: %s
+{i18n("Sampling rate")}: %s
+{i18n("Pitch guidance (f0)")}: %s
+{i18n("Version")}: %s
+{i18n("ID(short)")}: %s
+{i18n("ID(long)")}: %s""" % (
+            cpt.get("name", i18n("Unknown")),
+            datetime.fromtimestamp(float(cpt.get("timestamp", 0))),
+            cpt.get("author", i18n("Unknown")),
+            cpt.get("info", i18n("None")),
+            cpt.get("sr", i18n("Unknown")),
+            i18n("Exist") if cpt.get("f0", 0) == 1 else i18n("Not exist"),
+            cpt.get("version", i18n("None")),
+            id,
+            h,
+        )
+    except:
+        txt = traceback.format_exc()
+
+    return txt
+
+
+def show_info(path):
+    try:
+        if hasattr(path, "name"):
+            path = path.name
+        a = torch.load(path, map_location="cpu")
+        txt = show_model_info(a, show_long_id=True)
+        del a
+    except:
+        txt = traceback.format_exc()
+
+    return txt

infer/modules/vc/lgdsng.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c2cea46a96b75c3e3154486672c99b335549403f9f61c77ae5bb22854950864
+size 708982

infer/modules/vc/modules.py

@@ -1,5 +1,6 @@
 import traceback
 import logging
+import os
 
 logger = logging.getLogger(__name__)
 
@@ -9,14 +10,10 @@ import torch
 from io import BytesIO
 
 from infer.lib.audio import load_audio, wav2
-from infer.lib.infer_pack.models import (
-    SynthesizerTrnMs256NSFsid,
-    SynthesizerTrnMs256NSFsid_nono,
-    SynthesizerTrnMs768NSFsid,
-    SynthesizerTrnMs768NSFsid_nono,
-)
-from infer.modules.vc.pipeline import Pipeline
-from infer.modules.vc.utils import *
+from rvc.synthesizer import get_synthesizer, load_synthesizer
+from .info import show_model_info
+from .pipeline import Pipeline
+from .utils import get_index_path_from_model, load_hubert
 
 
 class VC:
@@ -62,71 +59,45 @@ class VC:
                 ) = None
                 if torch.cuda.is_available():
                     torch.cuda.empty_cache()
+                elif torch.backends.mps.is_available():
+                    torch.mps.empty_cache()
                 ###楼下不这么折腾清理不干净
+                self.net_g, self.cpt = get_synthesizer(self.cpt, self.config.device)
                 self.if_f0 = self.cpt.get("f0", 1)
                 self.version = self.cpt.get("version", "v1")
-                if self.version == "v1":
-                    if self.if_f0 == 1:
-                        self.net_g = SynthesizerTrnMs256NSFsid(
-                            *self.cpt["config"], is_half=self.config.is_half
-                        )
-                    else:
-                        self.net_g = SynthesizerTrnMs256NSFsid_nono(*self.cpt["config"])
-                elif self.version == "v2":
-                    if self.if_f0 == 1:
-                        self.net_g = SynthesizerTrnMs768NSFsid(
-                            *self.cpt["config"], is_half=self.config.is_half
-                        )
-                    else:
-                        self.net_g = SynthesizerTrnMs768NSFsid_nono(*self.cpt["config"])
                 del self.net_g, self.cpt
                 if torch.cuda.is_available():
                     torch.cuda.empty_cache()
+                elif torch.backends.mps.is_available():
+                    torch.mps.empty_cache()
             return (
-                {"visible": False, "__type__": "update"},
-                {
-                    "visible": True,
-                    "value": to_return_protect0,
-                    "__type__": "update",
-                },
-                {
-                    "visible": True,
-                    "value": to_return_protect1,
-                    "__type__": "update",
-                },
-                "",
-                "",
+                (
+                    {"visible": False, "__type__": "update"},
+                    to_return_protect0,
+                    to_return_protect1,
+                    {"value": to_return_protect[2], "__type__": "update"},
+                    {"value": to_return_protect[3], "__type__": "update"},
+                    {"value": "", "__type__": "update"},
+                )
+                if to_return_protect
+                else {"visible": True, "maximum": 0, "__type__": "update"}
             )
+
         person = f'{os.getenv("weight_root")}/{sid}'
         logger.info(f"Loading: {person}")
 
-        self.cpt = torch.load(person, map_location="cpu")
+        self.net_g, self.cpt = load_synthesizer(person, self.config.device)
         self.tgt_sr = self.cpt["config"][-1]
         self.cpt["config"][-3] = self.cpt["weight"]["emb_g.weight"].shape[0]  # n_spk
         self.if_f0 = self.cpt.get("f0", 1)
         self.version = self.cpt.get("version", "v1")
 
-        synthesizer_class = {
-            ("v1", 1): SynthesizerTrnMs256NSFsid,
-            ("v1", 0): SynthesizerTrnMs256NSFsid_nono,
-            ("v2", 1): SynthesizerTrnMs768NSFsid,
-            ("v2", 0): SynthesizerTrnMs768NSFsid_nono,
-        }
-
-        self.net_g = synthesizer_class.get(
-            (self.version, self.if_f0), SynthesizerTrnMs256NSFsid
-        )(*self.cpt["config"], is_half=self.config.is_half)
-
-        del self.net_g.enc_q
-
-        self.net_g.load_state_dict(self.cpt["weight"], strict=False)
-        self.net_g.eval().to(self.config.device)
         if self.config.is_half:
             self.net_g = self.net_g.half()
         else:
             self.net_g = self.net_g.float()
-
         self.pipeline = Pipeline(self.tgt_sr, self.config)
+
         n_spk = self.cpt["config"][-3]
         index = {"value": get_index_path_from_model(sid), "__type__": "update"}
         logger.info("Select index: " + index["value"])
@@ -138,6 +109,7 @@ class VC:
                 to_return_protect1,
                 index,
                 index,
+                show_model_info(self.cpt),
             )
             if to_return_protect
             else {"visible": True, "maximum": n_spk, "__type__": "update"}
@@ -160,18 +132,22 @@ class VC:
     ):
         if input_audio_path is None:
             return "You need to upload an audio", None
+        elif hasattr(input_audio_path, "name"):
+            input_audio_path = str(input_audio_path.name)
         f0_up_key = int(f0_up_key)
         try:
             audio = load_audio(input_audio_path, 16000)
             audio_max = np.abs(audio).max() / 0.95
             if audio_max > 1:
-                audio /= audio_max
+                np.divide(audio, audio_max, audio)
             times = [0, 0, 0]
 
             if self.hubert_model is None:
-                self.hubert_model = load_hubert(self.config)
+                self.hubert_model = load_hubert(self.config.device, self.config.is_half)
 
             if file_index:
+                if hasattr(file_index, "name"):
+                    file_index = str(file_index.name)
                 file_index = (
                     file_index.strip(" ")
                     .strip('"')
@@ -190,7 +166,6 @@ class VC:
                 self.net_g,
                 sid,
                 audio,
-                input_audio_path,
                 times,
                 f0_up_key,
                 f0_method,
@@ -204,25 +179,25 @@ class VC:
                 self.version,
                 protect,
                 f0_file,
-            )
+            ).astype(np.int16)
             if self.tgt_sr != resample_sr >= 16000:
                 tgt_sr = resample_sr
             else:
                 tgt_sr = self.tgt_sr
             index_info = (
-                "Index:\n%s." % file_index
+                "Index: %s." % file_index
                 if os.path.exists(file_index)
                 else "Index not used."
             )
             return (
-                "Success.\n%s\nTime:\nnpy: %.2fs, f0: %.2fs, infer: %.2fs."
+                "Success.\n%s\nTime: npy: %.2fs, f0: %.2fs, infer: %.2fs."
                 % (index_info, *times),
                 (tgt_sr, audio_opt),
             )
-        except:
+        except Exception as e:
             info = traceback.format_exc()
             logger.warning(info)
-            return info, (None, None)
+            return str(e), None
 
     def vc_multi(
         self,

infer/modules/vc/pipeline.py

@@ -5,40 +5,22 @@ import logging
 
 logger = logging.getLogger(__name__)
 
-from functools import lru_cache
-from time import time as ttime
+from time import time
 
 import faiss
 import librosa
 import numpy as np
-import parselmouth
-import pyworld
 import torch
 import torch.nn.functional as F
-import torchcrepe
 from scipy import signal
 
+from rvc.f0 import PM, Harvest, RMVPE, CRePE, Dio, FCPE
+
 now_dir = os.getcwd()
 sys.path.append(now_dir)
 
 bh, ah = signal.butter(N=5, Wn=48, btype="high", fs=16000)
 
-input_audio_path2wav = {}
-
-
-@lru_cache
-def cache_harvest_f0(input_audio_path, fs, f0max, f0min, frame_period):
-    audio = input_audio_path2wav[input_audio_path]
-    f0, t = pyworld.harvest(
-        audio,
-        fs=fs,
-        f0_ceil=f0max,
-        f0_floor=f0min,
-        frame_period=frame_period,
-    )
-    f0 = pyworld.stonemask(audio, f0, t, fs)
-    return f0
-
 
 def change_rms(data1, sr1, data2, sr2, rate):  # 1是输入音频，2是输出音频,rate是2的占比
     # print(data1.max(),data2.max())
@@ -83,7 +65,6 @@ class Pipeline(object):
 
     def get_f0(
         self,
-        input_audio_path,
         x,
         p_len,
         f0_up_key,
@@ -91,73 +72,62 @@ class Pipeline(object):
         filter_radius,
         inp_f0=None,
     ):
-        global input_audio_path2wav
-        time_step = self.window / self.sr * 1000
         f0_min = 50
         f0_max = 1100
         f0_mel_min = 1127 * np.log(1 + f0_min / 700)
         f0_mel_max = 1127 * np.log(1 + f0_max / 700)
         if f0_method == "pm":
-            f0 = (
-                parselmouth.Sound(x, self.sr)
-                .to_pitch_ac(
-                    time_step=time_step / 1000,
-                    voicing_threshold=0.6,
-                    pitch_floor=f0_min,
-                    pitch_ceiling=f0_max,
-                )
-                .selected_array["frequency"]
-            )
-            pad_size = (p_len - len(f0) + 1) // 2
-            if pad_size > 0 or p_len - len(f0) - pad_size > 0:
-                f0 = np.pad(
-                    f0, [[pad_size, p_len - len(f0) - pad_size]], mode="constant"
-                )
+            if not hasattr(self, "pm"):
+                self.pm = PM(self.window, f0_min, f0_max, self.sr)
+            f0 = self.pm.compute_f0(x, p_len=p_len)
+        if f0_method == "dio":
+            if not hasattr(self, "dio"):
+                self.dio = Dio(self.window, f0_min, f0_max, self.sr)
+            f0 = self.dio.compute_f0(x, p_len=p_len)
         elif f0_method == "harvest":
-            input_audio_path2wav[input_audio_path] = x.astype(np.double)
-            f0 = cache_harvest_f0(input_audio_path, self.sr, f0_max, f0_min, 10)
-            if filter_radius > 2:
-                f0 = signal.medfilt(f0, 3)
+            if not hasattr(self, "harvest"):
+                self.harvest = Harvest(self.window, f0_min, f0_max, self.sr)
+            f0 = self.harvest.compute_f0(x, p_len=p_len, filter_radius=filter_radius)
         elif f0_method == "crepe":
-            model = "full"
-            # Pick a batch size that doesn't cause memory errors on your gpu
-            batch_size = 512
-            # Compute pitch using first gpu
-            audio = torch.tensor(np.copy(x))[None].float()
-            f0, pd = torchcrepe.predict(
-                audio,
-                self.sr,
-                self.window,
-                f0_min,
-                f0_max,
-                model,
-                batch_size=batch_size,
-                device=self.device,
-                return_periodicity=True,
-            )
-            pd = torchcrepe.filter.median(pd, 3)
-            f0 = torchcrepe.filter.mean(f0, 3)
-            f0[pd < 0.1] = 0
-            f0 = f0[0].cpu().numpy()
+            if not hasattr(self, "crepe"):
+                self.crepe = CRePE(
+                    self.window,
+                    f0_min,
+                    f0_max,
+                    self.sr,
+                    self.device,
+                )
+            f0 = self.crepe.compute_f0(x, p_len=p_len)
         elif f0_method == "rmvpe":
-            if not hasattr(self, "model_rmvpe"):
-                from infer.lib.rmvpe import RMVPE
-
+            if not hasattr(self, "rmvpe"):
                 logger.info(
-                    "Loading rmvpe model,%s" % "%s/rmvpe.pt" % os.environ["rmvpe_root"]
+                    "Loading rmvpe model %s" % "%s/rmvpe.pt" % os.environ["rmvpe_root"]
                 )
-                self.model_rmvpe = RMVPE(
+                self.rmvpe = RMVPE(
                     "%s/rmvpe.pt" % os.environ["rmvpe_root"],
                     is_half=self.is_half,
                     device=self.device,
+                    # use_jit=self.config.use_jit,
                 )
-            f0 = self.model_rmvpe.infer_from_audio(x, thred=0.03)
+            f0 = self.rmvpe.compute_f0(x, p_len=p_len, filter_radius=0.03)
 
             if "privateuseone" in str(self.device):  # clean ortruntime memory
-                del self.model_rmvpe.model
-                del self.model_rmvpe
+                del self.rmvpe.model
+                del self.rmvpe
                 logger.info("Cleaning ortruntime memory")
 
+        elif f0_method == "fcpe":
+            if not hasattr(self, "model_fcpe"):
+                logger.info("Loading fcpe model")
+                self.model_fcpe = FCPE(
+                    self.window,
+                    f0_min,
+                    f0_max,
+                    self.sr,
+                    self.device,
+                )
+            f0 = self.model_fcpe.compute_f0(x, p_len=p_len)
+
         f0 *= pow(2, f0_up_key / 12)
         # with open("test.txt","w")as f:f.write("\n".join([str(i)for i in f0.tolist()]))
         tf0 = self.sr // self.window  # 每秒f0点数
@@ -214,7 +184,7 @@ class Pipeline(object):
             "padding_mask": padding_mask,
             "output_layer": 9 if version == "v1" else 12,
         }
-        t0 = ttime()
+        t0 = time()
         with torch.no_grad():
             logits = model.extract_features(**inputs)
             feats = model.final_proj(logits[0]) if version == "v1" else logits[0]
@@ -232,7 +202,10 @@ class Pipeline(object):
             # _, I = index.search(npy, 1)
             # npy = big_npy[I.squeeze()]
 
-            score, ix = index.search(npy, k=8)
+            try:
+                score, ix = index.search(npy, k=8)
+            except:
+                raise Exception("index mistatch")
             weight = np.square(1 / score)
             weight /= weight.sum(axis=1, keepdims=True)
             npy = np.sum(big_npy[ix] * np.expand_dims(weight, axis=2), axis=1)
@@ -249,7 +222,7 @@ class Pipeline(object):
             feats0 = F.interpolate(feats0.permute(0, 2, 1), scale_factor=2).permute(
                 0, 2, 1
             )
-        t1 = ttime()
+        t1 = time()
         p_len = audio0.shape[0] // self.window
         if feats.shape[1] < p_len:
             p_len = feats.shape[1]
@@ -266,14 +239,26 @@ class Pipeline(object):
             feats = feats.to(feats0.dtype)
         p_len = torch.tensor([p_len], device=self.device).long()
         with torch.no_grad():
-            hasp = pitch is not None and pitchf is not None
-            arg = (feats, p_len, pitch, pitchf, sid) if hasp else (feats, p_len, sid)
-            audio1 = (net_g.infer(*arg)[0][0, 0]).data.cpu().float().numpy()
-            del hasp, arg
+            audio1 = (
+                (
+                    net_g.infer(
+                        feats,
+                        p_len,
+                        sid,
+                        pitch=pitch,
+                        pitchf=pitchf,
+                    )[0, 0]
+                )
+                .data.cpu()
+                .float()
+                .numpy()
+            )
         del feats, p_len, padding_mask
         if torch.cuda.is_available():
             torch.cuda.empty_cache()
-        t2 = ttime()
+        elif torch.backends.mps.is_available():
+            torch.mps.empty_cache()
+        t2 = time()
         times[0] += t1 - t0
         times[2] += t2 - t1
         return audio1
@@ -284,7 +269,6 @@ class Pipeline(object):
         net_g,
         sid,
         audio,
-        input_audio_path,
         times,
         f0_up_key,
         f0_method,
@@ -308,7 +292,6 @@ class Pipeline(object):
         ):
             try:
                 index = faiss.read_index(file_index)
-                # big_npy = np.load(file_big_npy)
                 big_npy = index.reconstruct_n(0, index.ntotal)
             except:
                 traceback.print_exc()
@@ -334,7 +317,7 @@ class Pipeline(object):
         s = 0
         audio_opt = []
         t = None
-        t1 = ttime()
+        t1 = time()
         audio_pad = np.pad(audio, (self.t_pad, self.t_pad), mode="reflect")
         p_len = audio_pad.shape[0] // self.window
         inp_f0 = None
@@ -350,27 +333,29 @@ class Pipeline(object):
                 traceback.print_exc()
         sid = torch.tensor(sid, device=self.device).unsqueeze(0).long()
         pitch, pitchf = None, None
-        if if_f0 == 1:
-            pitch, pitchf = self.get_f0(
-                input_audio_path,
-                audio_pad,
-                p_len,
-                f0_up_key,
-                f0_method,
-                filter_radius,
-                inp_f0,
-            )
+        if if_f0:
+            if if_f0 == 1:
+                pitch, pitchf = self.get_f0(
+                    audio_pad,
+                    p_len,
+                    f0_up_key,
+                    f0_method,
+                    filter_radius,
+                    inp_f0,
+                )
+            elif if_f0 == 2:
+                pitch, pitchf = f0_method
             pitch = pitch[:p_len]
             pitchf = pitchf[:p_len]
             if "mps" not in str(self.device) or "xpu" not in str(self.device):
                 pitchf = pitchf.astype(np.float32)
             pitch = torch.tensor(pitch, device=self.device).unsqueeze(0).long()
             pitchf = torch.tensor(pitchf, device=self.device).unsqueeze(0).float()
-        t2 = ttime()
+        t2 = time()
         times[1] += t2 - t1
         for t in opt_ts:
             t = t // self.window * self.window
-            if if_f0 == 1:
+            if if_f0:
                 audio_opt.append(
                     self.vc(
                         model,
@@ -405,7 +390,7 @@ class Pipeline(object):
                     )[self.t_pad_tgt : -self.t_pad_tgt]
                 )
             s = t
-        if if_f0 == 1:
+        if if_f0:
             audio_opt.append(
                 self.vc(
                     model,
@@ -450,8 +435,10 @@ class Pipeline(object):
         max_int16 = 32768
         if audio_max > 1:
             max_int16 /= audio_max
-        audio_opt = (audio_opt * max_int16).astype(np.int16)
+        np.multiply(audio_opt, max_int16, audio_opt)
         del pitch, pitchf, sid
         if torch.cuda.is_available():
             torch.cuda.empty_cache()
+        elif torch.backends.mps.is_available():
+            torch.mps.empty_cache()
         return audio_opt

infer/modules/vc/utils.py

@@ -9,7 +9,8 @@ def get_index_path_from_model(sid):
             f
             for f in [
                 os.path.join(root, name)
-                for root, _, files in os.walk(os.getenv("index_root"), topdown=False)
+                for path in [os.getenv("outside_index_root"), os.getenv("index_root")]
+                for root, _, files in os.walk(path, topdown=False)
                 for name in files
                 if name.endswith(".index") and "trained" not in name
             ]
@@ -19,14 +20,14 @@ def get_index_path_from_model(sid):
     )
 
 
-def load_hubert(config):
+def load_hubert(device, is_half):
     models, _, _ = checkpoint_utils.load_model_ensemble_and_task(
         ["assets/hubert/hubert_base.pt"],
         suffix="",
     )
     hubert_model = models[0]
-    hubert_model = hubert_model.to(config.device)
-    if config.is_half:
+    hubert_model = hubert_model.to(device)
+    if is_half:
         hubert_model = hubert_model.half()
     else:
         hubert_model = hubert_model.float()