train index:auto kmeans when feature shape too large

train index:auto kmeans when feature shape too large

infer-web.py

@@ -35,8 +35,7 @@ from MDXNet import MDXNetDereverb
 from my_utils import load_audio
 from train.process_ckpt import change_info, extract_small_model, merge, show_info
 from vc_infer_pipeline import VC
-
+from sklearn.cluster import MiniBatchKMeans
-# from trainset_preprocess_pipeline import PreProcess
 
 logging.getLogger("numba").setLevel(logging.WARNING)
 
@@ -653,9 +652,13 @@ def change_sr2(sr2, if_f0_3, version19):
             "not exist, will not use pretrained model",
         )
     return (
-        "pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2) if if_pretrained_generator_exist else "",
+        "pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)
-        "pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2) if if_pretrained_discriminator_exist else "",
+        if if_pretrained_generator_exist
-        {"visible": True, "__type__": "update"}
+        else "",
+        "pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2)
+        if if_pretrained_discriminator_exist
+        else "",
+        {"visible": True, "__type__": "update"},
     )
 
 
@@ -679,8 +682,12 @@ def change_version19(sr2, if_f0_3, version19):
             "not exist, will not use pretrained model",
         )
     return (
-        "pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2) if if_pretrained_generator_exist else "",
+        "pretrained%s/%sG%s.pth" % (path_str, f0_str, sr2)
-        "pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2) if if_pretrained_discriminator_exist else "",
+        if if_pretrained_generator_exist
+        else "",
+        "pretrained%s/%sD%s.pth" % (path_str, f0_str, sr2)
+        if if_pretrained_discriminator_exist
+        else "",
     )
 
 
@@ -714,8 +721,12 @@ def change_f0(if_f0_3, sr2, version19):  # f0method8,pretrained_G14,pretrained_D
         )
     return (
         {"visible": False, "__type__": "update"},
-        "pretrained%s/G%s.pth" % (path_str, sr2) if if_pretrained_generator_exist else "",
+        ("pretrained%s/G%s.pth" % (path_str, sr2))
-        "pretrained%s/D%s.pth" % (path_str, sr2) if if_pretrained_discriminator_exist else "",
+        if if_pretrained_generator_exist
+        else "",
+        ("pretrained%s/D%s.pth" % (path_str, sr2))
+        if if_pretrained_discriminator_exist
+        else "",
     )
 
 
@@ -869,6 +880,7 @@ def train_index(exp_dir1, version19):
     listdir_res = list(os.listdir(feature_dir))
     if len(listdir_res) == 0:
         return "请先进行特征提取！"
+    infos = []
     npys = []
     for name in sorted(listdir_res):
         phone = np.load("%s/%s" % (feature_dir, name))
@@ -877,10 +889,20 @@ def train_index(exp_dir1, version19):
     big_npy_idx = np.arange(big_npy.shape[0])
     np.random.shuffle(big_npy_idx)
     big_npy = big_npy[big_npy_idx]
+    # if(big_npy.shape[0]>2e5):
+    if(1):
+        infos.append("Trying doing kmeans %s shape to 10k centers."%big_npy.shape[0])
+        yield "\n".join(infos)
+        try:
+            big_npy = MiniBatchKMeans(n_clusters=10000, verbose=True, batch_size=256 * config.n_cpu, compute_labels=False, init="random").fit(big_npy).cluster_centers_
+        except:
+            info=traceback.format_exc()
+            print(info)
+            infos.append(info)
+            yield "\n".join(infos)
+
     np.save("%s/total_fea.npy" % exp_dir, big_npy)
-    # n_ivf =  big_npy.shape[0] // 39
     n_ivf = min(int(16 * np.sqrt(big_npy.shape[0])), big_npy.shape[0] // 39)
-    infos = []
     infos.append("%s,%s" % (big_npy.shape, n_ivf))
     yield "\n".join(infos)
     index = faiss.index_factory(256 if version19 == "v1" else 768, "IVF%s,Flat" % n_ivf)
@@ -1120,6 +1142,19 @@ def train1key(
     big_npy_idx = np.arange(big_npy.shape[0])
     np.random.shuffle(big_npy_idx)
     big_npy = big_npy[big_npy_idx]
+
+    # if(big_npy.shape[0]>2e5):
+    if(1):
+        info="Trying doing kmeans %s shape to 10k centers."%big_npy.shape[0]
+        print(info)
+        yield get_info_str(info)
+        try:
+            big_npy = MiniBatchKMeans(n_clusters=10000, verbose=True, batch_size=256 * config.n_cpu, compute_labels=False, init="random").fit(big_npy).cluster_centers_
+        except:
+            info=traceback.format_exc()
+            print(info)
+            yield get_info_str(info)
+
     np.save("%s/total_fea.npy" % model_log_dir, big_npy)
 
     # n_ivf =  big_npy.shape[0] // 39
@@ -1565,7 +1600,7 @@ with gr.Blocks() as app:
                     maximum=config.n_cpu,
                     step=1,
                     label=i18n("提取音高和处理数据使用的CPU进程数"),
-                    value=config.n_cpu,
+                    value=int(np.ceil(config.n_cpu/1.5)),
                     interactive=True,
                 )
             with gr.Group():  # 暂时单人的, 后面支持最多4人的#数据处理