update new mix svs model (update ace-opencpop dataset)

exp/{svs_train_visinger2_raw_phn_none_mix/images/discriminator_loss.png → svs_stats_raw_phn_none_mix/train/feats_stats.npz}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0128375cff96feefa7bc083a580522796dc8abd9a301b94f0483d9acc199d19e
-size 69034
+oid sha256:95cbffa329e7b157b9c3481e7f8442a1867ead3b27518e7649ec486b76a56fab
+size 1402

exp/{svs_train_visinger2_raw_phn_none_mix/images/discriminator_backward_time.png → svs_stats_raw_phn_none_mix/train/pitch_stats.npz}

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:87c466c2258086564e173d691d245d8ed1a02d6baf873df3d2d58bc5f9b61090
-size 49691
+oid sha256:57b5a91ae0d5d359c421e234aadc10c44f3892fb73392cfb1cdb72810afbca1c
+size 770

exp/{svs_train_visinger2_raw_phn_none_mix → svs_train_visinger2_spk_embed_lang_raw_phn_none_mix}/500epoch.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2c2d019f47dce46ab6f9022606ba4d02365a39ba090093db562ec7f5ba0edd22
-size 448311886
+oid sha256:c93a08500a8958e48a57dd6780033635d11916de8e1d72a8691124ef2eb95f2a
+size 448412302

exp/{svs_train_visinger2_raw_phn_none_mix → svs_train_visinger2_spk_embed_lang_raw_phn_none_mix}/config.yaml

@@ -1,11 +1,11 @@
-config: conf/tuning/train_visinger2.yaml
+config: conf/tuning/train_visinger2_spk_embed_lang.yaml
 print_config: false
 log_level: INFO
 drop_last_iter: false
 dry_run: false
 iterator_type: sequence
 valid_iterator_type: null
-output_dir: exp/svs_train_visinger2_raw_phn_none_mix
+output_dir: exp/svs_train_visinger2_spk_embed_lang_raw_phn_none_mix
 ngpu: 1
 seed: 777
 num_workers: 0
@@ -224,8 +224,8 @@ token_list:
 - an@zh
 - en@zh
 - iou@zh
-- t@zh
 - ou@zh
+- t@zh
 - ao@zh
 - ong@zh
 - iang@zh
@@ -319,16 +319,15 @@ feats_extract_conf:
     fmin: 80
     fmax: 22050
     n_mels: 80
-normalize: global_mvn
-normalize_conf:
-    stats_file: exp/svs_stats_raw_phn_none_mix/train/feats_stats.npz
+normalize: null
+normalize_conf: {}
 svs: vits
 svs_conf:
     generator_type: visinger2
     vocoder_generator_type: visinger2
     generator_params:
         hidden_channels: 192
-        spks: 94
+        spk_embed_dim: 192
         langs: 3
         global_channels: 256
         segment_size: 20
@@ -538,5 +537,5 @@ energy_normalize_conf: {}
 required:
 - output_dir
 - token_list
-version: '202412'
+version: '202503'
 distributed: false