upload model with spemb

exp/svs_train_visinger2_raw_phn_none_mix/500epoch.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2c2d019f47dce46ab6f9022606ba4d02365a39ba090093db562ec7f5ba0edd22
-size 448311886
+oid sha256:3b2deb4f692ee1caf0e82a5c195dc23df5e9157ee7c3a2d81074a6c633fce972
+size 448413582

exp/svs_train_visinger2_raw_phn_none_mix/config.yaml

@@ -1,14 +1,14 @@
-config: conf/tuning/train_visinger2.yaml
+config: conf/tuning/train_visinger2_44k_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_44k_spk_embed_lang_0409
 ngpu: 1
 seed: 777
-num_workers: 0
+num_workers: 4
 num_att_plot: 0
 dist_backend: nccl
 dist_init_method: env://
@@ -23,8 +23,6 @@ unused_parameters: true
 sharded_ddp: false
 use_deepspeed: false
 deepspeed_config: null
-gradient_as_bucket_view: true
-ddp_comm_hook: null
 cudnn_enabled: true
 cudnn_benchmark: false
 cudnn_deterministic: false
@@ -80,11 +78,11 @@ batch_bins: 1000000
 valid_batch_bins: null
 category_sample_size: 10
 train_shape_file:
-- exp/svs_stats_raw_phn_none_mix/train/text_shape.phn
-- exp/svs_stats_raw_phn_none_mix/train/singing_shape
+- exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/text_shape.phn
+- exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/singing_shape
 valid_shape_file:
-- exp/svs_stats_raw_phn_none_mix/valid/text_shape.phn
-- exp/svs_stats_raw_phn_none_mix/valid/singing_shape
+- exp/svs_stats_raw_phn_None_zh_jp_44100Hz/valid/text_shape.phn
+- exp/svs_stats_raw_phn_None_zh_jp_44100Hz/valid/singing_shape
 batch_type: sorted
 valid_batch_type: null
 fold_length:
@@ -102,59 +100,53 @@ chunk_default_fs: null
 chunk_max_abs_length: null
 chunk_discard_short_samples: true
 train_data_path_and_name_and_type:
--   - dump_mix/raw/tr_no_dev/text
+-   - dump44100/raw/tr_no_dev/text
     - text
     - text
--   - dump_mix/raw/tr_no_dev/wav.scp
+-   - dump44100/raw/tr_no_dev/wav.scp
     - singing
     - sound
--   - dump_mix/raw/tr_no_dev/label
+-   - dump44100/raw/tr_no_dev/label
     - label
     - duration
--   - dump_mix/raw/tr_no_dev/score.scp
+-   - dump44100/raw/tr_no_dev/score.scp
     - score
     - score
--   - exp/svs_stats_raw_phn_none_mix/train/collect_feats/pitch.scp
+-   - exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/collect_feats/pitch.scp
     - pitch
     - npy
--   - exp/svs_stats_raw_phn_none_mix/train/collect_feats/feats.scp
+-   - exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/collect_feats/feats.scp
     - feats
     - npy
--   - dump_mix/espnet_spk/tr_no_dev/espnet_spk.scp
+-   - dump44100/raw/tr_no_dev/espnet_spk.scp
     - spembs
     - kaldi_ark
--   - dump_mix/raw/tr_no_dev/utt2sid
-    - sids
-    - text_int
--   - dump_mix/raw/tr_no_dev/utt2lid
+-   - dump44100/raw/tr_no_dev/utt2lid
     - lids
     - text_int
 valid_data_path_and_name_and_type:
--   - dump_mix/raw/dev/text
+-   - dump44100/raw/dev/text
     - text
     - text
--   - dump_mix/raw/dev/wav.scp
+-   - dump44100/raw/dev/wav.scp
     - singing
     - sound
--   - dump_mix/raw/dev/label
+-   - dump44100/raw/dev/label
     - label
     - duration
--   - dump_mix/raw/dev/score.scp
+-   - dump44100/raw/dev/score.scp
     - score
     - score
--   - exp/svs_stats_raw_phn_none_mix/valid/collect_feats/pitch.scp
+-   - exp/svs_stats_raw_phn_None_zh_jp_44100Hz/valid/collect_feats/pitch.scp
     - pitch
     - npy
--   - exp/svs_stats_raw_phn_none_mix/valid/collect_feats/feats.scp
+-   - exp/svs_stats_raw_phn_None_zh_jp_44100Hz/valid/collect_feats/feats.scp
     - feats
     - npy
--   - dump_mix/espnet_spk/dev/espnet_spk.scp
+-   - dump44100/raw/dev/espnet_spk.scp
     - spembs
     - kaldi_ark
--   - dump_mix/raw/dev/utt2sid
-    - sids
-    - text_int
--   - dump_mix/raw/dev/utt2lid
+-   - dump44100/raw/dev/utt2lid
     - lids
     - text_int
 multi_task_dataset: false
@@ -194,7 +186,6 @@ token_list:
 - <blank>
 - <unk>
 - SP
-- AP
 - i@zh
 - e@zh
 - d@zh
@@ -210,8 +201,8 @@ token_list:
 - j@zh
 - l@zh
 - h@zh
-- b@zh
 - iii@zh
+- b@zh
 - zh@zh
 - uei@zh
 - m@zh
@@ -224,9 +215,10 @@ token_list:
 - an@zh
 - en@zh
 - iou@zh
-- t@zh
 - ou@zh
+- t@zh
 - ao@zh
+- AP
 - ong@zh
 - iang@zh
 - ang@zh
@@ -237,13 +229,13 @@ token_list:
 - r@zh
 - k@zh
 - ch@zh
+- a@jp
 - v@zh
 - in@zh
-- a@jp
-- uan@zh
 - o@jp
-- c@zh
+- uan@zh
 - i@jp
+- c@zh
 - s@zh
 - uang@zh
 - ii@zh
@@ -253,17 +245,17 @@ token_list:
 - u@jp
 - e@jp
 - ia@zh
-- uen@zh
 - k@jp
+- uen@zh
 - ua@zh
 - n@jp
-- iong@zh
 - t@jp
 - r@jp
-- uai@zh
+- iong@zh
 - m@jp
-- er@zh
+- uai@zh
 - s@jp
+- er@zh
 - vn@zh
 - w@jp
 - d@jp
@@ -271,8 +263,8 @@ token_list:
 - sh@jp
 - g@jp
 - y@jp
-- o@zh
 - b@jp
+- o@zh
 - ts@jp
 - h@jp
 - cl@jp
@@ -302,6 +294,7 @@ non_linguistic_symbols: null
 cleaner: null
 g2p: null
 fs: 44100
+use_spk_contrastive_loss: null
 postfrontend: null
 postfrontend_conf: {}
 score_feats_extract: syllable_score_feats
@@ -321,14 +314,14 @@ feats_extract_conf:
     n_mels: 80
 normalize: global_mvn
 normalize_conf:
-    stats_file: exp/svs_stats_raw_phn_none_mix/train/feats_stats.npz
+    stats_file: exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/feats_stats.npz
 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
@@ -517,6 +510,7 @@ svs_conf:
     lambda_kl: 1.0
     sampling_rate: 44100
     cache_generator_outputs: true
+    device: cuda
 pitch_extract: dio
 pitch_extract_conf:
     use_token_averaged_f0: false
@@ -528,7 +522,7 @@ pitch_extract_conf:
     f0min: 80
 pitch_normalize: null
 pitch_normalize_conf:
-    stats_file: exp/svs_stats_raw_phn_none_mix/train/pitch_stats.npz
+    stats_file: exp/svs_stats_raw_phn_None_zh_jp_44100Hz/train/pitch_stats.npz
 ying_extract: null
 ying_extract_conf: {}
 energy_extract: null
@@ -538,5 +532,5 @@ energy_normalize_conf: {}
 required:
 - output_dir
 - token_list
-version: '202412'
+version: '202402'
 distributed: false