Added Pekora, updated images. Adjusted model loading for buggy test voices.

app.py

@@ -150,7 +150,6 @@ def tts_fn(
 
 def load_voicedata():
     print("Loading voice data...")
-    #voices = []
     envoices = []
     jpvoices = []
     styledict = {}
@@ -169,17 +168,20 @@ def load_voicedata():
         image = info['cover']
         if not os.path.exists(f"images/{image}"):
             image="none.png"
+        # for voices that are either known buggy or abnormal
+        nospace=False
+        if 'disableonspace' in info:
+            nospace=info['disableonspace']
         if not model_path in styledict.keys():
            conf=f"{model_dir}/{model_path}/config.json"
            hps = utils.get_hparams_from_file(conf)
            s2id = hps.data.style2id
            styledict[model_path] = s2id.keys()
            print(f"Set up hyperparameters for model {model_path}")
-        #print(f"Indexed voice {voice_name}")
         if(info['primarylang']=="JP"):
-            jpvoices.append((name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image))
+            jpvoices.append((name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image, nospace))
         else:
-            envoices.append((name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image))
+            envoices.append((name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image, nospace))
     return [envoices, jpvoices], styledict
         
 
@@ -301,7 +303,9 @@ if __name__ == "__main__":
         #for (name, model_path, voice_name, speakerid, datasetauthor, image) in voicedata:
         for vi in range(len(voicedata)):
             with gr.TabItem(langnames[vi]):
-                for (name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image) in voicedata[vi]:
+                for (name, model_path, model_path_full, voice_name, speakerid, datasetauthor, image, nospace) in voicedata[vi]:
+                    if(nospace and is_hf_spaces):
+                        continue
                     with gr.TabItem(name):
                         mn = gr.Textbox(value=model_path, visible=False, interactive=False)
                         mp = gr.Textbox(value=model_path_full, visible=False, interactive=False)

model_assets/SBV2_UsadaPekora/SBV2_UsadaPekora.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb16c8c20d9dcddb1dd7f871998730b6b115306a317a21ada4d64da0cb55c382
+size 198769212

model_assets/SBV2_UsadaPekora/config.json

@@ -0,0 +1,114 @@
+{
+  "model_name": "SBV2_UsadaPekora",
+  "train": {
+    "log_interval": 200,
+    "eval_interval": 1000,
+    "seed": 42,
+    "epochs": 100,
+    "learning_rate": 0.0002,
+    "betas": [
+      0.8,
+      0.99
+    ],
+    "eps": 1e-09,
+    "batch_size": 2,
+    "bf16_run": false,
+    "lr_decay": 0.99995,
+    "segment_size": 16384,
+    "init_lr_ratio": 1,
+    "warmup_epochs": 0,
+    "c_mel": 45,
+    "c_kl": 1.0,
+    "skip_optimizer": false,
+    "freeze_ZH_bert": false,
+    "freeze_JP_bert": false,
+    "freeze_EN_bert": false,
+    "freeze_style": false,
+    "freeze_encoder": false,
+    "freeze_decoder": false
+  },
+  "data": {
+    "use_jp_extra": false,
+    "training_files": "Data\\SBV2_UsadaPekora\\train.list",
+    "validation_files": "Data\\SBV2_UsadaPekora\\val.list",
+    "max_wav_value": 32768.0,
+    "sampling_rate": 44100,
+    "filter_length": 2048,
+    "hop_length": 512,
+    "win_length": 2048,
+    "n_mel_channels": 128,
+    "mel_fmin": 0.0,
+    "mel_fmax": null,
+    "add_blank": true,
+    "n_speakers": 2,
+    "cleaned_text": true,
+    "num_styles": 6,
+    "style2id": {
+      "Neutral": 0,
+      "Excited": 1,
+      "Thoughtful": 2,
+      "Explaining": 3,
+      "Mama": 4,
+      "Sad": 5
+    },
+    "spk2id": {
+      "UsadaPekora": 0,
+      "Pekomama": 1
+    }
+  },
+  "model": {
+    "use_spk_conditioned_encoder": true,
+    "use_noise_scaled_mas": true,
+    "use_mel_posterior_encoder": false,
+    "use_duration_discriminator": true,
+    "inter_channels": 192,
+    "hidden_channels": 192,
+    "filter_channels": 768,
+    "n_heads": 2,
+    "n_layers": 6,
+    "kernel_size": 3,
+    "p_dropout": 0.1,
+    "resblock": "1",
+    "resblock_kernel_sizes": [
+      3,
+      7,
+      11
+    ],
+    "resblock_dilation_sizes": [
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ],
+      [
+        1,
+        3,
+        5
+      ]
+    ],
+    "upsample_rates": [
+      8,
+      8,
+      2,
+      2,
+      2
+    ],
+    "upsample_initial_channel": 512,
+    "upsample_kernel_sizes": [
+      16,
+      16,
+      8,
+      2,
+      2
+    ],
+    "n_layers_q": 3,
+    "use_spectral_norm": false,
+    "gin_channels": 256
+  },
+  "version": "2.5.0"
+}

model_assets/SBV2_UsadaPekora/style_vectors.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12b73c03f08086906421f0942df5021672f6d49a0c5b7bea55455a3a3583de01
+size 6272

voicelist.json

@@ -179,6 +179,16 @@
 	"primarylang": "EN",
     "cover": "zeta.png"
   },
+  "Kobo": {
+    "enable": true,
+    "model_path": "SBV2_HoloESL",
+    "title": "Kobo Kanaeru",
+	"speakerid": "KoboKanaeru",
+	"datasetauthor": "megaaziib",
+	"primarylang": "EN",
+    "cover": "kobo.png",
+	"disableonspace": true
+  },
   "Sora": {
     "enable": true,
     "model_path": "SBV2_HoloJPTest2",
@@ -296,6 +306,25 @@
 	"primarylang": "JP",
     "cover": "okayu.png"
   },
+  "Pekora": {
+    "enable": true,
+    "model_path": "SBV2_UsadaPekora",
+    "title": "Usada Pekora",
+	"speakerid": "UsadaPekora",
+	"datasetauthor": "Kit Lemonfoot",
+	"primarylang": "JP",
+    "cover": "pekora.png"
+  },
+  "Pekomama": {
+    "enable": true,
+    "model_path": "SBV2_UsadaPekora",
+    "title": "Pekomama",
+	"speakerid": "Pekomama",
+	"datasetauthor": "dacoolkid44",
+	"primarylang": "JP",
+    "cover": "pekomama.png",
+	"disableonspace": true
+  },
   "Rushia": {
     "enable": true,
     "model_path": "SBV2_HoloJPTest3",