new TTS: Index TTS; SoVITS disabled due to mid perfo

app/models.py

@@ -73,7 +73,7 @@ AVAILABLE_MODELS = {
     # 'Svngoku/maskgct-audio-lab': 'Svngoku/maskgct-audio-lab', # DEMANDS 300 seconds of ZeroGPU!
 
     # GPT-SoVITS
-    'lj1995/GPT-SoVITS-v2': 'lj1995/GPT-SoVITS-v2',
+    # 'lj1995/GPT-SoVITS-v2': 'lj1995/GPT-SoVITS-v2',
 
     # OuteTTS 500M
     # 'OuteAI/OuteTTS-0.2-500M-Demo': 'OuteAI/OuteTTS-0.2-500M-Demo',
@@ -107,6 +107,9 @@ AVAILABLE_MODELS = {
     # Orpheus
     'MohamedRashad/Orpheus-TTS' : 'MohamedRashad/Orpheus-TTS',
 
+    # Index TTS
+    'IndexTeam/IndexTTS': 'IndexTeam/IndexTTS',
+
     # HF TTS w issues
     # 'LeeSangHoon/HierSpeech_TTS': 'LeeSangHoon/HierSpeech_TTS', # irresponsive to exclamation marks # 4.29
     # 'PolyAI/pheme': '/predict#0', # sleepy HF Space
@@ -505,6 +508,15 @@ HF_SPACES = {
         'is_zero_gpu_space': True,
         'series': 'Orpheus',
     },
+
+    'IndexTeam/IndexTTS' : {
+        'name': 'Index TTS',
+        'function': '/gen_single',
+        'text_param_index': 'text',
+        'return_audio_index': 0,
+        'is_zero_gpu_space': True,
+        'series': 'Index',
+    },
 }
 
 # for zero-shot TTS - voice sample used by XTTS (11 seconds)
@@ -806,7 +818,12 @@ OVERRIDE_INPUTS = {
 		'top_p': 0.95,
 		'repetition_penalty': 1.1,
 		'max_new_tokens': 1200,
-    }
+    },
+
+    # Index TTS
+    'IndexTeam/IndexTTS' : {
+		'prompt': DEFAULT_VOICE_SAMPLE, # voice
+    },
 }
 
 # minor mods to model from the same space
@@ -871,7 +888,7 @@ closed_source = [
 ]
 
 # top five models in order to always have one of them picked and scrutinized
-top_five = ['thunnai/SparkTTS']
+top_five = []
 
 # prioritize low vote models
 sql = 'SELECT name FROM model WHERE (upvote + downvote) < 700 ORDER BY (upvote + downvote) ASC'

test_tts_index.py

@@ -0,0 +1,50 @@
+import os
+from test_overrides import _get_param_examples, _override_params
+from gradio_client import Client, file
+
+model = "IndexTeam/IndexTTS"
+client = Client(model, hf_token=os.getenv('HF_TOKEN'))
+endpoints = client.view_api(all_endpoints=True, print_info=False, return_format='dict')
+# print(endpoints)
+
+api_name = '/gen_single'
+fn_index = None
+end_parameters = None
+text = 'This is what my voice sounds like.'
+
+end_parameters = _get_param_examples(
+	endpoints['named_endpoints'][api_name]['parameters']
+)
+print(end_parameters)
+
+
+space_inputs = end_parameters
+# override some or all default parameters
+space_inputs = _override_params(end_parameters, model)
+
+if(type(space_inputs) == dict):
+	space_inputs['text'] = text
+	result = client.predict(
+		**space_inputs,
+		api_name=api_name,
+		fn_index=fn_index
+	)
+else:
+	space_inputs[0] = text
+	result = client.predict(
+		*space_inputs,
+		api_name=api_name,
+		fn_index=fn_index
+	)
+	# space_inputs = {str(i): value for i, value in enumerate(space_inputs)}
+
+print(space_inputs)
+# print(*space_inputs)
+# print(**space_inputs)
+
+# result = client.predict(
+# 	**space_inputs,
+# 	api_name=api_name,
+#     fn_index=fn_index
+# )
+print(result)

test_tts_llasa.py

@@ -0,0 +1,50 @@
+import os
+from test_overrides import _get_param_examples, _override_params
+from gradio_client import Client, file
+
+model = "srinivasbilla/llasa-3b-tts"
+client = Client(model, hf_token=os.getenv('HF_TOKEN'))
+endpoints = client.view_api(all_endpoints=True, print_info=False, return_format='dict')
+# print(endpoints)
+
+api_name = '/infer'
+fn_index = None
+end_parameters = None
+text = 'This is what my voice sounds like.'
+
+end_parameters = _get_param_examples(
+	endpoints['named_endpoints'][api_name]['parameters']
+)
+print(end_parameters)
+
+
+space_inputs = end_parameters
+# override some or all default parameters
+space_inputs = _override_params(end_parameters, model)
+
+if(type(space_inputs) == dict):
+	space_inputs['target_text'] = text
+	result = client.predict(
+		**space_inputs,
+		api_name=api_name,
+		fn_index=fn_index
+	)
+else:
+	space_inputs[0] = text
+	result = client.predict(
+		*space_inputs,
+		api_name=api_name,
+		fn_index=fn_index
+	)
+	# space_inputs = {str(i): value for i, value in enumerate(space_inputs)}
+
+print(space_inputs)
+# print(*space_inputs)
+# print(**space_inputs)
+
+# result = client.predict(
+# 	**space_inputs,
+# 	api_name=api_name,
+#     fn_index=fn_index
+# )
+print(result)