Spaces:
Running
Running
added TTSs amphion, HierSpeech
Browse files
app.py
CHANGED
|
@@ -51,26 +51,38 @@ AVAILABLE_MODELS = {
|
|
| 51 |
# CoquiTTS (CPU)
|
| 52 |
# 'coqui/CoquiTTS': '0#0',
|
| 53 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
# 'pytorch/Tacotron2': '0#0', #old gradio
|
| 55 |
}
|
| 56 |
|
|
|
|
|
|
|
|
|
|
| 57 |
OVERRIDE_INPUTS = {
|
| 58 |
'coqui/xtts': {
|
| 59 |
1: 'en',
|
| 60 |
-
2:
|
| 61 |
-
3:
|
| 62 |
4: False, #use_mic
|
| 63 |
5: False, #cleanup_reference
|
| 64 |
6: False, #auto_detect
|
| 65 |
},
|
| 66 |
'collabora/WhisperSpeech': {
|
| 67 |
-
1:
|
| 68 |
-
2:
|
| 69 |
3: 14.0, #Tempo - Gradio Slider issue: takes min. rather than value
|
| 70 |
},
|
| 71 |
'myshell-ai/OpenVoice': {
|
| 72 |
1: 'default', # style
|
| 73 |
-
2:
|
| 74 |
},
|
| 75 |
'PolyAI/pheme': {
|
| 76 |
1: 'YOU1000000044_S0000798', # voice
|
|
@@ -81,6 +93,27 @@ OVERRIDE_INPUTS = {
|
|
| 81 |
1: 'ccby_nvidia_hifi_92_F', #fine-tuned voice model name
|
| 82 |
3: 1.0, #pacing/duration - Gradio Slider issue: takes min. rather than value
|
| 83 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 84 |
}
|
| 85 |
|
| 86 |
SPACE_ID = os.getenv('SPACE_ID')
|
|
@@ -339,9 +372,7 @@ model_licenses = {
|
|
| 339 |
'metavoice': 'Apache 2.0',
|
| 340 |
'elevenlabs': 'Proprietary',
|
| 341 |
'whisperspeech': 'MIT',
|
| 342 |
-
|
| 343 |
'Pendrokar/xVASynth': 'GPT3',
|
| 344 |
-
'Pendrokar/xVASynthStreaming': 'GPT3',
|
| 345 |
}
|
| 346 |
model_links = {
|
| 347 |
'styletts2': 'https://github.com/yl4579/StyleTTS2',
|
|
|
|
| 51 |
# CoquiTTS (CPU)
|
| 52 |
# 'coqui/CoquiTTS': '0#0',
|
| 53 |
|
| 54 |
+
'amphion/Text-to-Speech': '/predict#0',
|
| 55 |
+
'LeeSangHoon/HierSpeech_TTS': '/predict#0',
|
| 56 |
+
|
| 57 |
+
# TTS w issues
|
| 58 |
+
# 'suno/bark': '3#0', # Hallucinates
|
| 59 |
+
# 'shivammehta25/Matcha-TTS': '5#0', #seems to require multiple requests for setup
|
| 60 |
+
# 'styletts2/styletts2': '0#0', #API disabled
|
| 61 |
+
# 'mrfakename/MelloTTS': '0#0', #API disabled
|
| 62 |
+
# 'Manmay/tortoise-tts': '/predict#0', #Cannot skip text-from-file parameter
|
| 63 |
# 'pytorch/Tacotron2': '0#0', #old gradio
|
| 64 |
}
|
| 65 |
|
| 66 |
+
# for zero-shot TTS - voice sample of Scarlett Johanson
|
| 67 |
+
DEFAULT_VOICE_SAMPLE = 'https://cdn-uploads.huggingface.co/production/uploads/641de0213239b631552713e4/iKHHqWxWy6Zfmp6QP6CZZ.wav'
|
| 68 |
+
|
| 69 |
OVERRIDE_INPUTS = {
|
| 70 |
'coqui/xtts': {
|
| 71 |
1: 'en',
|
| 72 |
+
2: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 73 |
+
3: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 74 |
4: False, #use_mic
|
| 75 |
5: False, #cleanup_reference
|
| 76 |
6: False, #auto_detect
|
| 77 |
},
|
| 78 |
'collabora/WhisperSpeech': {
|
| 79 |
+
1: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 80 |
+
2: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 81 |
3: 14.0, #Tempo - Gradio Slider issue: takes min. rather than value
|
| 82 |
},
|
| 83 |
'myshell-ai/OpenVoice': {
|
| 84 |
1: 'default', # style
|
| 85 |
+
2: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 86 |
},
|
| 87 |
'PolyAI/pheme': {
|
| 88 |
1: 'YOU1000000044_S0000798', # voice
|
|
|
|
| 93 |
1: 'ccby_nvidia_hifi_92_F', #fine-tuned voice model name
|
| 94 |
3: 1.0, #pacing/duration - Gradio Slider issue: takes min. rather than value
|
| 95 |
},
|
| 96 |
+
'suno/bark': {
|
| 97 |
+
1: 'Speaker 3 (en)',
|
| 98 |
+
},
|
| 99 |
+
'amphion/Text-to-Speech': {
|
| 100 |
+
1: 'LikeManyWaters',
|
| 101 |
+
},
|
| 102 |
+
'LeeSangHoon/HierSpeech_TTS': {
|
| 103 |
+
1: DEFAULT_VOICE_SAMPLE, # voice sample
|
| 104 |
+
2: 0.333,
|
| 105 |
+
3: 0.333,
|
| 106 |
+
4: 1,
|
| 107 |
+
5: 1,
|
| 108 |
+
6: 0,
|
| 109 |
+
7: 1111,
|
| 110 |
+
},
|
| 111 |
+
'Manmay/tortoise-tts': {
|
| 112 |
+
1: None, # text-from-file; FIXME: cannot skip and doesn't work without
|
| 113 |
+
2: 'angie',
|
| 114 |
+
3: None,
|
| 115 |
+
4: 'No',
|
| 116 |
+
},
|
| 117 |
}
|
| 118 |
|
| 119 |
SPACE_ID = os.getenv('SPACE_ID')
|
|
|
|
| 372 |
'metavoice': 'Apache 2.0',
|
| 373 |
'elevenlabs': 'Proprietary',
|
| 374 |
'whisperspeech': 'MIT',
|
|
|
|
| 375 |
'Pendrokar/xVASynth': 'GPT3',
|
|
|
|
| 376 |
}
|
| 377 |
model_links = {
|
| 378 |
'styletts2': 'https://github.com/yl4579/StyleTTS2',
|