├── audios └── tempfile.mp3 ├── images └── Masahiro.png ├── videos ├── Masahiro.mp4 └── tempfile.mp4 ├── requirements.txt ├── README.md ├── .gitattributes ├── azure_utils.py ├── polly_utils.py └── app.py /audios/tempfile.mp3: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/alitrack/Chat-GPT-LangChain/HEAD/audios/tempfile.mp3 -------------------------------------------------------------------------------- /images/Masahiro.png: -------------------------------------------------------------------------------- 1 | version https://git-lfs.github.com/spec/v1 2 | oid sha256:215bfaa1bdb0ee4852988b29d480e2d1c2d9669eaa907ba25cc2d3dfa6ebfa4e 3 | size 4392529 4 | -------------------------------------------------------------------------------- /videos/Masahiro.mp4: -------------------------------------------------------------------------------- 1 | version https://git-lfs.github.com/spec/v1 2 | oid sha256:ca886517414fae8bcea5a5130ae1d01ef3ead7aed437203abebf032217fd0be6 3 | size 2425212 4 | -------------------------------------------------------------------------------- /videos/tempfile.mp4: -------------------------------------------------------------------------------- 1 | version https://git-lfs.github.com/spec/v1 2 | oid sha256:4579f43ce3f39906372a8f52b0858510321807df2ed98d4a539a991986ab0cc7 3 | size 103305 4 | -------------------------------------------------------------------------------- /requirements.txt: -------------------------------------------------------------------------------- 1 | openai==0.27.0 2 | gradio==3.19.1 3 | google-search-results 4 | google-api-python-client==2.80.0 5 | wolframalpha 6 | langchain==0.0.98 7 | requests==2.28.2 8 | git+https://github.com/openai/whisper.git 9 | boto3==1.26.82 10 | faiss-cpu 11 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | --- 2 | title: GPT+WolframAlpha+Whisper 3 | emoji: 👀 4 | colorFrom: red 5 | colorTo: gray 6 | sdk: gradio 7 | sdk_version: 3.16.1 8 | app_file: app.py 9 | pinned: false 10 | license: apache-2.0 11 | --- 12 | 13 | Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference 14 | -------------------------------------------------------------------------------- /.gitattributes: -------------------------------------------------------------------------------- 1 | *.7z filter=lfs diff=lfs merge=lfs -text 2 | *.arrow filter=lfs diff=lfs merge=lfs -text 3 | *.bin filter=lfs diff=lfs merge=lfs -text 4 | *.bz2 filter=lfs diff=lfs merge=lfs -text 5 | *.ckpt filter=lfs diff=lfs merge=lfs -text 6 | *.ftz filter=lfs diff=lfs merge=lfs -text 7 | *.gz filter=lfs diff=lfs merge=lfs -text 8 | *.h5 filter=lfs diff=lfs merge=lfs -text 9 | *.joblib filter=lfs diff=lfs merge=lfs -text 10 | *.lfs.* filter=lfs diff=lfs merge=lfs -text 11 | *.mlmodel filter=lfs diff=lfs merge=lfs -text 12 | *.model filter=lfs diff=lfs merge=lfs -text 13 | *.msgpack filter=lfs diff=lfs merge=lfs -text 14 | *.npy filter=lfs diff=lfs merge=lfs -text 15 | *.npz filter=lfs diff=lfs merge=lfs -text 16 | *.onnx filter=lfs diff=lfs merge=lfs -text 17 | *.ot filter=lfs diff=lfs merge=lfs -text 18 | *.parquet filter=lfs diff=lfs merge=lfs -text 19 | *.pb filter=lfs diff=lfs merge=lfs -text 20 | *.pickle filter=lfs diff=lfs merge=lfs -text 21 | *.pkl filter=lfs diff=lfs merge=lfs -text 22 | *.pt filter=lfs diff=lfs merge=lfs -text 23 | *.pth filter=lfs diff=lfs merge=lfs -text 24 | *.rar filter=lfs diff=lfs merge=lfs -text 25 | *.safetensors filter=lfs diff=lfs merge=lfs -text 26 | saved_model/**/* filter=lfs diff=lfs merge=lfs -text 27 | *.tar.* filter=lfs diff=lfs merge=lfs -text 28 | *.tflite filter=lfs diff=lfs merge=lfs -text 29 | *.tgz filter=lfs diff=lfs merge=lfs -text 30 | *.wasm filter=lfs diff=lfs merge=lfs -text 31 | *.xz filter=lfs diff=lfs merge=lfs -text 32 | *.zip filter=lfs diff=lfs merge=lfs -text 33 | *.zst filter=lfs diff=lfs merge=lfs -text 34 | *tfevents* filter=lfs diff=lfs merge=lfs -text 35 | *.mp4 filter=lfs diff=lfs merge=lfs -text 36 | *.png filter=lfs diff=lfs merge=lfs -text 37 | -------------------------------------------------------------------------------- /azure_utils.py: -------------------------------------------------------------------------------- 1 | # This class stores Azure voice data. Specifically, the class stores several records containing 2 | # language, lang_code, gender, voice_id and engine. The class also has a method to return the 3 | # voice_id, lang_code and engine given a language and gender. 4 | 5 | NEURAL_ENGINE = "neural" 6 | STANDARD_ENGINE = "standard" 7 | 8 | 9 | class AzureVoiceData: 10 | def get_voice(self, language, gender): 11 | for voice in self.voice_data: 12 | if voice['language'] == language and voice['gender'] == gender: 13 | return voice['azure_voice'] 14 | return None 15 | 16 | def __init__(self): 17 | self.voice_data = [ 18 | {'language': 'Arabic', 19 | 'azure_voice': 'ar-EG-ShakirNeural', 20 | 'gender': 'Male'}, 21 | {'language': 'Arabic (Gulf)', 22 | 'azure_voice': 'ar-KW-FahedNeural', 23 | 'gender': 'Male'}, 24 | {'language': 'Catalan', 25 | 'azure_voice': 'ca-ES-EnricNeural', 26 | 'gender': 'Male'}, 27 | {'language': 'Chinese (Cantonese)', 28 | 'azure_voice': 'yue-CN-YunSongNeural', 29 | 'gender': 'Male'}, 30 | {'language': 'Chinese (Mandarin)', 31 | 'azure_voice': 'zh-CN-YunxiNeural', 32 | 'gender': 'Male'}, 33 | {'language': 'Danish', 34 | 'azure_voice': 'da-DK-JeppeNeural', 35 | 'gender': 'Male'}, 36 | {'language': 'Dutch', 37 | 'azure_voice': 'nl-NL-MaartenNeural', 38 | 'gender': 'Male'}, 39 | {'language': 'English (Australian)', 40 | 'azure_voice': 'en-AU-KenNeural', 41 | 'gender': 'Male'}, 42 | {'language': 'English (British)', 43 | 'azure_voice': 'en-GB-RyanNeural', 44 | 'gender': 'Male'}, 45 | {'language': 'English (Indian)', 46 | 'azure_voice': 'en-IN-PrabhatNeural', 47 | 'gender': 'Male'}, 48 | {'language': 'English (New Zealand)', 49 | 'azure_voice': 'en-NZ-MitchellNeural', 50 | 'gender': 'Male'}, 51 | {'language': 'English (South African)', 52 | 'azure_voice': 'en-ZA-LukeNeural', 53 | 'gender': 'Male'}, 54 | {'language': 'English (US)', 55 | 'azure_voice': 'en-US-ChristopherNeural', 56 | 'gender': 'Male'}, 57 | {'language': 'English (Welsh)', 58 | 'azure_voice': 'cy-GB-AledNeural', 59 | 'gender': 'Male'}, 60 | {'language': 'Finnish', 61 | 'azure_voice': 'fi-FI-HarriNeural', 62 | 'gender': 'Male'}, 63 | {'language': 'French', 64 | 'azure_voice': 'fr-FR-HenriNeural', 65 | 'gender': 'Male'}, 66 | {'language': 'French (Canadian)', 67 | 'azure_voice': 'fr-CA-AntoineNeural', 68 | 'gender': 'Male'}, 69 | {'language': 'German', 70 | 'azure_voice': 'de-DE-KlausNeural', 71 | 'gender': 'Male'}, 72 | {'language': 'German (Austrian)', 73 | 'azure_voice': 'de-AT-JonasNeural', 74 | 'gender': 'Male'}, 75 | {'language': 'Hindi', 76 | 'azure_voice': 'hi-IN-MadhurNeural', 77 | 'gender': 'Male'}, 78 | {'language': 'Icelandic', 79 | 'azure_voice': 'is-IS-GunnarNeural', 80 | 'gender': 'Male'}, 81 | {'language': 'Italian', 82 | 'azure_voice': 'it-IT-GianniNeural', 83 | 'gender': 'Male'}, 84 | {'language': 'Japanese', 85 | 'azure_voice': 'ja-JP-KeitaNeural', 86 | 'gender': 'Male'}, 87 | {'language': 'Korean', 88 | 'azure_voice': 'ko-KR-GookMinNeural', 89 | 'gender': 'Male'}, 90 | {'language': 'Norwegian', 91 | 'azure_voice': 'nb-NO-FinnNeural', 92 | 'gender': 'Male'}, 93 | {'language': 'Polish', 94 | 'azure_voice': 'pl-PL-MarekNeural', 95 | 'gender': 'Male'}, 96 | {'language': 'Portuguese (Brazilian)', 97 | 'azure_voice': 'pt-BR-NicolauNeural', 98 | 'gender': 'Male'}, 99 | {'language': 'Portuguese (European)', 100 | 'azure_voice': 'pt-PT-DuarteNeural', 101 | 'gender': 'Male'}, 102 | {'language': 'Romanian', 103 | 'azure_voice': 'ro-RO-EmilNeural', 104 | 'gender': 'Male'}, 105 | {'language': 'Russian', 106 | 'azure_voice': 'ru-RU-DmitryNeural', 107 | 'gender': 'Male'}, 108 | {'language': 'Spanish (European)', 109 | 'azure_voice': 'es-ES-TeoNeural', 110 | 'gender': 'Male'}, 111 | {'language': 'Spanish (Mexican)', 112 | 'azure_voice': 'es-MX-LibertoNeural', 113 | 'gender': 'Male'}, 114 | {'language': 'Spanish (US)', 115 | 'azure_voice': 'es-US-AlonsoNeural"', 116 | 'gender': 'Male'}, 117 | {'language': 'Swedish', 118 | 'azure_voice': 'sv-SE-MattiasNeural', 119 | 'gender': 'Male'}, 120 | {'language': 'Turkish', 121 | 'azure_voice': 'tr-TR-AhmetNeural', 122 | 'gender': 'Male'}, 123 | {'language': 'Welsh', 124 | 'azure_voice': 'cy-GB-AledNeural', 125 | 'gender': 'Male'}, 126 | ] 127 | 128 | 129 | # Run from the command-line 130 | if __name__ == '__main__': 131 | azure_voice_data = AzureVoiceData() 132 | 133 | azure_voice = azure_voice_data.get_voice('English (US)', 'Male') 134 | print('English (US)', 'Male', azure_voice) 135 | 136 | azure_voice = azure_voice_data.get_voice('English (US)', 'Female') 137 | print('English (US)', 'Female', azure_voice) 138 | 139 | azure_voice = azure_voice_data.get_voice('French', 'Female') 140 | print('French', 'Female', azure_voice) 141 | 142 | azure_voice = azure_voice_data.get_voice('French', 'Male') 143 | print('French', 'Male', azure_voice) 144 | 145 | azure_voice = azure_voice_data.get_voice('Japanese', 'Female') 146 | print('Japanese', 'Female', azure_voice) 147 | 148 | azure_voice = azure_voice_data.get_voice('Japanese', 'Male') 149 | print('Japanese', 'Male', azure_voice) 150 | 151 | azure_voice = azure_voice_data.get_voice('Hindi', 'Female') 152 | print('Hindi', 'Female', azure_voice) 153 | 154 | azure_voice = azure_voice_data.get_voice('Hindi', 'Male') 155 | print('Hindi', 'Male', azure_voice) 156 | -------------------------------------------------------------------------------- /polly_utils.py: -------------------------------------------------------------------------------- 1 | # This class stores Polly voice data. Specifically, the class stores several records containing 2 | # language, lang_code, gender, voice_id and engine. The class also has a method to return the 3 | # voice_id, lang_code and engine given a language and gender. 4 | 5 | NEURAL_ENGINE = "neural" 6 | STANDARD_ENGINE = "standard" 7 | 8 | 9 | class PollyVoiceData: 10 | def get_voice(self, language, gender): 11 | for voice in self.voice_data: 12 | if voice['language'] == language and voice['gender'] == gender: 13 | if voice['neural'] == 'Yes': 14 | return voice['voice_id'], voice['lang_code'], NEURAL_ENGINE 15 | for voice in self.voice_data: 16 | if voice['language'] == language and voice['gender'] == gender: 17 | if voice['standard'] == 'Yes': 18 | return voice['voice_id'], voice['lang_code'], STANDARD_ENGINE 19 | return None, None, None 20 | 21 | def get_whisper_lang_code(self, language): 22 | for voice in self.voice_data: 23 | if voice['language'] == language: 24 | return voice['whisper_lang_code'] 25 | return "en" 26 | 27 | def __init__(self): 28 | self.voice_data = [ 29 | {'language': 'Arabic', 30 | 'lang_code': 'arb', 31 | 'whisper_lang_code': 'ar', 32 | 'voice_id': 'Zeina', 33 | 'gender': 'Female', 34 | 'neural': 'No', 35 | 'standard': 'Yes'}, 36 | {'language': 'Arabic (Gulf)', 37 | 'lang_code': 'ar-AE', 38 | 'whisper_lang_code': 'ar', 39 | 'voice_id': 'Hala', 40 | 'gender': 'Female', 41 | 'neural': 'Yes', 42 | 'standard': 'No'}, 43 | {'language': 'Catalan', 44 | 'lang_code': 'ca-ES', 45 | 'whisper_lang_code': 'ca', 46 | 'voice_id': 'Arlet', 47 | 'gender': 'Female', 48 | 'neural': 'Yes', 49 | 'standard': 'No'}, 50 | {'language': 'Chinese (Cantonese)', 51 | 'lang_code': 'yue-CN', 52 | 'whisper_lang_code': 'zh', 53 | 'voice_id': 'Hiujin', 54 | 'gender': 'Female', 55 | 'neural': 'Yes', 56 | 'standard': 'No'}, 57 | {'language': 'Chinese (Mandarin)', 58 | 'lang_code': 'cmn-CN', 59 | 'whisper_lang_code': 'zh', 60 | 'voice_id': 'Zhiyu', 61 | 'gender': 'Female', 62 | 'neural': 'Yes', 63 | 'standard': 'No'}, 64 | {'language': 'Danish', 65 | 'lang_code': 'da-DK', 66 | 'whisper_lang_code': 'da', 67 | 'voice_id': 'Naja', 68 | 'gender': 'Female', 69 | 'neural': 'No', 70 | 'standard': 'Yes'}, 71 | {'language': 'Danish', 72 | 'lang_code': 'da-DK', 73 | 'whisper_lang_code': 'da', 74 | 'voice_id': 'Mads', 75 | 'gender': 'Male', 76 | 'neural': 'No', 77 | 'standard': 'Yes'}, 78 | {'language': 'Dutch', 79 | 'lang_code': 'nl-NL', 80 | 'whisper_lang_code': 'nl', 81 | 'voice_id': 'Laura', 82 | 'gender': 'Female', 83 | 'neural': 'Yes', 84 | 'standard': 'No'}, 85 | {'language': 'Dutch', 86 | 'lang_code': 'nl-NL', 87 | 'whisper_lang_code': 'nl', 88 | 'voice_id': 'Lotte', 89 | 'gender': 'Female', 90 | 'neural': 'No', 91 | 'standard': 'Yes'}, 92 | {'language': 'Dutch', 93 | 'lang_code': 'nl-NL', 94 | 'whisper_lang_code': 'nl', 95 | 'voice_id': 'Ruben', 96 | 'gender': 'Male', 97 | 'neural': 'No', 98 | 'standard': 'Yes'}, 99 | {'language': 'English (Australian)', 100 | 'lang_code': 'en-AU', 101 | 'whisper_lang_code': 'en', 102 | 'voice_id': 'Nicole', 103 | 'gender': 'Female', 104 | 'neural': 'No', 105 | 'standard': 'Yes'}, 106 | {'language': 'English (Australian)', 107 | 'lang_code': 'en-AU', 108 | 'whisper_lang_code': 'en', 109 | 'voice_id': 'Olivia', 110 | 'gender': 'Female', 111 | 'neural': 'Yes', 112 | 'standard': 'No'}, 113 | {'language': 'English (Australian)', 114 | 'lang_code': 'en-AU', 115 | 'whisper_lang_code': 'en', 116 | 'voice_id': 'Russell', 117 | 'gender': 'Male', 118 | 'neural': 'No', 119 | 'standard': 'Yes'}, 120 | {'language': 'English (British)', 121 | 'lang_code': 'en-GB', 122 | 'whisper_lang_code': 'en', 123 | 'voice_id': 'Amy', 124 | 'gender': 'Female', 125 | 'neural': 'Yes', 126 | 'standard': 'Yes'}, 127 | {'language': 'English (British)', 128 | 'lang_code': 'en-GB', 129 | 'whisper_lang_code': 'en', 130 | 'voice_id': 'Emma', 131 | 'gender': 'Female', 132 | 'neural': 'Yes', 133 | 'standard': 'Yes'}, 134 | {'language': 'English (British)', 135 | 'lang_code': 'en-GB', 136 | 'whisper_lang_code': 'en', 137 | 'voice_id': 'Brian', 138 | 'gender': 'Male', 139 | 'neural': 'Yes', 140 | 'standard': 'Yes'}, 141 | {'language': 'English (British)', 142 | 'lang_code': 'en-GB', 143 | 'whisper_lang_code': 'en', 144 | 'voice_id': 'Arthur', 145 | 'gender': 'Male', 146 | 'neural': 'Yes', 147 | 'standard': 'No'}, 148 | {'language': 'English (Indian)', 149 | 'lang_code': 'en-IN', 150 | 'whisper_lang_code': 'en', 151 | 'voice_id': 'Aditi', 152 | 'gender': 'Female', 153 | 'neural': 'No', 154 | 'standard': 'Yes'}, 155 | {'language': 'English (Indian)', 156 | 'lang_code': 'en-IN', 157 | 'whisper_lang_code': 'en', 158 | 'voice_id': 'Raveena', 159 | 'gender': 'Female', 160 | 'neural': 'No', 161 | 'standard': 'Yes'}, 162 | {'language': 'English (Indian)', 163 | 'lang_code': 'en-IN', 164 | 'whisper_lang_code': 'en', 165 | 'voice_id': 'Kajal', 166 | 'gender': 'Female', 167 | 'neural': 'Yes', 168 | 'standard': 'No'}, 169 | {'language': 'English (New Zealand)', 170 | 'lang_code': 'en-NZ', 171 | 'whisper_lang_code': 'en', 172 | 'voice_id': 'Aria', 173 | 'gender': 'Female', 174 | 'neural': 'Yes', 175 | 'standard': 'No'}, 176 | {'language': 'English (South African)', 177 | 'lang_code': 'en-ZA', 178 | 'whisper_lang_code': 'en', 179 | 'voice_id': 'Ayanda', 180 | 'gender': 'Female', 181 | 'neural': 'Yes', 182 | 'standard': 'No'}, 183 | {'language': 'English (US)', 184 | 'lang_code': 'en-US', 185 | 'whisper_lang_code': 'en', 186 | 'voice_id': 'Ivy', 187 | 'gender': 'Female (child)', 188 | 'neural': 'Yes', 189 | 'standard': 'Yes'}, 190 | {'language': 'English (US)', 191 | 'lang_code': 'en-US', 192 | 'whisper_lang_code': 'en', 193 | 'voice_id': 'Joanna', 194 | 'gender': 'Female', 195 | 'neural': 'Yes', 196 | 'standard': 'Yes'}, 197 | {'language': 'English (US)', 198 | 'lang_code': 'en-US', 199 | 'whisper_lang_code': 'en', 200 | 'voice_id': 'Kendra', 201 | 'gender': 'Female', 202 | 'neural': 'Yes', 203 | 'standard': 'Yes'}, 204 | {'language': 'English (US)', 205 | 'lang_code': 'en-US', 206 | 'whisper_lang_code': 'en', 207 | 'voice_id': 'Kimberly', 208 | 'gender': 'Female', 209 | 'neural': 'Yes', 210 | 'standard': 'Yes'}, 211 | {'language': 'English (US)', 212 | 'lang_code': 'en-US', 213 | 'whisper_lang_code': 'en', 214 | 'voice_id': 'Salli', 215 | 'gender': 'Female', 216 | 'neural': 'Yes', 217 | 'standard': 'Yes'}, 218 | {'language': 'English (US)', 219 | 'lang_code': 'en-US', 220 | 'whisper_lang_code': 'en', 221 | 'voice_id': 'Joey', 222 | 'gender': 'Male', 223 | 'neural': 'Yes', 224 | 'standard': 'Yes'}, 225 | {'language': 'English (US)', 226 | 'lang_code': 'en-US', 227 | 'whisper_lang_code': 'en', 228 | 'voice_id': 'Justin', 229 | 'gender': 'Male (child)', 230 | 'neural': 'Yes', 231 | 'standard': 'Yes'}, 232 | {'language': 'English (US)', 233 | 'lang_code': 'en-US', 234 | 'whisper_lang_code': 'en', 235 | 'voice_id': 'Kevin', 236 | 'gender': 'Male (child)', 237 | 'neural': 'Yes', 238 | 'standard': 'No'}, 239 | {'language': 'English (US)', 240 | 'lang_code': 'en-US', 241 | 'whisper_lang_code': 'en', 242 | 'voice_id': 'Matthew', 243 | 'gender': 'Male', 244 | 'neural': 'Yes', 245 | 'standard': 'Yes'}, 246 | {'language': 'English (Welsh)', 247 | 'lang_code': 'en-GB-WLS', 248 | 'whisper_lang_code': 'en', 249 | 'voice_id': 'Geraint', 250 | 'gender': 'Male', 251 | 'neural': 'No', 252 | 'standard': 'Yes'}, 253 | {'language': 'Finnish', 254 | 'lang_code': 'fi-FI', 255 | 'whisper_lang_code': 'fi', 256 | 'voice_id': 'Suvi', 257 | 'gender': 'Female', 258 | 'neural': 'Yes', 259 | 'standard': 'No'}, 260 | {'language': 'French', 261 | 'lang_code': 'fr-FR', 262 | 'whisper_lang_code': 'fr', 263 | 'voice_id': 'Celine', 264 | 'gender': 'Female', 265 | 'neural': 'No', 266 | 'standard': 'Yes'}, 267 | {'language': 'French', 268 | 'lang_code': 'fr-FR', 269 | 'whisper_lang_code': 'fr', 270 | 'voice_id': 'Lea', 271 | 'gender': 'Female', 272 | 'neural': 'Yes', 273 | 'standard': 'Yes'}, 274 | {'language': 'French', 275 | 'lang_code': 'fr-FR', 276 | 'whisper_lang_code': 'fr', 277 | 'voice_id': 'Mathieu', 278 | 'gender': 'Male', 279 | 'neural': 'No', 280 | 'standard': 'Yes'}, 281 | {'language': 'French (Canadian)', 282 | 'lang_code': 'fr-CA', 283 | 'whisper_lang_code': 'fr', 284 | 'voice_id': 'Chantal', 285 | 'gender': 'Female', 286 | 'neural': 'No', 287 | 'standard': 'Yes'}, 288 | {'language': 'French (Canadian)', 289 | 'lang_code': 'fr-CA', 290 | 'whisper_lang_code': 'fr', 291 | 'voice_id': 'Gabrielle', 292 | 'gender': 'Female', 293 | 'neural': 'Yes', 294 | 'standard': 'No'}, 295 | {'language': 'French (Canadian)', 296 | 'lang_code': 'fr-CA', 297 | 'whisper_lang_code': 'fr', 298 | 'voice_id': 'Liam', 299 | 'gender': 'Male', 300 | 'neural': 'Yes', 301 | 'standard': 'No'}, 302 | {'language': 'German', 303 | 'lang_code': 'de-DE', 304 | 'whisper_lang_code': 'de', 305 | 'voice_id': 'Marlene', 306 | 'gender': 'Female', 307 | 'neural': 'No', 308 | 'standard': 'Yes'}, 309 | {'language': 'German', 310 | 'lang_code': 'de-DE', 311 | 'whisper_lang_code': 'de', 312 | 'voice_id': 'Vicki', 313 | 'gender': 'Female', 314 | 'neural': 'Yes', 315 | 'standard': 'Yes'}, 316 | {'language': 'German', 317 | 'lang_code': 'de-DE', 318 | 'whisper_lang_code': 'de', 319 | 'voice_id': 'Hans', 320 | 'gender': 'Male', 321 | 'neural': 'No', 322 | 'standard': 'Yes'}, 323 | {'language': 'German', 324 | 'lang_code': 'de-DE', 325 | 'whisper_lang_code': 'de', 326 | 'voice_id': 'Daniel', 327 | 'gender': 'Male', 328 | 'neural': 'Yes', 329 | 'standard': 'No'}, 330 | {'language': 'German (Austrian)', 331 | 'lang_code': 'de-AT', 332 | 'whisper_lang_code': 'de', 333 | 'voice_id': 'Hannah', 334 | 'gender': 'Female', 335 | 'neural': 'Yes', 336 | 'standard': 'No'}, 337 | {'language': 'Hindi', 338 | 'lang_code': 'hi-IN', 339 | 'whisper_lang_code': 'hi', 340 | 'voice_id': 'Aditi', 341 | 'gender': 'Female', 342 | 'neural': 'No', 343 | 'standard': 'Yes'}, 344 | {'language': 'Hindi', 345 | 'lang_code': 'hi-IN', 346 | 'whisper_lang_code': 'hi', 347 | 'voice_id': 'Kajal', 348 | 'gender': 'Female', 349 | 'neural': 'Yes', 350 | 'standard': 'No'}, 351 | {'language': 'Icelandic', 352 | 'lang_code': 'is-IS', 353 | 'whisper_lang_code': 'is', 354 | 'voice_id': 'Dora', 355 | 'gender': 'Female', 356 | 'neural': 'No', 357 | 'standard': 'Yes'}, 358 | {'language': 'Icelandic', 359 | 'lang_code': 'is-IS', 360 | 'whisper_lang_code': 'is', 361 | 'voice_id': 'Karl', 362 | 'gender': 'Male', 363 | 'neural': 'No', 364 | 'standard': 'Yes'}, 365 | {'language': 'Italian', 366 | 'lang_code': 'it-IT', 367 | 'whisper_lang_code': 'it', 368 | 'voice_id': 'Carla', 369 | 'gender': 'Female', 370 | 'neural': 'No', 371 | 'standard': 'Yes'}, 372 | {'language': 'Italian', 373 | 'lang_code': 'it-IT', 374 | 'whisper_lang_code': 'it', 375 | 'voice_id': 'Bianca', 376 | 'gender': 'Female', 377 | 'neural': 'Yes', 378 | 'standard': 'Yes'}, 379 | {'language': 'Japanese', 380 | 'lang_code': 'ja-JP', 381 | 'whisper_lang_code': 'ja', 382 | 'voice_id': 'Mizuki', 383 | 'gender': 'Female', 384 | 'neural': 'No', 385 | 'standard': 'Yes'}, 386 | {'language': 'Japanese', 387 | 'lang_code': 'ja-JP', 388 | 'whisper_lang_code': 'ja', 389 | 'voice_id': 'Takumi', 390 | 'gender': 'Male', 391 | 'neural': 'Yes', 392 | 'standard': 'Yes'}, 393 | {'language': 'Korean', 394 | 'lang_code': 'ko-KR', 395 | 'whisper_lang_code': 'ko', 396 | 'voice_id': 'Seoyeon', 397 | 'gender': 'Female', 398 | 'neural': 'Yes', 399 | 'standard': 'Yes'}, 400 | {'language': 'Norwegian', 401 | 'lang_code': 'nb-NO', 402 | 'whisper_lang_code': 'no', 403 | 'voice_id': 'Liv', 404 | 'gender': 'Female', 405 | 'neural': 'No', 406 | 'standard': 'Yes'}, 407 | {'language': 'Norwegian', 408 | 'lang_code': 'nb-NO', 409 | 'whisper_lang_code': 'no', 410 | 'voice_id': 'Ida', 411 | 'gender': 'Female', 412 | 'neural': 'Yes', 413 | 'standard': 'No'}, 414 | {'language': 'Polish', 415 | 'lang_code': 'pl-PL', 416 | 'whisper_lang_code': 'pl', 417 | 'voice_id': 'Ewa', 418 | 'gender': 'Female', 419 | 'neural': 'No', 420 | 'standard': 'Yes'}, 421 | {'language': 'Polish', 422 | 'lang_code': 'pl-PL', 423 | 'whisper_lang_code': 'pl', 424 | 'voice_id': 'Maja', 425 | 'gender': 'Female', 426 | 'neural': 'No', 427 | 'standard': 'Yes'}, 428 | {'language': 'Polish', 429 | 'lang_code': 'pl-PL', 430 | 'whisper_lang_code': 'pl', 431 | 'voice_id': 'Jacek', 432 | 'gender': 'Male', 433 | 'neural': 'No', 434 | 'standard': 'Yes'}, 435 | {'language': 'Polish', 436 | 'lang_code': 'pl-PL', 437 | 'whisper_lang_code': 'pl', 438 | 'voice_id': 'Jan', 439 | 'gender': 'Male', 440 | 'neural': 'No', 441 | 'standard': 'Yes'}, 442 | {'language': 'Polish', 443 | 'lang_code': 'pl-PL', 444 | 'whisper_lang_code': 'pl', 445 | 'voice_id': 'Ola', 446 | 'gender': 'Female', 447 | 'neural': 'Yes', 448 | 'standard': 'No'}, 449 | {'language': 'Portuguese (Brazilian)', 450 | 'lang_code': 'pt-BR', 451 | 'whisper_lang_code': 'pt', 452 | 'voice_id': 'Camila', 453 | 'gender': 'Female', 454 | 'neural': 'Yes', 455 | 'standard': 'Yes'}, 456 | {'language': 'Portuguese (Brazilian)', 457 | 'lang_code': 'pt-BR', 458 | 'whisper_lang_code': 'pt', 459 | 'voice_id': 'Vitoria', 460 | 'gender': 'Female', 461 | 'neural': 'Yes', 462 | 'standard': 'Yes'}, 463 | {'language': 'Portuguese (Brazilian)', 464 | 'lang_code': 'pt-BR', 465 | 'whisper_lang_code': 'pt', 466 | 'voice_id': 'Ricardo', 467 | 'gender': 'Male', 468 | 'neural': 'No', 469 | 'standard': 'Yes'}, 470 | {'language': 'Portuguese (European)', 471 | 'lang_code': 'pt-PT', 472 | 'whisper_lang_code': 'pt', 473 | 'voice_id': 'Ines', 474 | 'gender': 'Female', 475 | 'neural': 'Yes', 476 | 'standard': 'Yes'}, 477 | {'language': 'Portuguese (European)', 478 | 'lang_code': 'pt-PT', 479 | 'whisper_lang_code': 'pt', 480 | 'voice_id': 'Cristiano', 481 | 'gender': 'Male', 482 | 'neural': 'No', 483 | 'standard': 'Yes'}, 484 | {'language': 'Romanian', 485 | 'lang_code': 'ro-RO', 486 | 'whisper_lang_code': 'ro', 487 | 'voice_id': 'Carmen', 488 | 'gender': 'Female', 489 | 'neural': 'No', 490 | 'standard': 'Yes'}, 491 | {'language': 'Russian', 492 | 'lang_code': 'ru-RU', 493 | 'whisper_lang_code': 'ru', 494 | 'voice_id': 'Tatyana', 495 | 'gender': 'Female', 496 | 'neural': 'No', 497 | 'standard': 'Yes'}, 498 | {'language': 'Russian', 499 | 'lang_code': 'ru-RU', 500 | 'whisper_lang_code': 'ru', 501 | 'voice_id': 'Maxim', 502 | 'gender': 'Male', 503 | 'neural': 'No', 504 | 'standard': 'Yes'}, 505 | {'language': 'Spanish (European)', 506 | 'lang_code': 'es-ES', 507 | 'whisper_lang_code': 'es', 508 | 'voice_id': 'Conchita', 509 | 'gender': 'Female', 510 | 'neural': 'No', 511 | 'standard': 'Yes'}, 512 | {'language': 'Spanish (European)', 513 | 'lang_code': 'es-ES', 514 | 'whisper_lang_code': 'es', 515 | 'voice_id': 'Lucia', 516 | 'gender': 'Female', 517 | 'neural': 'Yes', 518 | 'standard': 'Yes'}, 519 | {'language': 'Spanish (European)', 520 | 'lang_code': 'es-ES', 521 | 'whisper_lang_code': 'es', 522 | 'voice_id': 'Enrique', 523 | 'gender': 'Male', 524 | 'neural': 'No', 525 | 'standard': 'Yes'}, 526 | {'language': 'Spanish (Mexican)', 527 | 'lang_code': 'es-MX', 528 | 'whisper_lang_code': 'es', 529 | 'voice_id': 'Mia', 530 | 'gender': 'Female', 531 | 'neural': 'Yes', 532 | 'standard': 'Yes'}, 533 | {'language': 'Spanish (US)', 534 | 'lang_code': 'es-US', 535 | 'whisper_lang_code': 'es', 536 | 'voice_id': 'Lupe', 537 | 'gender': 'Female', 538 | 'neural': 'Yes', 539 | 'standard': 'Yes'}, 540 | {'language': 'Spanish (US)', 541 | 'lang_code': 'es-US', 542 | 'whisper_lang_code': 'es', 543 | 'voice_id': 'Penelope', 544 | 'gender': 'Female', 545 | 'neural': 'No', 546 | 'standard': 'Yes'}, 547 | {'language': 'Spanish (US)', 548 | 'lang_code': 'es-US', 549 | 'whisper_lang_code': 'es', 550 | 'voice_id': 'Miguel', 551 | 'gender': 'Male', 552 | 'neural': 'No', 553 | 'standard': 'Yes'}, 554 | {'language': 'Spanish (US)', 555 | 'lang_code': 'es-US', 556 | 'whisper_lang_code': 'es', 557 | 'voice_id': 'Pedro', 558 | 'gender': 'Male', 559 | 'neural': 'Yes', 560 | 'standard': 'No'}, 561 | {'language': 'Swedish', 562 | 'lang_code': 'sv-SE', 563 | 'whisper_lang_code': 'sv', 564 | 'voice_id': 'Astrid', 565 | 'gender': 'Female', 566 | 'neural': 'No', 567 | 'standard': 'Yes'}, 568 | {'language': 'Swedish', 569 | 'lang_code': 'sv-SE', 570 | 'whisper_lang_code': 'sv', 571 | 'voice_id': 'Elin', 572 | 'gender': 'Female', 573 | 'neural': 'Yes', 574 | 'standard': 'No'}, 575 | {'language': 'Turkish', 576 | 'lang_code': 'tr-TR', 577 | 'whisper_lang_code': 'tr', 578 | 'voice_id': 'Filiz', 579 | 'gender': 'Female', 580 | 'neural': 'No', 581 | 'standard': 'Yes'}, 582 | {'language': 'Welsh', 583 | 'lang_code': 'cy-GB', 584 | 'whisper_lang_code': 'cy', 585 | 'voice_id': 'Gwyneth', 586 | 'gender': 'Female', 587 | 'neural': 'No', 588 | 'standard': 'Yes'} 589 | ] 590 | 591 | 592 | # Run from the command-line 593 | if __name__ == '__main__': 594 | polly_voice_data = PollyVoiceData() 595 | 596 | voice_id, language_code, engine = polly_voice_data.get_voice('English (US)', 'Male') 597 | print('English (US)', 'Male', voice_id, language_code, engine) 598 | 599 | voice_id, language_code, engine = polly_voice_data.get_voice('English (US)', 'Female') 600 | print('English (US)', 'Female', voice_id, language_code, engine) 601 | 602 | voice_id, language_code, engine = polly_voice_data.get_voice('French', 'Female') 603 | print('French', 'Female', voice_id, language_code, engine) 604 | 605 | voice_id, language_code, engine = polly_voice_data.get_voice('French', 'Male') 606 | print('French', 'Male', voice_id, language_code, engine) 607 | 608 | voice_id, language_code, engine = polly_voice_data.get_voice('Japanese', 'Female') 609 | print('Japanese', 'Female', voice_id, language_code, engine) 610 | 611 | voice_id, language_code, engine = polly_voice_data.get_voice('Japanese', 'Male') 612 | print('Japanese', 'Male', voice_id, language_code, engine) 613 | 614 | voice_id, language_code, engine = polly_voice_data.get_voice('Hindi', 'Female') 615 | print('Hindi', 'Female', voice_id, language_code, engine) 616 | 617 | voice_id, language_code, engine = polly_voice_data.get_voice('Hindi', 'Male') 618 | print('Hindi', 'Male', voice_id, language_code, engine) 619 | 620 | whisper_lang_code = polly_voice_data.get_whisper_lang_code('English (US)') 621 | print('English (US) whisper_lang_code:', whisper_lang_code) 622 | 623 | whisper_lang_code = polly_voice_data.get_whisper_lang_code('Chinese (Mandarin)') 624 | print('Chinese (Mandarin) whisper_lang_code:', whisper_lang_code) 625 | 626 | whisper_lang_code = polly_voice_data.get_whisper_lang_code('Norwegian') 627 | print('Norwegian whisper_lang_code:', whisper_lang_code) 628 | 629 | whisper_lang_code = polly_voice_data.get_whisper_lang_code('Dutch') 630 | print('Dutch whisper_lang_code:', whisper_lang_code) 631 | 632 | whisper_lang_code = polly_voice_data.get_whisper_lang_code('Foo') 633 | print('Foo whisper_lang_code:', whisper_lang_code) 634 | 635 | 636 | -------------------------------------------------------------------------------- /app.py: -------------------------------------------------------------------------------- 1 | import io 2 | import os 3 | import ssl 4 | from contextlib import closing 5 | from typing import Optional, Tuple 6 | import datetime 7 | 8 | import boto3 9 | import gradio as gr 10 | import requests 11 | 12 | # UNCOMMENT TO USE WHISPER 13 | import warnings 14 | import whisper 15 | 16 | from langchain import ConversationChain, LLMChain 17 | 18 | from langchain.agents import load_tools, initialize_agent 19 | from langchain.chains.conversation.memory import ConversationBufferMemory 20 | from langchain.llms import OpenAI 21 | from threading import Lock 22 | 23 | # Console to variable 24 | from io import StringIO 25 | import sys 26 | import re 27 | 28 | from openai.error import AuthenticationError, InvalidRequestError, RateLimitError 29 | 30 | # Pertains to Express-inator functionality 31 | from langchain.prompts import PromptTemplate 32 | 33 | from polly_utils import PollyVoiceData, NEURAL_ENGINE 34 | from azure_utils import AzureVoiceData 35 | 36 | # Pertains to question answering functionality 37 | from langchain.embeddings.openai import OpenAIEmbeddings 38 | from langchain.text_splitter import CharacterTextSplitter 39 | from langchain.vectorstores.faiss import FAISS 40 | from langchain.docstore.document import Document 41 | from langchain.chains.question_answering import load_qa_chain 42 | 43 | news_api_key = os.environ["NEWS_API_KEY"] 44 | tmdb_bearer_token = os.environ["TMDB_BEARER_TOKEN"] 45 | 46 | TOOLS_LIST = ['serpapi', 'wolfram-alpha', 'pal-math', 'pal-colored-objects'] #'google-search','news-api','tmdb-api','open-meteo-api' 47 | TOOLS_DEFAULT_LIST = ['serpapi', 'pal-math'] 48 | BUG_FOUND_MSG = "Congratulations, you've found a bug in this application!" 49 | # AUTH_ERR_MSG = "Please paste your OpenAI key from openai.com to use this application. It is not necessary to hit a button or key after pasting it." 50 | AUTH_ERR_MSG = "Please paste your OpenAI key from openai.com to use this application. " 51 | MAX_TOKENS = 512 52 | 53 | LOOPING_TALKING_HEAD = "videos/Masahiro.mp4" 54 | TALKING_HEAD_WIDTH = "192" 55 | MAX_TALKING_HEAD_TEXT_LENGTH = 155 56 | 57 | # Pertains to Express-inator functionality 58 | NUM_WORDS_DEFAULT = 0 59 | MAX_WORDS = 400 60 | FORMALITY_DEFAULT = "N/A" 61 | TEMPERATURE_DEFAULT = 0.5 62 | EMOTION_DEFAULT = "N/A" 63 | LANG_LEVEL_DEFAULT = "N/A" 64 | TRANSLATE_TO_DEFAULT = "N/A" 65 | LITERARY_STYLE_DEFAULT = "N/A" 66 | PROMPT_TEMPLATE = PromptTemplate( 67 | input_variables=["original_words", "num_words", "formality", "emotions", "lang_level", "translate_to", 68 | "literary_style"], 69 | template="Restate {num_words}{formality}{emotions}{lang_level}{translate_to}{literary_style}the following: \n{original_words}\n", 70 | ) 71 | 72 | POLLY_VOICE_DATA = PollyVoiceData() 73 | AZURE_VOICE_DATA = AzureVoiceData() 74 | 75 | # Pertains to WHISPER functionality 76 | WHISPER_DETECT_LANG = "Detect language" 77 | 78 | 79 | # UNCOMMENT TO USE WHISPER 80 | warnings.filterwarnings("ignore") 81 | WHISPER_MODEL = whisper.load_model("tiny") 82 | print("WHISPER_MODEL", WHISPER_MODEL) 83 | 84 | 85 | # UNCOMMENT TO USE WHISPER 86 | def transcribe(aud_inp, whisper_lang): 87 | if aud_inp is None: 88 | return "" 89 | aud = whisper.load_audio(aud_inp) 90 | aud = whisper.pad_or_trim(aud) 91 | mel = whisper.log_mel_spectrogram(aud).to(WHISPER_MODEL.device) 92 | _, probs = WHISPER_MODEL.detect_language(mel) 93 | options = whisper.DecodingOptions() 94 | if whisper_lang != WHISPER_DETECT_LANG: 95 | whisper_lang_code = POLLY_VOICE_DATA.get_whisper_lang_code(whisper_lang) 96 | options = whisper.DecodingOptions(language=whisper_lang_code) 97 | result = whisper.decode(WHISPER_MODEL, mel, options) 98 | print("result.text", result.text) 99 | result_text = "" 100 | if result and result.text: 101 | result_text = result.text 102 | return result_text 103 | 104 | 105 | # Temporarily address Wolfram Alpha SSL certificate issue 106 | ssl._create_default_https_context = ssl._create_unverified_context 107 | 108 | 109 | # TEMPORARY FOR TESTING 110 | def transcribe_dummy(aud_inp_tb, whisper_lang): 111 | if aud_inp_tb is None: 112 | return "" 113 | # aud = whisper.load_audio(aud_inp) 114 | # aud = whisper.pad_or_trim(aud) 115 | # mel = whisper.log_mel_spectrogram(aud).to(WHISPER_MODEL.device) 116 | # _, probs = WHISPER_MODEL.detect_language(mel) 117 | # options = whisper.DecodingOptions() 118 | # options = whisper.DecodingOptions(language="ja") 119 | # result = whisper.decode(WHISPER_MODEL, mel, options) 120 | result_text = "Whisper will detect language" 121 | if whisper_lang != WHISPER_DETECT_LANG: 122 | whisper_lang_code = POLLY_VOICE_DATA.get_whisper_lang_code(whisper_lang) 123 | result_text = f"Whisper will use lang code: {whisper_lang_code}" 124 | print("result_text", result_text) 125 | return aud_inp_tb 126 | 127 | 128 | # Pertains to Express-inator functionality 129 | def transform_text(desc, express_chain, num_words, formality, 130 | anticipation_level, joy_level, trust_level, 131 | fear_level, surprise_level, sadness_level, disgust_level, anger_level, 132 | lang_level, translate_to, literary_style): 133 | num_words_prompt = "" 134 | if num_words and int(num_words) != 0: 135 | num_words_prompt = "using up to " + str(num_words) + " words, " 136 | 137 | # Change some arguments to lower case 138 | formality = formality.lower() 139 | anticipation_level = anticipation_level.lower() 140 | joy_level = joy_level.lower() 141 | trust_level = trust_level.lower() 142 | fear_level = fear_level.lower() 143 | surprise_level = surprise_level.lower() 144 | sadness_level = sadness_level.lower() 145 | disgust_level = disgust_level.lower() 146 | anger_level = anger_level.lower() 147 | 148 | formality_str = "" 149 | if formality != "n/a": 150 | formality_str = "in a " + formality + " manner, " 151 | 152 | # put all emotions into a list 153 | emotions = [] 154 | if anticipation_level != "n/a": 155 | emotions.append(anticipation_level) 156 | if joy_level != "n/a": 157 | emotions.append(joy_level) 158 | if trust_level != "n/a": 159 | emotions.append(trust_level) 160 | if fear_level != "n/a": 161 | emotions.append(fear_level) 162 | if surprise_level != "n/a": 163 | emotions.append(surprise_level) 164 | if sadness_level != "n/a": 165 | emotions.append(sadness_level) 166 | if disgust_level != "n/a": 167 | emotions.append(disgust_level) 168 | if anger_level != "n/a": 169 | emotions.append(anger_level) 170 | 171 | emotions_str = "" 172 | if len(emotions) > 0: 173 | if len(emotions) == 1: 174 | emotions_str = "with emotion of " + emotions[0] + ", " 175 | else: 176 | emotions_str = "with emotions of " + ", ".join(emotions[:-1]) + " and " + emotions[-1] + ", " 177 | 178 | lang_level_str = "" 179 | if lang_level != LANG_LEVEL_DEFAULT: 180 | lang_level_str = "at a level that a person in " + lang_level + " can easily comprehend, " if translate_to == TRANSLATE_TO_DEFAULT else "" 181 | 182 | translate_to_str = "" 183 | if translate_to != TRANSLATE_TO_DEFAULT: 184 | translate_to_str = "translated to " + translate_to + ( 185 | "" if lang_level == TRANSLATE_TO_DEFAULT else " at a level that a person in " + lang_level + " can easily comprehend") + ", " 186 | 187 | literary_style_str = "" 188 | if literary_style != LITERARY_STYLE_DEFAULT: 189 | if literary_style == "Prose": 190 | literary_style_str = "as prose, " 191 | if literary_style == "Story": 192 | literary_style_str = "as a story, " 193 | elif literary_style == "Summary": 194 | literary_style_str = "as a summary, " 195 | elif literary_style == "Outline": 196 | literary_style_str = "as an outline numbers and lower case letters, " 197 | elif literary_style == "Bullets": 198 | literary_style_str = "as bullet points using bullets, " 199 | elif literary_style == "Poetry": 200 | literary_style_str = "as a poem, " 201 | elif literary_style == "Haiku": 202 | literary_style_str = "as a haiku, " 203 | elif literary_style == "Limerick": 204 | literary_style_str = "as a limerick, " 205 | elif literary_style == "Rap": 206 | literary_style_str = "as a rap, " 207 | elif literary_style == "Joke": 208 | literary_style_str = "as a very funny joke with a setup and punchline, " 209 | elif literary_style == "Knock-knock": 210 | literary_style_str = "as a very funny knock-knock joke, " 211 | elif literary_style == "FAQ": 212 | literary_style_str = "as a FAQ with several questions and answers, " 213 | 214 | formatted_prompt = PROMPT_TEMPLATE.format( 215 | original_words=desc, 216 | num_words=num_words_prompt, 217 | formality=formality_str, 218 | emotions=emotions_str, 219 | lang_level=lang_level_str, 220 | translate_to=translate_to_str, 221 | literary_style=literary_style_str 222 | ) 223 | 224 | trans_instr = num_words_prompt + formality_str + emotions_str + lang_level_str + translate_to_str + literary_style_str 225 | if express_chain and len(trans_instr.strip()) > 0: 226 | generated_text = express_chain.run( 227 | {'original_words': desc, 'num_words': num_words_prompt, 'formality': formality_str, 228 | 'emotions': emotions_str, 'lang_level': lang_level_str, 'translate_to': translate_to_str, 229 | 'literary_style': literary_style_str}).strip() 230 | else: 231 | print("Not transforming text") 232 | generated_text = desc 233 | 234 | # replace all newlines with
in generated_text 235 | generated_text = generated_text.replace("\n", "\n\n") 236 | 237 | prompt_plus_generated = "GPT prompt: " + formatted_prompt + "\n\n" + generated_text 238 | 239 | print("\n==== date/time: " + str(datetime.datetime.now() - datetime.timedelta(hours=5)) + " ====") 240 | print("prompt_plus_generated: " + prompt_plus_generated) 241 | 242 | return generated_text 243 | 244 | 245 | def load_chain(tools_list, llm): 246 | chain = None 247 | express_chain = None 248 | memory = None 249 | if llm: 250 | print("\ntools_list", tools_list) 251 | tool_names = tools_list 252 | tools = load_tools(tool_names, llm=llm, news_api_key=news_api_key, tmdb_bearer_token=tmdb_bearer_token) 253 | 254 | memory = ConversationBufferMemory(memory_key="chat_history") 255 | 256 | chain = initialize_agent(tools, llm, agent="conversational-react-description", verbose=True, memory=memory) 257 | express_chain = LLMChain(llm=llm, prompt=PROMPT_TEMPLATE, verbose=True) 258 | return chain, express_chain, memory 259 | 260 | 261 | def set_openai_api_key(api_key): 262 | """Set the api key and return chain. 263 | If no api_key, then None is returned. 264 | """ 265 | if api_key and api_key.startswith("sk-") and len(api_key) > 50: 266 | os.environ["OPENAI_API_KEY"] = api_key 267 | print("\n\n ++++++++++++++ Setting OpenAI API key ++++++++++++++ \n\n") 268 | print(str(datetime.datetime.now()) + ": Before OpenAI, OPENAI_API_KEY length: " + str( 269 | len(os.environ["OPENAI_API_KEY"]))) 270 | llm = OpenAI(temperature=0, max_tokens=MAX_TOKENS) 271 | print(str(datetime.datetime.now()) + ": After OpenAI, OPENAI_API_KEY length: " + str( 272 | len(os.environ["OPENAI_API_KEY"]))) 273 | chain, express_chain, memory = load_chain(TOOLS_DEFAULT_LIST, llm) 274 | 275 | # Pertains to question answering functionality 276 | embeddings = OpenAIEmbeddings() 277 | qa_chain = load_qa_chain(OpenAI(temperature=0), chain_type="stuff") 278 | 279 | print(str(datetime.datetime.now()) + ": After load_chain, OPENAI_API_KEY length: " + str( 280 | len(os.environ["OPENAI_API_KEY"]))) 281 | os.environ["OPENAI_API_KEY"] = "" 282 | return chain, express_chain, llm, embeddings, qa_chain, memory 283 | return None, None, None, None, None, None 284 | 285 | 286 | def run_chain(chain, inp, capture_hidden_text): 287 | output = "" 288 | hidden_text = None 289 | if capture_hidden_text: 290 | error_msg = None 291 | tmp = sys.stdout 292 | hidden_text_io = StringIO() 293 | sys.stdout = hidden_text_io 294 | 295 | try: 296 | output = chain.run(input=inp) 297 | except AuthenticationError as ae: 298 | error_msg = AUTH_ERR_MSG + str(datetime.datetime.now()) + ". " + str(ae) 299 | print("error_msg", error_msg) 300 | except RateLimitError as rle: 301 | error_msg = "\n\nRateLimitError: " + str(rle) 302 | except ValueError as ve: 303 | error_msg = "\n\nValueError: " + str(ve) 304 | except InvalidRequestError as ire: 305 | error_msg = "\n\nInvalidRequestError: " + str(ire) 306 | except Exception as e: 307 | error_msg = "\n\n" + BUG_FOUND_MSG + ":\n\n" + str(e) 308 | 309 | sys.stdout = tmp 310 | hidden_text = hidden_text_io.getvalue() 311 | 312 | # remove escape characters from hidden_text 313 | hidden_text = re.sub(r'\x1b[^m]*m', '', hidden_text) 314 | 315 | # remove "Entering new AgentExecutor chain..." from hidden_text 316 | hidden_text = re.sub(r"Entering new AgentExecutor chain...\n", "", hidden_text) 317 | 318 | # remove "Finished chain." from hidden_text 319 | hidden_text = re.sub(r"Finished chain.", "", hidden_text) 320 | 321 | # Add newline after "Thought:" "Action:" "Observation:" "Input:" and "AI:" 322 | hidden_text = re.sub(r"Thought:", "\n\nThought:", hidden_text) 323 | hidden_text = re.sub(r"Action:", "\n\nAction:", hidden_text) 324 | hidden_text = re.sub(r"Observation:", "\n\nObservation:", hidden_text) 325 | hidden_text = re.sub(r"Input:", "\n\nInput:", hidden_text) 326 | hidden_text = re.sub(r"AI:", "\n\nAI:", hidden_text) 327 | 328 | if error_msg: 329 | hidden_text += error_msg 330 | 331 | print("hidden_text: ", hidden_text) 332 | else: 333 | try: 334 | output = chain.run(input=inp) 335 | except AuthenticationError as ae: 336 | output = AUTH_ERR_MSG + str(datetime.datetime.now()) + ". " + str(ae) 337 | print("output", output) 338 | except RateLimitError as rle: 339 | output = "\n\nRateLimitError: " + str(rle) 340 | except ValueError as ve: 341 | output = "\n\nValueError: " + str(ve) 342 | except InvalidRequestError as ire: 343 | output = "\n\nInvalidRequestError: " + str(ire) 344 | except Exception as e: 345 | output = "\n\n" + BUG_FOUND_MSG + ":\n\n" + str(e) 346 | 347 | return output, hidden_text 348 | 349 | 350 | def reset_memory(history, memory): 351 | memory.clear() 352 | history = [] 353 | return history, history, memory 354 | 355 | 356 | class ChatWrapper: 357 | 358 | def __init__(self): 359 | self.lock = Lock() 360 | 361 | def __call__( 362 | self, api_key: str, inp: str, history: Optional[Tuple[str, str]], chain: Optional[ConversationChain], 363 | trace_chain: bool, speak_text: bool, talking_head: bool, monologue: bool, express_chain: Optional[LLMChain], 364 | num_words, formality, anticipation_level, joy_level, trust_level, 365 | fear_level, surprise_level, sadness_level, disgust_level, anger_level, 366 | lang_level, translate_to, literary_style, qa_chain, docsearch, use_embeddings 367 | ): 368 | """Execute the chat functionality.""" 369 | self.lock.acquire() 370 | try: 371 | print("\n==== date/time: " + str(datetime.datetime.now()) + " ====") 372 | print("inp: " + inp) 373 | print("trace_chain: ", trace_chain) 374 | print("speak_text: ", speak_text) 375 | print("talking_head: ", talking_head) 376 | print("monologue: ", monologue) 377 | history = history or [] 378 | # If chain is None, that is because no API key was provided. 379 | output = "Please paste your OpenAI key from openai.com to use this app. " + str(datetime.datetime.now()) 380 | hidden_text = output 381 | 382 | if chain: 383 | # Set OpenAI key 384 | import openai 385 | openai.api_key = api_key 386 | if not monologue: 387 | if use_embeddings: 388 | if inp and inp.strip() != "": 389 | if docsearch: 390 | docs = docsearch.similarity_search(inp) 391 | output = str(qa_chain.run(input_documents=docs, question=inp)) 392 | else: 393 | output, hidden_text = "Please supply some text in the the Embeddings tab.", None 394 | else: 395 | output, hidden_text = "What's on your mind?", None 396 | else: 397 | output, hidden_text = run_chain(chain, inp, capture_hidden_text=trace_chain) 398 | else: 399 | output, hidden_text = inp, None 400 | 401 | output = transform_text(output, express_chain, num_words, formality, anticipation_level, joy_level, 402 | trust_level, 403 | fear_level, surprise_level, sadness_level, disgust_level, anger_level, 404 | lang_level, translate_to, literary_style) 405 | 406 | text_to_display = output 407 | if trace_chain: 408 | text_to_display = hidden_text + "\n\n" + output 409 | history.append((inp, text_to_display)) 410 | 411 | html_video, temp_file, html_audio, temp_aud_file = None, None, None, None 412 | if speak_text: 413 | if talking_head: 414 | if len(output) <= MAX_TALKING_HEAD_TEXT_LENGTH: 415 | html_video, temp_file = do_html_video_speak(output, translate_to) 416 | else: 417 | temp_file = LOOPING_TALKING_HEAD 418 | html_video = create_html_video(temp_file, TALKING_HEAD_WIDTH) 419 | html_audio, temp_aud_file = do_html_audio_speak(output, translate_to) 420 | else: 421 | html_audio, temp_aud_file = do_html_audio_speak(output, translate_to) 422 | else: 423 | if talking_head: 424 | temp_file = LOOPING_TALKING_HEAD 425 | html_video = create_html_video(temp_file, TALKING_HEAD_WIDTH) 426 | else: 427 | # html_audio, temp_aud_file = do_html_audio_speak(output, translate_to) 428 | # html_video = create_html_video(temp_file, "128") 429 | pass 430 | 431 | except Exception as e: 432 | raise e 433 | finally: 434 | self.lock.release() 435 | return history, history, html_video, temp_file, html_audio, temp_aud_file, "" 436 | # return history, history, html_audio, temp_aud_file, "" 437 | 438 | 439 | chat = ChatWrapper() 440 | 441 | 442 | def do_html_audio_speak(words_to_speak, polly_language): 443 | polly_client = boto3.Session( 444 | aws_access_key_id=os.environ["AWS_ACCESS_KEY_ID"], 445 | aws_secret_access_key=os.environ["AWS_SECRET_ACCESS_KEY"], 446 | region_name=os.environ["AWS_DEFAULT_REGION"] 447 | ).client('polly') 448 | 449 | # voice_id, language_code, engine = POLLY_VOICE_DATA.get_voice(polly_language, "Female") 450 | voice_id, language_code, engine = POLLY_VOICE_DATA.get_voice(polly_language, "Male") 451 | if not voice_id: 452 | # voice_id = "Joanna" 453 | voice_id = "Matthew" 454 | language_code = "en-US" 455 | engine = NEURAL_ENGINE 456 | response = polly_client.synthesize_speech( 457 | Text=words_to_speak, 458 | OutputFormat='mp3', 459 | VoiceId=voice_id, 460 | LanguageCode=language_code, 461 | Engine=engine 462 | ) 463 | 464 | html_audio = '
no audio
' 465 | 466 | # Save the audio stream returned by Amazon Polly on Lambda's temp directory 467 | if "AudioStream" in response: 468 | with closing(response["AudioStream"]) as stream: 469 | # output = os.path.join("/tmp/", "speech.mp3") 470 | 471 | try: 472 | with open('audios/tempfile.mp3', 'wb') as f: 473 | f.write(stream.read()) 474 | temp_aud_file = gr.File("audios/tempfile.mp3") 475 | temp_aud_file_url = "/file=" + temp_aud_file.value['name'] 476 | html_audio = f'' 477 | except IOError as error: 478 | # Could not write to file, exit gracefully 479 | print(error) 480 | return None, None 481 | else: 482 | # The response didn't contain audio data, exit gracefully 483 | print("Could not stream audio") 484 | return None, None 485 | 486 | return html_audio, "audios/tempfile.mp3" 487 | 488 | 489 | def create_html_video(file_name, width): 490 | temp_file_url = "/file=" + tmp_file.value['name'] 491 | html_video = f'' 492 | return html_video 493 | 494 | 495 | def do_html_video_speak(words_to_speak, azure_language): 496 | azure_voice = AZURE_VOICE_DATA.get_voice(azure_language, "Male") 497 | if not azure_voice: 498 | azure_voice = "en-US-ChristopherNeural" 499 | 500 | headers = {"Authorization": f"Bearer {os.environ['EXHUMAN_API_KEY']}"} 501 | body = { 502 | 'bot_name': 'Masahiro', 503 | 'bot_response': words_to_speak, 504 | 'azure_voice': azure_voice, 505 | 'azure_style': 'friendly', 506 | 'animation_pipeline': 'high_speed', 507 | } 508 | api_endpoint = "https://api.exh.ai/animations/v1/generate_lipsync" 509 | res = requests.post(api_endpoint, json=body, headers=headers) 510 | print("res.status_code: ", res.status_code) 511 | 512 | html_video = '
no video
' 513 | if isinstance(res.content, bytes): 514 | response_stream = io.BytesIO(res.content) 515 | print("len(res.content)): ", len(res.content)) 516 | 517 | with open('videos/tempfile.mp4', 'wb') as f: 518 | f.write(response_stream.read()) 519 | temp_file = gr.File("videos/tempfile.mp4") 520 | temp_file_url = "/file=" + temp_file.value['name'] 521 | html_video = f'' 522 | else: 523 | print('video url unknown') 524 | return html_video, "videos/tempfile.mp4" 525 | 526 | 527 | def update_selected_tools(widget, state, llm): 528 | if widget: 529 | state = widget 530 | chain, express_chain, memory = load_chain(state, llm) 531 | return state, llm, chain, express_chain 532 | 533 | 534 | def update_talking_head(widget, state): 535 | if widget: 536 | state = widget 537 | 538 | video_html_talking_head = create_html_video(LOOPING_TALKING_HEAD, TALKING_HEAD_WIDTH) 539 | return state, video_html_talking_head 540 | else: 541 | # return state, create_html_video(LOOPING_TALKING_HEAD, "32") 542 | return None, "
"
543 | 
544 | 
545 | def update_foo(widget, state):
546 |     if widget:
547 |         state = widget
548 |         return state
549 | 
550 | 
551 | # Pertains to question answering functionality
552 | def update_embeddings(embeddings_text, embeddings, qa_chain):
553 |     if embeddings_text:
554 |         text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
555 |         texts = text_splitter.split_text(embeddings_text)
556 | 
557 |         docsearch = FAISS.from_texts(texts, embeddings)
558 |         print("Embeddings updated")
559 |         return docsearch
560 | 
561 | 
562 | # Pertains to question answering functionality
563 | def update_use_embeddings(widget, state):
564 |     if widget:
565 |         state = widget
566 |         return state
567 | 
568 | 
569 | with gr.Blocks(css=".gradio-container {background-color: lightgray}") as block:
570 |     llm_state = gr.State()
571 |     history_state = gr.State()
572 |     chain_state = gr.State()
573 |     express_chain_state = gr.State()
574 |     tools_list_state = gr.State(TOOLS_DEFAULT_LIST)
575 |     trace_chain_state = gr.State(False)
576 |     speak_text_state = gr.State(False)
577 |     talking_head_state = gr.State(True)
578 |     monologue_state = gr.State(False)  # Takes the input and repeats it back to the user, optionally transforming it.
579 |     memory_state = gr.State()
580 | 
581 |     # Pertains to Express-inator functionality
582 |     num_words_state = gr.State(NUM_WORDS_DEFAULT)
583 |     formality_state = gr.State(FORMALITY_DEFAULT)
584 |     anticipation_level_state = gr.State(EMOTION_DEFAULT)
585 |     joy_level_state = gr.State(EMOTION_DEFAULT)
586 |     trust_level_state = gr.State(EMOTION_DEFAULT)
587 |     fear_level_state = gr.State(EMOTION_DEFAULT)
588 |     surprise_level_state = gr.State(EMOTION_DEFAULT)
589 |     sadness_level_state = gr.State(EMOTION_DEFAULT)
590 |     disgust_level_state = gr.State(EMOTION_DEFAULT)
591 |     anger_level_state = gr.State(EMOTION_DEFAULT)
592 |     lang_level_state = gr.State(LANG_LEVEL_DEFAULT)
593 |     translate_to_state = gr.State(TRANSLATE_TO_DEFAULT)
594 |     literary_style_state = gr.State(LITERARY_STYLE_DEFAULT)
595 | 
596 |     # Pertains to WHISPER functionality
597 |     whisper_lang_state = gr.State(WHISPER_DETECT_LANG)
598 | 
599 |     # Pertains to question answering functionality
600 |     embeddings_state = gr.State()
601 |     qa_chain_state = gr.State()
602 |     docsearch_state = gr.State()
603 |     use_embeddings_state = gr.State(False)
604 | 
605 |     with gr.Tab("Chat"):
606 |         with gr.Row():
607 |             with gr.Column():
608 |                 gr.HTML(
609 |                     """
GPT + WolframAlpha + Whisper

610 |                     
Hit Enter after pasting your OpenAI API key
""")
611 | 
612 |             openai_api_key_textbox = gr.Textbox(placeholder="Paste your OpenAI API key (sk-...) and hit Enter",
613 |                                                 show_label=False, lines=1, type='password')
614 | 
615 |         with gr.Row():
616 |             with gr.Column(scale=1, min_width=TALKING_HEAD_WIDTH, visible=True):
617 |                 speak_text_cb = gr.Checkbox(label="Enable speech", value=False)
618 |                 speak_text_cb.change(update_foo, inputs=[speak_text_cb, speak_text_state],
619 |                                      outputs=[speak_text_state])
620 | 
621 |                 my_file = gr.File(label="Upload a file", type="file", visible=False)
622 |                 tmp_file = gr.File(LOOPING_TALKING_HEAD, visible=False)
623 |                 # tmp_file_url = "/file=" + tmp_file.value['name']
624 |                 htm_video = create_html_video(LOOPING_TALKING_HEAD, TALKING_HEAD_WIDTH)
625 |                 video_html = gr.HTML(htm_video)
626 | 
627 |                 # my_aud_file = gr.File(label="Audio file", type="file", visible=True)
628 |                 tmp_aud_file = gr.File("audios/tempfile.mp3", visible=False)
629 |                 tmp_aud_file_url = "/file=" + tmp_aud_file.value['name']
630 |                 htm_audio = f''
631 |                 audio_html = gr.HTML(htm_audio)
632 | 
633 |             with gr.Column(scale=7):
634 |                 chatbot = gr.Chatbot()
635 | 
636 |         with gr.Row():
637 |             message = gr.Textbox(label="What's on your mind??",
638 |                                  placeholder="What's the answer to life, the universe, and everything?",
639 |                                  lines=1)
640 |             submit = gr.Button(value="Send", variant="secondary").style(full_width=False)
641 | 
642 |         # UNCOMMENT TO USE WHISPER
643 |         with gr.Row():
644 |             audio_comp = gr.Microphone(source="microphone", type="filepath", label="Just say it!",
645 |                                        interactive=True, streaming=False)
646 |             audio_comp.change(transcribe, inputs=[audio_comp, whisper_lang_state], outputs=[message])
647 | 
648 |         # TEMPORARY FOR TESTING
649 |         # with gr.Row():
650 |         #     audio_comp_tb = gr.Textbox(label="Just say it!", lines=1)
651 |         #     audio_comp_tb.submit(transcribe_dummy, inputs=[audio_comp_tb, whisper_lang_state], outputs=[message])
652 | 
653 |         gr.Examples(
654 |             examples=["How many people live in Canada?",
655 |                       "What is 2 to the 30th power?",
656 |                       "If x+y=10 and x-y=4, what are x and y?",
657 |                       "How much did it rain in SF today?",
658 |                       "Get me information about the movie 'Avatar'",
659 |                       "What are the top tech headlines in the US?",
660 |                       "On the desk, you see two blue booklets, two purple booklets, and two yellow pairs of sunglasses - "
661 |                       "if I remove all the pairs of sunglasses from the desk, how many purple items remain on it?"],
662 |             inputs=message
663 |         )
664 | 
665 |     with gr.Tab("Settings"):
666 |         tools_cb_group = gr.CheckboxGroup(label="Tools:", choices=TOOLS_LIST,
667 |                                           value=TOOLS_DEFAULT_LIST)
668 |         tools_cb_group.change(update_selected_tools,
669 |                               inputs=[tools_cb_group, tools_list_state, llm_state],
670 |                               outputs=[tools_list_state, llm_state, chain_state, express_chain_state])
671 | 
672 |         trace_chain_cb = gr.Checkbox(label="Show reasoning chain in chat bubble", value=False)
673 |         trace_chain_cb.change(update_foo, inputs=[trace_chain_cb, trace_chain_state],
674 |                               outputs=[trace_chain_state])
675 | 
676 |         # speak_text_cb = gr.Checkbox(label="Speak text from agent", value=False)
677 |         # speak_text_cb.change(update_foo, inputs=[speak_text_cb, speak_text_state],
678 |         #                      outputs=[speak_text_state])
679 | 
680 |         talking_head_cb = gr.Checkbox(label="Show talking head", value=True)
681 |         talking_head_cb.change(update_talking_head, inputs=[talking_head_cb, talking_head_state],
682 |                                outputs=[talking_head_state, video_html])
683 | 
684 |         monologue_cb = gr.Checkbox(label="Babel fish mode (translate/restate what you enter, no conversational agent)",
685 |                                    value=False)
686 |         monologue_cb.change(update_foo, inputs=[monologue_cb, monologue_state],
687 |                             outputs=[monologue_state])
688 | 
689 |         reset_btn = gr.Button(value="Reset chat", variant="secondary").style(full_width=False)
690 |         reset_btn.click(reset_memory, inputs=[history_state, memory_state], outputs=[chatbot, history_state, memory_state])
691 | 
692 |     with gr.Tab("Whisper STT"):
693 |         whisper_lang_radio = gr.Radio(label="Whisper speech-to-text language:", choices=[
694 |             WHISPER_DETECT_LANG, "Arabic", "Arabic (Gulf)", "Catalan", "Chinese (Cantonese)", "Chinese (Mandarin)",
695 |             "Danish", "Dutch", "English (Australian)", "English (British)", "English (Indian)", "English (New Zealand)",
696 |             "English (South African)", "English (US)", "English (Welsh)", "Finnish", "French", "French (Canadian)",
697 |             "German", "German (Austrian)", "Georgian", "Hindi", "Icelandic", "Indonesian", "Italian", "Japanese",
698 |             "Korean", "Norwegian", "Polish",
699 |             "Portuguese (Brazilian)", "Portuguese (European)", "Romanian", "Russian", "Spanish (European)",
700 |             "Spanish (Mexican)", "Spanish (US)", "Swedish", "Turkish", "Ukrainian", "Welsh"],
701 |                                       value=WHISPER_DETECT_LANG)
702 | 
703 |         whisper_lang_radio.change(update_foo,
704 |                                   inputs=[whisper_lang_radio, whisper_lang_state],
705 |                                   outputs=[whisper_lang_state])
706 | 
707 |     with gr.Tab("Translate to"):
708 |         lang_level_radio = gr.Radio(label="Language level:", choices=[
709 |             LANG_LEVEL_DEFAULT, "1st grade", "2nd grade", "3rd grade", "4th grade", "5th grade", "6th grade",
710 |             "7th grade", "8th grade", "9th grade", "10th grade", "11th grade", "12th grade", "University"],
711 |                                     value=LANG_LEVEL_DEFAULT)
712 |         lang_level_radio.change(update_foo, inputs=[lang_level_radio, lang_level_state],
713 |                                 outputs=[lang_level_state])
714 | 
715 |         translate_to_radio = gr.Radio(label="Language:", choices=[
716 |             TRANSLATE_TO_DEFAULT, "Arabic", "Arabic (Gulf)", "Catalan", "Chinese (Cantonese)", "Chinese (Mandarin)",
717 |             "Danish", "Dutch", "English (Australian)", "English (British)", "English (Indian)", "English (New Zealand)",
718 |             "English (South African)", "English (US)", "English (Welsh)", "Finnish", "French", "French (Canadian)",
719 |             "German", "German (Austrian)", "Georgian", "Hindi", "Icelandic", "Indonesian", "Italian", "Japanese",
720 |             "Korean", "Norwegian", "Polish",
721 |             "Portuguese (Brazilian)", "Portuguese (European)", "Romanian", "Russian", "Spanish (European)",
722 |             "Spanish (Mexican)", "Spanish (US)", "Swedish", "Turkish", "Ukrainian", "Welsh",
723 |             "emojis", "Gen Z slang", "how the stereotypical Karen would say it", "Klingon", "Neanderthal",
724 |             "Pirate", "Strange Planet expospeak technical talk", "Yoda"],
725 |                                       value=TRANSLATE_TO_DEFAULT)
726 | 
727 |         translate_to_radio.change(update_foo,
728 |                                   inputs=[translate_to_radio, translate_to_state],
729 |                                   outputs=[translate_to_state])
730 | 
731 |     with gr.Tab("Formality"):
732 |         formality_radio = gr.Radio(label="Formality:",
733 |                                    choices=[FORMALITY_DEFAULT, "Casual", "Polite", "Honorific"],
734 |                                    value=FORMALITY_DEFAULT)
735 |         formality_radio.change(update_foo,
736 |                                inputs=[formality_radio, formality_state],
737 |                                outputs=[formality_state])
738 | 
739 |     with gr.Tab("Lit style"):
740 |         literary_style_radio = gr.Radio(label="Literary style:", choices=[
741 |             LITERARY_STYLE_DEFAULT, "Prose", "Story", "Summary", "Outline", "Bullets", "Poetry", "Haiku", "Limerick", "Rap",
742 |             "Joke", "Knock-knock", "FAQ"],
743 |                                         value=LITERARY_STYLE_DEFAULT)
744 | 
745 |         literary_style_radio.change(update_foo,
746 |                                     inputs=[literary_style_radio, literary_style_state],
747 |                                     outputs=[literary_style_state])
748 | 
749 |     with gr.Tab("Emotions"):
750 |         anticipation_level_radio = gr.Radio(label="Anticipation level:",
751 |                                             choices=[EMOTION_DEFAULT, "Interest", "Anticipation", "Vigilance"],
752 |                                             value=EMOTION_DEFAULT)
753 |         anticipation_level_radio.change(update_foo,
754 |                                         inputs=[anticipation_level_radio, anticipation_level_state],
755 |                                         outputs=[anticipation_level_state])
756 | 
757 |         joy_level_radio = gr.Radio(label="Joy level:",
758 |                                    choices=[EMOTION_DEFAULT, "Serenity", "Joy", "Ecstasy"],
759 |                                    value=EMOTION_DEFAULT)
760 |         joy_level_radio.change(update_foo,
761 |                                inputs=[joy_level_radio, joy_level_state],
762 |                                outputs=[joy_level_state])
763 | 
764 |         trust_level_radio = gr.Radio(label="Trust level:",
765 |                                      choices=[EMOTION_DEFAULT, "Acceptance", "Trust", "Admiration"],
766 |                                      value=EMOTION_DEFAULT)
767 |         trust_level_radio.change(update_foo,
768 |                                  inputs=[trust_level_radio, trust_level_state],
769 |                                  outputs=[trust_level_state])
770 | 
771 |         fear_level_radio = gr.Radio(label="Fear level:",
772 |                                     choices=[EMOTION_DEFAULT, "Apprehension", "Fear", "Terror"],
773 |                                     value=EMOTION_DEFAULT)
774 |         fear_level_radio.change(update_foo,
775 |                                 inputs=[fear_level_radio, fear_level_state],
776 |                                 outputs=[fear_level_state])
777 | 
778 |         surprise_level_radio = gr.Radio(label="Surprise level:",
779 |                                         choices=[EMOTION_DEFAULT, "Distraction", "Surprise", "Amazement"],
780 |                                         value=EMOTION_DEFAULT)
781 |         surprise_level_radio.change(update_foo,
782 |                                     inputs=[surprise_level_radio, surprise_level_state],
783 |                                     outputs=[surprise_level_state])
784 | 
785 |         sadness_level_radio = gr.Radio(label="Sadness level:",
786 |                                        choices=[EMOTION_DEFAULT, "Pensiveness", "Sadness", "Grief"],
787 |                                        value=EMOTION_DEFAULT)
788 |         sadness_level_radio.change(update_foo,
789 |                                    inputs=[sadness_level_radio, sadness_level_state],
790 |                                    outputs=[sadness_level_state])
791 | 
792 |         disgust_level_radio = gr.Radio(label="Disgust level:",
793 |                                        choices=[EMOTION_DEFAULT, "Boredom", "Disgust", "Loathing"],
794 |                                        value=EMOTION_DEFAULT)
795 |         disgust_level_radio.change(update_foo,
796 |                                    inputs=[disgust_level_radio, disgust_level_state],
797 |                                    outputs=[disgust_level_state])
798 | 
799 |         anger_level_radio = gr.Radio(label="Anger level:",
800 |                                      choices=[EMOTION_DEFAULT, "Annoyance", "Anger", "Rage"],
801 |                                      value=EMOTION_DEFAULT)
802 |         anger_level_radio.change(update_foo,
803 |                                  inputs=[anger_level_radio, anger_level_state],
804 |                                  outputs=[anger_level_state])
805 | 
806 |     with gr.Tab("Max words"):
807 |         num_words_slider = gr.Slider(label="Max number of words to generate (0 for don't care)",
808 |                                      value=NUM_WORDS_DEFAULT, minimum=0, maximum=MAX_WORDS, step=10)
809 |         num_words_slider.change(update_foo,
810 |                                 inputs=[num_words_slider, num_words_state],
811 |                                 outputs=[num_words_state])
812 | 
813 |     with gr.Tab("Embeddings"):
814 |         embeddings_text_box = gr.Textbox(label="Enter text for embeddings and hit Create:",
815 |                                          lines=20)
816 | 
817 |         with gr.Row():
818 |             use_embeddings_cb = gr.Checkbox(label="Use embeddings", value=False)
819 |             use_embeddings_cb.change(update_use_embeddings, inputs=[use_embeddings_cb, use_embeddings_state],
820 |                                      outputs=[use_embeddings_state])
821 | 
822 |             embeddings_text_submit = gr.Button(value="Create", variant="secondary").style(full_width=False)
823 |             embeddings_text_submit.click(update_embeddings,
824 |                                          inputs=[embeddings_text_box, embeddings_state, qa_chain_state],
825 |                                          outputs=[docsearch_state])
826 | 
827 |     gr.HTML("""
828 |         
This application, developed by James L. Weaver, 
829 |         demonstrates a conversational agent implemented with OpenAI GPT-3.5 and LangChain. 
830 |         When necessary, it leverages tools for complex math, searching the internet, and accessing news and weather.
831 |         Uses talking heads from Ex-Human.
832 |         For faster inference without waiting in queue, you may duplicate the space.
833 |         
""")
834 | 
835 |     gr.HTML("""
836 | 

837 | 
838 | 
839 | 
840 | 
841 | 
842 | 
843 | 

844 |     """)
845 | 
846 |     gr.HTML("""

847 |         
848 |         Duplicate Space
849 |         Powered by LangChain 🦜️🔗
850 |         
""")
851 | 
852 |     message.submit(chat, inputs=[openai_api_key_textbox, message, history_state, chain_state, trace_chain_state,
853 |                                  speak_text_state, talking_head_state, monologue_state,
854 |                                  express_chain_state, num_words_state, formality_state,
855 |                                  anticipation_level_state, joy_level_state, trust_level_state, fear_level_state,
856 |                                  surprise_level_state, sadness_level_state, disgust_level_state, anger_level_state,
857 |                                  lang_level_state, translate_to_state, literary_style_state,
858 |                                  qa_chain_state, docsearch_state, use_embeddings_state],
859 |                    outputs=[chatbot, history_state, video_html, my_file, audio_html, tmp_aud_file, message])
860 |     # outputs=[chatbot, history_state, audio_html, tmp_aud_file, message])
861 | 
862 |     submit.click(chat, inputs=[openai_api_key_textbox, message, history_state, chain_state, trace_chain_state,
863 |                                speak_text_state, talking_head_state, monologue_state,
864 |                                express_chain_state, num_words_state, formality_state,
865 |                                anticipation_level_state, joy_level_state, trust_level_state, fear_level_state,
866 |                                surprise_level_state, sadness_level_state, disgust_level_state, anger_level_state,
867 |                                lang_level_state, translate_to_state, literary_style_state,
868 |                                qa_chain_state, docsearch_state, use_embeddings_state],
869 |                  outputs=[chatbot, history_state, video_html, my_file, audio_html, tmp_aud_file, message])
870 |     # outputs=[chatbot, history_state, audio_html, tmp_aud_file, message])
871 | 
872 |     openai_api_key_textbox.change(set_openai_api_key,
873 |                                   inputs=[openai_api_key_textbox],
874 |                                   outputs=[chain_state, express_chain_state, llm_state, embeddings_state,
875 |                                            qa_chain_state, memory_state])
876 |     openai_api_key_textbox.submit(set_openai_api_key,
877 |                                   inputs=[openai_api_key_textbox],
878 |                                   outputs=[chain_state, express_chain_state, llm_state, embeddings_state,
879 |                                            qa_chain_state, memory_state])
880 | 
881 | block.launch(debug=True)
882 | 


--------------------------------------------------------------------------------