whisper apple.mp3<\/pre>\n![\"\"](\"https:\/\/blog.hoyo.idv.tw\/wp-content\/uploads\/2023\/07\/2023-07-20-13-19-24.png\")
<\/p>\n
\u6307\u4ee4\u5b8c\u6210\u9810\u8a2d\u6703\u8f38\u51fa\u6240\u6709\u683c\u5f0f\u6a94\u6848\uff1ajson, srt, tsv, txt, vtt<\/p>\n
![\"\"](\"https:\/\/blog.hoyo.idv.tw\/wp-content\/uploads\/2023\/07\/2023-07-20-13-23-12.png\")
<\/p>\n
\u6307\u5b9a\u6a21\u578b\uff0c\u521d\u6b21\u4f7f\u7528\u6703\u81ea\u52d5\u4e0b\u8f09<\/p>\n
whisper OSR_us_000_0061_8k.wav --model medium\r\nC:\\Users\\chen\\AppData\\Local\\Programs\\Python\\Python310\\lib\\site-packages\\whisper\\timing.py:58: NumbaDeprecationWarning: The 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https:\/\/numba.readthedocs.io\/en\/stable\/reference\/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.\r\n def backtrace(trace: np.ndarray):\r\n 11%|\u2588\u2588\u2588\u2588\u258e | 167M\/1.42G [01:12<08:20, 2.70MiB\/s]<\/pre>\n\u6307\u5b9a\u82f1\u6587\u6700\u5c0f\u6a21\u578b\u5feb\u901f\u8fa8\u8b58<\/p>\n
whisper apple.mp3 --model tiny.en --language English --fp16 False --output_format txt<\/pre>\n--<\/p>\n
Apache \u7db2\u9801\u57f7\u884c<\/h2>\n
\u8981\u5f9e Apache \u7db2\u9801\u547c\u53eb\u57f7\u884c\uff0c\u9700\u8981\u6ce8\u610f\u7684\u90fd\u662f\u8def\u5f91\u554f\u984c\uff0c\u6c92\u6709\u4f7f\u7528\u8005\u6b0a\u9650\u554f\u984c\u3002\u6a21\u578b\u8981\u8907\u88fd\u5230\u53e6\u5916\u7684\u76ee\u9304<\/p>\n
\/usr\/local\/bin\/whisper \/tmp\/audio.webm --model tiny.en --language English --fp16 False --output_format txt --output_dir \/tmp --model_dir \/data\/model<\/pre>\n--<\/p>\n
\u53c3\u6578<\/h2>\n\n- --model {tiny.en,tiny,base.en,base,small.en,small,medium.en,medium,large-v1,large-v2,large}
\nname of the Whisper model to use (default: small)
\n.en \u662f\u82f1\u6587\u5c08\u5c6c\u6a21\u578b<\/li>\n - --model_dir MODEL_DIR
\nthe path to save model files; uses ~\/.cache\/whisper by default (default: None)
\n\u6a21\u578b\u8def\u5f91<\/li>\n - --language {af,am,ar,as,az,ba,be,bg,bn,bo,br,bs,ca,cs,cy,da,de,el,en,es,et,eu,fa,fi,fo,fr,gl,gu,ha,haw,he,hi,hr,ht,hu,hy,id,is,it,ja,jw,ka,kk,km,kn,ko,la,lb,ln,lo,lt,lv,mg,mi,mk,ml,mn,mr,ms,mt,my,ne,nl,nn,no,oc,pa,pl,ps,pt,ro,ru,sa,sd,si,sk,sl,sn,so,sq,sr,su,sv,sw,ta,te,tg,th,tk,tl,tr,tt,uk,ur,uz,vi,yi,yo,zh,Afrikaans,Albanian,Amharic,Arabic,Armenian,Assamese,Azerbaijani,Bashkir,Basque,Belarusian,Bengali,Bosnian,Breton,Bulgarian,Burmese,Castilian,Catalan,Chinese,Croatian,Czech,Danish,Dutch,English,Estonian,Faroese,Finnish,Flemish,French,Galician,Georgian,German,Greek,Gujarati,Haitian,Haitian Creole,Hausa,Hawaiian,Hebrew,Hindi,Hungarian,Icelandic,Indonesian,Italian,Japanese,Javanese,Kannada,Kazakh,Khmer,Korean,Lao,Latin,Latvian,Letzeburgesch,Lingala,Lithuanian,Luxembourgish,Macedonian,Malagasy,Malay,Malayalam,Maltese,Maori,Marathi,Moldavian,Moldovan,Mongolian,Myanmar,Nepali,Norwegian,Nynorsk,Occitan,Panjabi,Pashto,Persian,Polish,Portuguese,Punjabi,Pushto,Romanian,Russian,Sanskrit,Serbian,Shona,Sindhi,Sinhala,Sinhalese,Slovak,Slovenian,Somali,Spanish,Sundanese,Swahili,Swedish,Tagalog,Tajik,Tamil,Tatar,Telugu,Thai,Tibetan,Turkish,Turkmen,Ukrainian,Urdu,Uzbek,Valencian,Vietnamese,Welsh,Yiddish,Yoruba}
\n\u70ba\u8a2d\u5b9a\u6642\uff0c\u57f7\u884c\u6642\u6703\u81ea\u52d5\u5075\u6e2c<\/li>\n - --output_format {txt,vtt,srt,tsv,json,all}, -f {txt,vtt,srt,tsv,json,all}
\nformat of the output file; if not specified, all available formats will be produced (default:all)
\n\u6a94\u6848\u8f38\u51fa\u683c\u5f0f\uff0c\u548c\u756b\u9762\u8f38\u51fa\u7121\u95dc<\/li>\n - --threads THREADS number of threads used by torch for CPU inference; supercedes MKL_NUM_THREADS\/OMP_NUM_THREADS
\n(default: 0)
\n\u4f7f\u7528 CPU \u6642\uff0c\u6307\u5b9a\u4f7f\u7528\u6838\u5fc3\u6578<\/li>\n - <\/li>\n<\/ul>\n
--<\/p>\n
\u4e0d\u51fa\u73fe\u8b66\u544a<\/h2>\nPYTHONWARNINGS=ignore whisper \/data1\/tmp\/audio.webm --model small --language Chinese --output_format txt --fp16 False<\/pre>\n--<\/p>\n
\u786c\u9ad4\u898f\u683c\u5efa\u8b70<\/h2>\n
\ud83e\udde0 \u4e0d\u540c\u6a21\u578b\u5927\u5c0f\u8207\u5ef6\u9072\u4f30\u7b97\uff08\u4ee5 30 \u79d2\u97f3\u8a0a\u70ba\u4f8b\uff09<\/p>\n
\n\n\n| \u6a21\u578b<\/th>\n | \u5e73\u5747\u8655\u7406\u6642\u9593\uff08CPU\uff09<\/th>\n | GPU (\u5982 RTX 3060+) \u8655\u7406\u6642\u9593<\/th>\n | \u79d2\u56de\u5efa\u8b70<\/th>\n<\/tr>\n<\/thead>\n |
\n\ntiny<\/code><\/td>\n| 4\u20136 \u79d2<\/td>\n | < 1 \u79d2<\/td>\n | \u2705 \u53ef\u79d2\u56de<\/td>\n<\/tr>\n | \nbase<\/code><\/td>\n| 8\u201312 \u79d2<\/td>\n | ~1 \u79d2<\/td>\n | \u2705 \u53ef\u79d2\u56de<\/td>\n<\/tr>\n | \nsmall<\/code><\/td>\n| 15\u201320 \u79d2<\/td>\n | ~2\u20133 \u79d2<\/td>\n | \u26a0\ufe0f \u8fd1\u5373\u6642<\/td>\n<\/tr>\n | \nmedium<\/code><\/td>\n| 30\u201350 \u79d2<\/td>\n | ~5\u20137 \u79d2<\/td>\n | \u274c \u4e0d\u9069\u5408\u79d2\u56de<\/td>\n<\/tr>\n | \nlarge<\/code><\/td>\n| 1\u20132 \u5206\u9418\u4ee5\u4e0a<\/td>\n | ~10 \u79d2\u4ee5\u4e0a<\/td>\n | \u274c \u4e0d\u9069\u5408\u79d2\u56de<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n \u2705 GPU \u52a0\u901f\uff08\u5f37\u70c8\u5efa\u8b70\uff09<\/p>\n \n\n\n| \u5143\u4ef6<\/th>\n | \u63a8\u85a6\u898f\u683c<\/th>\n<\/tr>\n<\/thead>\n | \n\nGPU<\/strong><\/td>\n| NVIDIA RTX 3060 \/ 4060 \u4ee5\u4e0a\uff08CUDA \u652f\u63f4\uff09<\/td>\n<\/tr>\n | \nCUDA \u9a45\u52d5<\/strong><\/td>\n| CUDA Toolkit 11.x + cuDNN<\/td>\n<\/tr>\n | \nRAM<\/strong><\/td>\n| 16 GB \u4ee5\u4e0a<\/td>\n<\/tr>\n | \nCPU<\/strong><\/td>\n| Intel i5\/Ryzen 5\uff08\u6216\u66f4\u597d\uff09<\/td>\n<\/tr>\n | \nSSD<\/strong><\/td>\n| \u5feb\u901f NVMe SSD<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n \u2705 \u6574\u9ad4\u7d50\u8ad6<\/strong><\/p>\n\n\n\n| \u7528\u9014<\/th>\n | \u5efa\u8b70\u9078\u64c7<\/th>\n<\/tr>\n<\/thead>\n | \n\n| \u6253\u904a\u6232\uff081080p\uff09<\/td>\n | RTX 4060<\/strong>\uff08\u6548\u80fd\u66f4\u597d\u3001\u529f\u8017\u66f4\u4f4e\uff09<\/td>\n<\/tr>\n\n| \u6253\u904a\u6232\uff082K \u6216\u6709\u9ad8\u6750\u8ceaMOD\uff09<\/td>\n | RTX 3060 12GB<\/strong>\uff08\u986f\u5b58\u5920\u7528\uff09<\/td>\n<\/tr>\n\n| AI\u3001Stable Diffusion\u3001\u5275\u4f5c\u7528\u9014<\/td>\n | RTX 3060 12GB<\/strong>\uff08\u986f\u5b58\u91cd\u8981\uff09<\/td>\n<\/tr>\n\n| \u5e0c\u671b\u9577\u6642\u9593\u4f7f\u7528\u3001\u7bc0\u80fd\u7701\u96fb<\/td>\n | RTX 4060<\/strong>\uff08\u65b0\u67b6\u69cb\u66f4\u7701\u96fb\uff09<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n \ud83d\udcca \u6548\u80fd\u6bd4\u8f03<\/strong><\/p>\n\n\n\n| \u9805\u76ee<\/th>\n | RTX 3060 12GB<\/th>\n | RTX 4060 8GB<\/th>\n<\/tr>\n<\/thead>\n | \n\n| CUDA \u6838\u5fc3<\/td>\n | 3584<\/td>\n | 3072<\/td>\n<\/tr>\n | \n| \u986f\u5b58<\/td>\n | 12GB GDDR6<\/td>\n | 8GB GDDR6\uff08\u4f46\u983b\u5bec\u5c0f\uff09<\/td>\n<\/tr>\n | \n| \u8a18\u61b6\u9ad4\u4ecb\u9762<\/td>\n | 192-bit<\/td>\n | 128-bit\uff08\u8f03\u5c0f\uff09<\/strong><\/td>\n<\/tr>\n\n| \u529f\u8017<\/td>\n | \u7d04 170W<\/td>\n | 115W\uff08\u7bc0\u80fd\uff09<\/strong><\/td>\n<\/tr>\n\n| \u67b6\u69cb<\/td>\n | Ampere\uff08\u820a\uff09<\/td>\n | Ada Lovelace\uff08\u65b0\uff09<\/strong><\/td>\n<\/tr>\n\n| \u5149\u8ffd\u6548\u80fd<\/td>\n | \u8f03\u5dee<\/td>\n | \u8f03\u597d\uff08\u7b2c\u4e09\u4ee3 RT \u6838\u5fc3\uff09<\/strong><\/td>\n<\/tr>\n\n| DLSS<\/td>\n | \u6709\uff08DLSS 2\uff09<\/td>\n | \u6709\uff08DLSS 3\uff09<\/strong><\/td>\n<\/tr>\n\n| \u6548\u80fd\uff08\u5927\u591a\u6578\u904a\u6232\uff09<\/td>\n | \u7a0d\u4f4e<\/td>\n | \u7565\u9ad8\uff085~15%\uff09<\/strong><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n --<\/p>\n | | | | | | | | | | | | | | | | | | | | | |