\n
nvidia-smi\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nAnd you should see something like this:<\/p>\n
![\"\u201cHow](\"https:\/\/www.jeremymorgan.com\/images\/tutorials\/generative-ai\/how-to-transcribe-audio\/how-to-transcribe-audio-01.webp\")
<\/p>\n
Install Whisper<\/h2>\n
Whisper runs as an executable within your Python environment. It\u2019s pretty cool.<\/p>\n
The best way to install it is:<\/p>\n
\n
pip install -U openai-whisper\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nBut you can also pull the latest version straight from the repository if you like:<\/p>\n
\n
pip install git+https:\/\/github.com\/openai\/whisper.git\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nEither way, it will install a bunch of packages, so go get some ice water. When it\u2019s done, the whisper executable will be installed.<\/p>\n
I recorded a sample file, and here\u2019s how we can run it.<\/p>\n
\n
whisper [<\/span>audio.flac audio.mp3 audio.wav]<\/span> --model [<\/span>model size]<\/span>\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nI will start with the tiny model just to see how it performs. Here\u2019s a list of available models<\/p>\n
\n\n\n| \u00a0Size<\/th>\n | Parameters<\/th>\n | English-only model<\/th>\n | Multilingual model<\/th>\n | Required VRAM<\/th>\n | Relative speed<\/th>\n<\/tr>\n<\/thead>\n |
\n\n| \u00a0tiny<\/td>\n | \u00a0 \u00a039 M<\/td>\n | \u00a0 \u00a0\u00a0tiny.en<\/code><\/td>\n\u00a0 \u00a0 \u00a0\u00a0tiny<\/code><\/td>\n| \u00a0 \u00a0 ~1 GB<\/td>\n | \u00a0 \u00a0 \u00a0~32x<\/td>\n<\/tr>\n | \n| \u00a0base<\/td>\n | \u00a0 \u00a074 M<\/td>\n | \u00a0 \u00a0\u00a0base.en<\/code><\/td>\n\u00a0 \u00a0 \u00a0\u00a0base<\/code><\/td>\n| \u00a0 \u00a0 ~1 GB<\/td>\n | \u00a0 \u00a0 \u00a0~16x<\/td>\n<\/tr>\n | \n| small<\/td>\n | \u00a0 244 M<\/td>\n | \u00a0 \u00a0\u00a0small.en<\/code><\/td>\n\u00a0 \u00a0 \u00a0small<\/code><\/td>\n| \u00a0 \u00a0 ~2 GB<\/td>\n | \u00a0 \u00a0 \u00a0~6x<\/td>\n<\/tr>\n | \n| medium<\/td>\n | \u00a0 769 M<\/td>\n | \u00a0 \u00a0medium.en<\/code><\/td>\n\u00a0 \u00a0 \u00a0medium<\/code><\/td>\n| \u00a0 \u00a0 ~5 GB<\/td>\n | \u00a0 \u00a0 \u00a0~2x<\/td>\n<\/tr>\n | \n| large<\/td>\n | \u00a0 1550 M<\/td>\n | \u00a0 \u00a0 \u00a0 \u00a0N\/A<\/td>\n | \u00a0 \u00a0 \u00a0large<\/code><\/td>\n| \u00a0 \u00a0~10 GB<\/td>\n | \u00a0 \u00a0 \u00a0 1x<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n I\u2019ll start with the smallest model and see its accuracy, then work my way up if needed.<\/p>\n Here\u2019s the command I ran to parse and extract from my sample file:<\/p>\n \n whisper sample-audio.wav --model tiny\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nAnd lucky for me, it was transcribed perfectly:<\/p>\n ![\"\u201cHow](\"https:\/\/www.jeremymorgan.com\/images\/tutorials\/generative-ai\/how-to-transcribe-audio\/how-to-transcribe-audio-02.webp\") <\/p>\n
Your results will vary. If you don\u2019t like the output you can always step it up to a larger model, which will take more memory and a longer amount of time.<\/p>\n So, what else can you do with this tool?<\/p>\n Building a Cool Python Script<\/h2>\nThe Whisper service has a bunch of cool features that I don\u2019t use, like translation! But what if we want to script this stuff, like processing 100 audio files or something? Building a Python script to run it is easy.<\/p>\n Here\u2019s a script straight from the GitHub page:<\/p>\n \n import<\/span> whisper<\/span>\r\n<\/span><\/span>\r\nmodel<\/span> =<\/span> whisper<\/span>.<\/span>load_model<\/span>(<\/span>\"base\"<\/span>)<\/span>\r\n<\/span><\/span>result<\/span> =<\/span> model<\/span>.<\/span>transcribe<\/span>(<\/span>\"audio.mp3\"<\/span>)<\/span>\r\n<\/span><\/span>print<\/span>(<\/span>result<\/span>[<\/span>\"text\"<\/span>])<\/span>\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nAnd when I run it, it shows clean text output.<\/p>\n ![\"\u201cHow](\"https:\/\/www.jeremymorgan.com\/images\/tutorials\/generative-ai\/how-to-transcribe-audio\/how-to-transcribe-audio-03.webp\") <\/p>\n
You can of course, write this to its own text file:<\/p>\n \n with<\/span> open<\/span>(<\/span>\"output.txt\"<\/span>,<\/span> \"w\"<\/span>)<\/span> as<\/span> file<\/span>:<\/span>\r\n<\/span><\/span> file<\/span>.<\/span>write<\/span>(<\/span>result<\/span>[<\/span>\"text\"<\/span>])<\/span>\r\n<\/span><\/span><\/code><\/pre>\n<\/div>\nThere are tons of options available. It also does transcriptions in other languages as well.<\/p>\n <\/p>\n <\/p>\n <\/p>\n","protected":false},"excerpt":{"rendered":" Le besoin => enregistrement de r\u00e9union => R\u00e9cup\u00e9rer un transcript => faire une synth\u00e8se (MAIA \/ BPCE ou\u00a0 Chat GPT ou Gemini […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-5423","post","type-post","status-publish","format-standard","hentry","category-non-classe"],"_links":{"self":[{"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/5423","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5423"}],"version-history":[{"count":4,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/5423\/revisions"}],"predecessor-version":[{"id":5490,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=\/wp\/v2\/posts\/5423\/revisions\/5490"}],"wp:attachment":[{"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5423"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5423"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.blue-bears.com\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5423"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
| | | | | | | | | |
![\"\u201cHow](\"https:\/\/www.jeremymorgan.com\/images\/tutorials\/generative-ai\/how-to-transcribe-audio\/how-to-transcribe-audio-00.webp\")
<\/p>\n