FREE & OFFLINE Audio to Text | Whisper: Install Guide | OpenAI Whisper | ASR

[@morbid.curiosity]

Incredibly thorough, broke down an "intimidating" task into managable chunks, edited well with fantastic audio. I'm glad I was linked this video! Subscribed!!!

[@thanatos2185]

My father-in-law routinely has a lot of interviews to transcribe. Recently stuck with an increased workload, I went on a quest to find a tool, and stumbled upon whisper, and then your video. I wouldn't have known where to start with python and the cmd window, despite knowing computers well enough. This was of immense help and extremely easy to follow (the only missing thing for me was that I was missing Git. But adding it and also adding it to path was very easy (a bonus thing I learned thanks to this video)). Thank you so much. You just saved an old man a week worth of work and many more in the future.

[@gustavoreche597]

how do i add this git? i am receiving the error

[@1samc]

@@gustavoreche597 same here, i posted steps which worked for me if you havent figured it out yet. good luck!

[@bmqww223]

hello can anyone help me using gpu by default to process instead of cpu , my cpu is very fast but i think gpu would be better suited for this kind of operation......

[@IR240474]

This was wonderful, I followed along using my Macbook Air, and it worked. Thank you so much. Brilliant! Edit - I could transcribe a 45-minute audio recording of a meditation I had to learn. Another idea is to record your own voice for a project and transcribe it. This was so cool, and I thank you for the effort in making this video!

[@BenjaminK123]

Great video, i have been using the wisper API for testing but that costs i didnt know this was open source, your video got me up and running in no time at all, thank you for taking the time to make this its very cool stuff

[@ejohn951]

How was the resources use? is it worth it? I'm also planning on moving my website from the API to selfhosted Whisper.

[@BenjaminK123]

@@ejohn951 its not very fast when self hosted unless you have a power house of a pc, using the whisper api is the way to go is you need to it very quick, if time is not a concern then locally is the way

[@anthologyofheroespodcast]

Very useful, particularly the multiple Python instances. Thanks!

[@nauikunart]

@TheTopMan just press windows key then type cmd

[@MrFerocius]

Awesome quality! Suscribed! Keep them coming.

[@usernotfound.3837]

i have seen so many tutorial videos in my life, this one, is on my top 3 list. just WOW.. perfect everything perfect, thanks for the great work

[@johnbradshaw3001]

Yeah but there are many great GUI for whisper ai. Should use that instead of this command base cmd

[@MondSemmel]

10:38: Note that the "large" Whisper model is now "large-v2", and the old "large" model has been renamed to "large-v1" or something.

[@palmykim1025]

Thanks!

[@TradTalk]

Thanks a lot! I was having a lot of troubles with the PowerShell running of the program, but after this tutorial all my problems were solved.

[@weekendmakeit7760]

you sir are the best. I had to install "GIT" as well

[@MRAHBB]

Thanks. This took me about 2 hours and some extra installs for missing scripts, moving scripts and paths, but I got it.

[@jonasb4395]

Damm thanks for telling how u fixed the missing scripts

[@sergiomendez3647]

Thank you very much!!! Never stops sharing please 🙏

[@oldmanbeta]

Thanks!

[@lazyhomeowner]

2023. things have still not gotten simpler. this was like a speed run on how to set up an ide.

[@martinaheinle8702]

help! when I want to install whisper i get the following error: cannot find command `git`- do you have `git` installed and in your path?

[@ciroramirez6439]

Hey! It has been a while since you posted this. Have you been able to fix it?

[@PCtutorijali]

Can someone please help me, everything works but my entire text gets under one single long timestamp ? how can i fix it

[@klauser10]

You only missed the 'git' installation but everything else was great. I got it to work. Thanks.

[@uninconnu7112]

how did you get it? Im stucked right there

[@1samc]

@@uninconnu7112 i posted steps to get around it if you're still stuck...i was stuck there too for a bit.

[@rashid8646]

@@1samc Posted them where???????

[@srjhr]

good lord, this tutorial is mildly infuriating

[@dec35]

when running the command that clones the github repo, i get an error that sais git is not installed

[@thanatos2185]

You probably fixed this since, but just in case: install Git (not to confound with Github). Read carefully though the installation options. Make sure it's added to "path" just like with the rest.

[@srjhr]

for some reason, when I type 'python -v' after first installation it returns a huge amount of lines saying "import bla-bla-bla" with "Python 3.9.9 (tags/v3.9.9:ccb0e6a, Nov 15 2021, 18:08:50) [MSC v.1929 64 bit (AMD64)] on win32" at the very end. Not sure if it's normal

[@sithumiperera5212]

This happened to me too.Have u found a solution for that?😮

[@rishitaumasankaran7159]

try capital V python -V

[@BirjloxuBR]

Sorry, but this tutorial is confusing. First the missing git thing, that you didn't explain. Second, it now refuses to use my NVIDIA GPU as the renderer thing or whatever. Even though I installed CUDA. So something is missing. For someone who is not into coding, python or anything related this tutorial is just lacking and does not provide all the correct info and troubleshoots, no pun intended.

[@Here_is_Waldo]

I hope you got it figured out. Annoyingly, many of these Hugging Face and Github stuff seems to assume you are a professional IT person, so it seems to rarely explain some things that (I can only assume) they think are simple.

[@elrambler]

Yes. it works but you have to go through every single step I did it before and didn't change the name under my python and then keep throwing me back at the last part. when you install the whispered code from Github, it kept throwing me an error after I went back and did everything exactly Descriptive on the video it finally work. PS also you can just try to press your windows button and your age button. and it's also a new AI service from. the collab between Microsoft and open AI and you can actually move that app. around to use it on different multiple. text prompts.

[@williamvanlaere]

how can I make sure the system uses GPU and not CPU? So it's much faster.

[@Grapegum]

I would've preferred the venv or even conda approach but still, it was a pretty good tutorial. Thank you!

[@BeTheFeatureNotTheBug]

What do you mean by cuda approach?

[@Grapegum]

@@BeTheFeatureNotTheBug I said conda, not cuda, as in using a separate python environment to do the installation.

[@BeTheFeatureNotTheBug]

Thank you! Is there a simple way to speed up whisper I am getting 4 seconds transcribing per minute of audio.

[@rich6476]

one obstacle to clarify is that the "path" folder isn't actually in his "ESD" folder, i messed that up

[@Caio_Freitas]

Hi. I got a message saying GIT is not installed. I tried installing it but i continue getting error message

[@MuhammadAli-fw5bm]

same here

[@matthewdavis1236]

Thank you for the tutorial. I've tried following along, but I continually get the following error message: " raise RuntimeError( RuntimeError: Model has been downloaded but the SHA256 checksum does not not match. Please retry loading the model." Has anyone any suggestions on how to fix this? I've already tried reinstalling Whisper twice.

[@GonzaloGDiezyt]

paste the errore to chatgpt, it usually tells you what to do.

[@officialluxxii327]

i can't seem to get it i keep getting a message that says ERROR: Error [WinError 2] The system cannot find the file specified while executing command git version ERROR: Cannot find command 'git' - do you have 'git' installed and in your PATH? not sure how to fix this error.

[@viacheslavnozdrin9469]

Do you have the resolve now?

[@Game7Pr0]

i got the same

[@Chubaddie]

after going through each step I get "whisper" is not recognized as command

[@husinhashim5723]

it's working great job many thanks 👍

[@mmlufer]

Hola, gracias por tu tutorial! Saludos desde Colombia!

[@JackOW3]

open ai whisper: I am better (says in homelander voice)

[@KryzysX]

I'm smarter, I'm stronger, I'm better... I'm BETTER

[@ppnepali7536]

fp16 is not supported on cpu using fp32 instead whisper pleae help

[@WolstonLobo]

Would you be able to show how to batch process / transcribe multiple audio files?

[@luke9822]

I came here looking for this. I guess I need to find a different video.

[@heitorcosta8592]

Não funcionou pra mim a parte do ffmpeg

[@chocofro3]

Do you know any parameters for controlling time stamps or subtitle length/line breaks? Thanks, and thanks for this video.

[@funderburk5]

I would like to know as well.

[@peterandras61]

@TroubleChute do you have a video about how to automate speaker Diarization or how to combine Whisper with Pyannote?

[@marcovillanuevalegler]

I've followed all the steps. However, whisper is still using my CPU instead of my GPU. How can I solve this?

[@gamer-mt8pe]

after 5:44 i got so confused since it just got confusing. great video i just couldn't understnad past that time period

[@sergeytsygankov]

It refuses to run on GPU (cuda of course), only on CPU. I've tried everything, nothing works. So frustrating. Something in this tutorial is really missing.

[@nezukokawaii2372]

I am having a problem where it says that [WinError 126] The specified module could not be found. Error loading ....... Can someone help me with this problem?

[@INezhynsky13]

C:\Windows\System32>ffmpeg 'ffmpeg' is not recognized as an internal or external command, operable program or batch file.

[@youtubeisamazing9591]

thank you so much works great.

[@RiCARdo_Racing]

Is it a way to host OpenAI? But my country not support it

[@jasonrist6582]

VPN

[@slishnevsky]

Yeah, but how do you do voice typing, not transcription?

[@avamaria8447]

I'm using an old 'puter to test. it's now writing out my file. Thank you.

[@IdontknowV]

When my check my scipts folder, whisper.exe is not there . help please?

[@TroubleChute]

Check which python it installed with, and vavigate to that path. Otherwise search your whole pc for whisper.exe.

[@TroubleChute]

And if it installed properly.

[@DJRayBoston]

same happened to me, it was installed in appdata and not in the program files folder location, find it in appdata and add that script folder to your path

[@pau_gz6899]

can it recognize different people in a recorded meeting?

[@ensarguler7684]

good question, did you find an answer?

[@RyanHunt]

​@@ensarguler7684 I'm wondering as well

[@magicmulder]

No. That would require additional training. I’m currently working on something like that.

[@britneytaylor9929]

Hi, trying to install this however getting stuck at the python install stage. When I hit python - V a massive block of text shows up starting with ; import _frozen_importlib # frozen import _imp # builtin import '_thread' # import '_warnings' # import '_weakref' # etc etc I have installed python 3.9.9 and uninstalled others but not sure what to do from here?

[@nasinpa]

mine either

[@ethanho9096]

are you capitalizing the V?

[@taktaz]

instead of typing -v, do --version

[@vladimirlazar8949]

Thank you for the great instruction. I've installed everything, following your steps. But when I try to use Whisper, I get this error: torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 102.00 MiB (GPU 0; 2.00 GiB total capacity; 1.26 GiB already allocated; 0 bytes free; 1.69 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF How can I fix it (I'm not good in Python at all)?

[@Chris-jg5wr]

Is there any alternative to CUDA for AMD?

[@prophetasi]

9:03 how to add script directory to my path

[@iDaNGeR_GaMeR]

Great thing to use, on KZbin videos Thank you for your effort 👏

[@krn12]

Excellent video managed and I finally ran it from the command line. There is also a translation of the file - srt. Via cmd. ?

[@NoName-zl5vz]

this tutorial support for my old laptop, thanks .... but where is folder the model tiny or large.bin file save ?

[@jonathan-6948]

is there a difference between the Wisper API of Open AI and this one?

[@ygoryarz]

Thanks for the tutorial, I have a problem, it doesn't work offline, why could it be?

[@Warriorbr-un7ew]

Dont know why there is no python folder in my case

[@Jrv175]

Hi, is there any way to use GPU on an AMD graphics card? (instead of CPU)

[@lovielol3935]

hi i tried to download Whisper and cmd gives an error that I need to install git. Should I go git first then?

[@yereloyun]

Hey! Can you make a video showing us how to add Whisper to DaVinci Resolve using the StoryToolkitAI tool?

[@andreydmitriyev4582]

Tnx! And if the video card is built-in, do you need to install nvidia software from the section?

[@nagatouzumaki8750]

You have tried to go thoroughly and I do appreciate it but besides english it shows rubbish in command prompt when transcribing in other languages

[@magicmulder]

German works great. Have you set the language with --language German (or whatever your language is)?

[@Mazufa]

@TroubleChute Hello! I was able to install Whisper on my Windows 11 computer with this installation guide of yours! :) But I have a problem. I have created an mp3 audio file from the movie file on my computer with VLC, and I have tried to make Finnish subtitles with whisper with this command Whisper Hangar.1-The.UFO.Files.S02E01.UFOs.at.War.720p.HDTV.x264-DHD.mp3 -- task translate --language fi. For some reason I always get English subtitles even though everything should be correct, what am I doing wrong or why is this? Thanks for your help!

[@ROBBIEP]

is this easier then cloning?

[@erentekin5134]

Thanks dude keep it goob job

[@michaeltubel]

amazing. can this be used with MP4 or need to convert first to mp3? i am not a programmer but was able to follow the steps.

[@magicmulder]

You can always use ffmpeg to extract audio from video if Whisper doesn’t take it.

[@gebasy]

This is how to make a 0 python experience bro doing the magic in 15 min.🤩

[@chaitanyapatel5876]

A question for you, what exactly do you use PC for? your PC storage is damn high

[@ohsrybro]

Awesome tutorial thanks!!! But I was wondering can I only use this on MP3 or MP4s as well?

[@bytewings3378]

You can use with several file format - 'File uploads are currently limited to 25 MB and the following input file types are supported: mp3, mp4, mpeg, mpga, m4a, wav, and webm.' Said by Open AI

[@melechhaarayot9273]

Can the Updated Whisper operate with newer Python?

[@davidq3083]

Hello, how can I use on MacBook Pro ? How can I Install? Have you time to explain? Great vidéo thanks

[@kobericha]

has whisper been taken down? or will this still work?

[@TS-fh4un]

ありがとうございます。 git のところでつまずいたけれど下記を参考にしてインストールしました。 Windows 10に最新版Git for Windowsをインストールする | ジコログ srt ファイルについては Subtitle Edit で開くと良いです。

[@gregoriobolajuzon8418]

Hi! extreamly usefull video. Im from Argentina and I started to work with it. I just got a problem, and I would like to know if you can help me. When I try to launch whipser it states "failed to create process." Looking around, python39 (the folder whe I installed Python which happends to be on program files) doesnt have the "whisper.EXE" file. The files is on C:/python39. I tried to move it, but keeps on putting this error. Do you know how to solve it?

[@immanueldomeniussihombing_6953]

ERROR: Error [WinError 2] The system cannot find the file specified while executing command git version ERROR: Cannot find command 'git' - do you have 'git' installed and in your PATH?

[@taktaz]

I got the same error as well...don't know what went wrong...

[@Raffa987]

try that one pip install -U openai-whisper

[@taktaz]

Install Git will resolve this issue.

[@romannavratilid]

I have GTX 1070 8GB... how well (or badly) will it run on it...? (and will it even run on it?) it has CUDA but doesnt have RTX if that is needed...? Thank you

[@magicmulder]

You’ll maybe need an older CUDA version. It will still run circles around your CPU unless you have a massive Threadripper.

[@romanbolgar]

Слишком сложно. Раньше можно было просто нажать на exe файл и само все установит. Теперь куча манипуляций. И вообще Насколько я понял работает только с английским. Как дообучить на другом языке со своим голосом? Почему оно не может использовать Видеокарту AMD

[@nearata01]

thank you very useful

[@Creem16]

my torch version fails when i try this tutorial

[@xtrading888]

is whisper ai free to use? do we have to pay fees per audio minute?

[@RyanHunt]

He literally says in the video that it is completely free

[@jefferc8550]

hello!, its this method still working?

[@TroubleChute]

Yes, it is

[@jefferc8550]

@@TroubleChute yep, it does. Tnx!🥰 new sub

[@dr.ignacioglez.9677]

Lo interesante sería que fuera un programa en una ventana normal, donde simplemente arrastraras el .MP3 o el vídeo, y te diera los archivos.

[@Torres-tw3ic]

search for "freesubtitles ai"

[@youdrive311]

esta usando cmd, la clasica pantalla de programacion

[@tohfatulfatima2565]

thank you so much

[@292hello]

king

[@7u5ay]

Genereal question here: why dont u consider using Linux? As I can see u r fairly aware of computers while refusing to adapt to any propiertary Services/Products (your keePass Video for reference) ...Is there anything except probably Gaming that keeps u away from using Linux?

[@batb0y43]

its a good idea to use windows (11 for that matter) because it is the most popular OS in the world hence making it the best to make tutorials about. chances are if you are learning to install python/ relativly simple programs then you are likely not using linux. Windows is just the most relatable for people who need the tutorial. For personal use definatly would recomend linux :)

[@mazdai2]

Better question is what keeps u away from using Windows? There are 1001 reasons for not using Linux.

[@BRICKSINSILK]

anyone can install it....few can stop it holding onto an end-time stamp for 9 years

[@Zayedl]

Thanks for your videos! Is it possible to host ChatGPT?

[@TroubleChute]

You can host much lesser versions, as GPT-J for example. ChatGPT is beyond massive and runs on likely hundreds of gigabytes of VRAM. Think of how much it 'knows'

[@visionaryhera]

@@TroubleChute I actually heard it was only a hundred gigabytes. Still needs alot of processing power to run but you could hypothetically store it on computer.

[@Rocksite1]

Disappointing. Like Rust, but unlike C++ (excepting Qt), they keep breaking breaking backwards compatibility, so yer old stuff doesn't run, and you gotta have a certain version for it to work. Hard to get so good a mix of mature libraries, backwards compatibility, and strong performance out of other programming languages.

[@KryzysX]

❤

[@sivasrinivas3716]

install git and add to path.

[@Oscar08814]

I fucking love you, bye

[@TebogoMotlhale]

I just noticed what appears to be a BIG fatal error/mistake/blunder in both the video as well as in the installation instruction/webpage at the OPs website. He skipped the installation of Git :(( while the information in this video is extremely important and of immense intrinsic value, what is LACKING bigggggg TIME in these sort of videos, is the ability to UNIFY all the installation artifacts/materials into a single package. The fact that EVERYTHING is SCATTERED across VAST spaces on the internet, is a true absurdity in the way these innovative and immensely-valuable Ai tools are PACKAGED (very dismal packaging). For this reason, I am going to create my own video that PACKAGES everything into a single combined package which someone can download from ONE location, and then run a Wizard-drive installation (by using things like the built-in Windows installer system or the Mac Package Installer system). Think about it, intorducing this SIMPLE improvement, is gonna propel this product/solution TO THE MOON !!

[@paulof2028]

WTF are my whisper uding CPU?

[@v01d_r34l1ty]

I'm too early for HD

[@uno9915]

11:15

[@rocren6246]

Enhanced bass in audio makes me throw up.

[@hjoseph777]

About real audio transcribing

[@waynegilbert9517]

a lot of the points are not clear and misleading!

[@boudmohmohboud1007]

complicated

[@sebastianmarkowski3335]

5/5