Run Whisper AI locally on your PC (includes additional instructions to transcribe multiple audio files and different languages and how to use different models): ruclips.net/video/ABFqbY_rmEk/видео.html Article that walks through how to use Whisper AI in the cloud: kevinstratvert.com/2023/01/19/best-free-speech-to-text-ai-whisper-ai/
@@KevinStratvert Outstanding video. I was looking to purchase a product for transcription. As I see it you may have just saved me almost $100.00. Thank You.
Thank you! I am doing some transcribing that I just calculated would take me over 200 hours because of all the background noise and quiet speech, and a billion other issues. You are a life saver and this was able to transcribe one of the files in 10 minutes. It took me 8 hours to transcribe the same file with about 2 words difference! Amazing!!!
Being able to quickly transcribe audio files makes a huge difference in my work writing magazine articles that are based on recorded oral interviews. Thank you SO MUCH for making this accessible! Legitimately game-changing. You're the best.
"unexpected indent" Just to point this out in case it happened to anyone else: When entering that code in the description, it kept failing, getting this "unexpected indent" nonsense. It drove me crazy for a while, but then I figured out the solution: just before the, " !sudo apt update && sudo apt install ffmpeg" part, was a blank character. Once I deleted that blank character (right before " !sudo......"), everything worked fine. Thanks Todd
Kevin this is life saving. I have to transcribe dozens of interview for a masters dissertation. This is sooooo easy and soooo good!! This is the most powerful and high quality free tool on the internet, bar none. Thank you for sharing this!
Thank you SO much. I processed a file on Google yesterday, manually made all the changes - which took about an hour and several re-listens to the audio. This morning I followed your excellent instructions and the model turned out a perfect transcription! I could have saved 1 hr yesterday. Bless you.
This is so helpful! Working in the qualitative research field for years...this just makes the entire process of transcribing so much easier. Thank you!
This is the most beneficial video I have watched so far. Thank you Kelvin for your selfless commitment. I just love the way you take time to explain things making it so easy to understand. God bless you and increase you more.
This is GREAT. I LOVE it!! I have done over 30 transcripts - some way over a hour without an issue. I have not paid anything. I convert all my files to mp3 before transcribing. I use the Lage Model. THANKS!!!
Because of you, I'm able to create subtitles for multiple production houses and have been doing a business for myself using this! Excellent!!! I'll love to explore your channel and see what other magic tricks i can learn!
I must admit, I was a bit intimidated by all the code at first but it was super simple to set up! As a next step I'm going to try to get the translation working between English and Chinese!
@@tdyrc I'm a developer so obviously I know it is, but 99.9% of the world hasn't a clue what Linux is, and that's perfectly fine. Frankly, it's job security for people like me. Admitting that you don't know what something is or that you don't know how to do something is respectable, mature, professional adult behavior.
Thanks! Clear and accurate ... just enough detail ... his instructions actually work! Well done! (The only thing I'd suggest, go into a little more detail into how to insert parms into the command line.)
The fact that it outperforms most human transcribers and other speech-to-text tools in various environments is truly impressive. Thanks for sharing this valuable tutorial and shedding light!
I just tried this tool to get lecture notes and it works very well. It didn't work with mp4 file so I had to convert it to an mp3 file. Thanks for introducing this useful tool.
A bit off topic, but when I was playing around a Speak and Spell IC in the 1980s, I noticed that whispering really helped when trying to encode speech for the IC. Speak and Spell used linear predictive coding to synthesize speech. My analysis code could not handle voiced segments, but worked well on whispering. Back then it would take many hours to encode a couple of seconds of speech on my pitiful 8080 PC. Gotta say I am really impressed with the way speech technology has advanced over the years.
Awesome work, Kevin. Thank you very much for taking the time to make it possible for all of us to follow in your footsteps. My plan for your excellent tutorial is to apply it to two immersive language learning courses I've struggled with. Immersive sounds important but it's meaningless. What it really means is that we'll find ourselves stuck in a language course where only the teacher was capable of understanding & translating the target language. The rest of us will have to guess meanings from the clues and contexts in the material, like a newborn infant. It takes thousands of incidental exposures to build a large vocab. And like me half the other students often only speak their native languages fluently. This is how languages should be taught. Bless you, Kevin.
Hey man, I came back to this video as I am jumping back into my yt journey and want to give you a huge shout for such an indispensable tool! You did an awesome job explaining everything in layman's terms and making this accessible for everyone. Thanks a million 💛
If you're getting the error "/bin/bash: line 1: whisper: command not found" when trying to use Whisper AI, here's a quick fix: Make sure you have the Whisper AI library installed. Just paste this command into your console: pip install openai-whisper Hope this helps!
TY! Kevin, thank you! You have just saved me hours of transcribing the interviews I have with some of the musicians I have spent time with recently! I cannot thank you enough for this. Your tutorials are easy to follow, and you never make me feel like an idiot as you walk through these things. TY!
Wow Kevin this is a game changer. You video explained so well even for non geeky people. I got the following message and I wonder if you can shed light onto it: Change to a standard runtime You are connected to a GPU runtime, but not utilising the GPU. To avoid hitting GPU usage limits, switch to a standard runtime.
Absolutely brilliant! Especially after spending lots of time on sites claiming to transcribe audio -- only to find about the limitations applied to free options once my files were uploaded...
Hi again, I was wondering whether you would suggest new options nowadays to have (even live) audio transcribed faster with Whisper AI. Huggingface seems to offer some great ways to do that but I couldn't figure out what to do exactly. Also, is there a way to "batch process" multiple (small) audio files using Google Colab by a slight modification of the code above? Thank you!
I just love how you thoroughly explained this thing and provided a super easy step-by-step tutorial of this. You're so greattttt!!!! I'm so glad it worked! I have been spending DAYS jotting down all the notes about our company meeting but some words are just hard to comprehend! Hence, I'm indeed thankful for this video! You just earned a new subscriber here! Looking forward to more helpful and practical videos of you in the future!
Excellent Kevin! I've been struggling to transcribe large interviews for our publication.I'm sure this will help solve the problem. You're a great educator! Kudos!
Man, this is absolutely stunning! 😮 I've been working for years now trying to find a solution to accurately transcribe my voice. I suffer from muscular dystrophy, so my voice is really low. I've tried dozens of microphones, programs and interfaces but never found something as powerful and accurate as Whisper. It's completely unbelievable how it can be so accurate! I've tried to transcribe my worst-quality recordings and it got almost 100% accuracy every time, especially when using the "large" model which gives even more impressive results. Thank you very much for sharing and don't hesitate to give us more tips about speech recognition. it's getting better lately but there still is a lot of work to do. By the way, do you know any way to obtain a file without automatic line breaks in the .txt file? It gives me a lot of work to do after transcribing to format the text. (this whole paragraph has been transcribed using whisper ; 100% correct! And I'm french and speak english as a foreign language. And I've got a terrible sore throat killing my voice)
I have the same condition so I feel your pain. My problem with the current speech recognition solutions like DragonDictate is that they struggle a bit with the sound of the ventilator. I think this sort of thing will make a massive difference to people with disabilities.
@@robertsleight8013 The sound of my ventilator often appears as "fart noise" in the transcription I get from MacWhisper 😂( I swear my breath is not that bad. My girlfriend would have told me!) Even though it's pretty funny my texts are generally better without it ; it is very easy to remove from the transcript as it appears on a separate line. I really face no issue issue to transcript anything with that program and I can write in French, English and Spanish again, which I have not been able to do for years now. What a relief!
You just saved me so much time and money for the fan project I'm working on (a tumblr of favorite quotes from podcasts). Thanks so much for this step by step video!
Very informative. However a few things have changed since making this video. When changing runtime type you have two options to chose from now not shown in the video. You have to chose either Python3 or R from the runtime options. Then you can chose between CPU, T4-GPU, A100-GPU and V100-GPU or TPU. So far I tried picking Python3 and t4-GPU and it didn't give me any results. Are you using this, Have you seen this and what do you chose to get your results?
Thank you, Kevin, for assisting me in understanding Whisper AI and installing it. The step-by-step video was enjoyable, and despite the fact that I have no IT expertise of my own, it was straightforward and easy to understand. I had to convert the M4A files MP3 which was also easy for me to understand.
Excellent Kevin, I was a bit daunted by the code etc, but by following your simple instructions step by step it worked. You're a great educator great Thanks!
Very grateful for the quality and clarity of your video. This helped me to transcribe files with all simplicity. Looking forward to discovering more of your content !
Thank you so much! I have just started my Master's course in Japanese language and this was truly life saving tutorial for me. I am grateful for your efforts for teaching us!
I just need to thank you, Kevin, for the large amount of patience, hardwork, kindness and empathy shown in this video. You are a true artist of education. You have become my new standard of a true passionate teacher. I wish I could hug you because the amount of perfectly explained information has moved me so much I want to cry. I hope one day you'll be able to understand how generous this is of you to give to mankind. Srly tough, I think I love you. 👏👏👏👏👏
Edit to my comment two months ago. YES! Your process worked for my Russian video chats! Thank you! My friends in Russia are incredible and now I can read their words in English! THANK YOU!
I just realized that if OpenAI would combine the capabilities of whisper and chat GPT that would make the best assistant imaginable! Like a million times better than Google Assistant
Thanks for the tutorial, Kevin. Note that of the five models, the largest, called large, will not take and extension, like large.en. large.en will throw an error. You have to use just large, like so. : --model large
Thank you very much Kevin for the video and the well-written article. Your instructions helped me a lot. As a result, now I can better understand the content of three online courses and their many video sessions. As a student, paying this to a transcription service would have been prohibitive. Keep up the good work! 👍
Ok, I'm speachless. This is gold! J.R.Ward wasn't counting on that when she took too much to make Darius readable... And not just... Audiobookable. Not going to wait until July. Next book is already here. Thank you so very much
I love your videos, Kevin. They're informative and useful, as well as the step by step instructions and every day examples. This is very cool technology!
I am absolutely gobsmacked! This video and the tools you have provided are AMAZING. I have been struggling for a week to find a free way using AI to transcribe lengthy interviews and have had nothing but failures, or options that are only free trials. The output is flawless. I will have to do some more work organizing who is saying what etc, which requires me to listen to the file and make edits, but the most time consuming heavy lifting work is done. Thank you and I can't wait to check out your other videos!
Omg, as a student in sociology this is definitely going to save me hours of retranscription work... I need to try it out as soon as possible ! Thank you for your great video !
Hi Kevin - thank you! Excellent tutorial. I am not well-versed in code work, but I was able to follow your action steps because your visual navigation from one window to the next was fairly easy to follow. Well done! Look forward to seeing what else I can learn from you in the future.
Thank you so much, Kevin! I am writing a book and doing tons of interviews and was looking for softwares and apps that could help me with the transcription and wouldn't break the bank (I live in Brazil and right now 1 dollar is worth 5 reais) and this is not only free but the quality is so much better than the paid options I was trying out! This will save me dozens, even hundreds of hours! Thank you!
Hi Kevin..Thank you for such clear step-by-step instructions. The first couple of times I was able to do it but then when I tried a .m4p file (about 45 min long) I am getting a "/bin/bash: line 1: whisper: command not found". Can you please explain this in layman's terms? what do I do to fix this? Thank you
Magnificent tool Kevin! Thanks a lot. Suggestion: If you wanna make a transcript in another language I would use the following command !whisper "NAME OF THE FILE.EXTENSION" --model medium --language es This is an example for a Spanish file, besides you can see that all the other languages are in the list with the !whisper -h command that you explained. Greetings from Bolivia!
This is one of the best hands-on videos on showing you "how to do". Thank you so much for sharing this with us, Kevin. I have a quick question. What is the Whisper command syntax if I would like to translate the orginal mp3 (English) to German? Thank you.
Hi Kevin, Can you please explain more about transcribing in different language, where do we need to type the language that we want to transcribe? Thanks in advance. Keep up good work.
This is so helpful for my dissertation research. Thank you Kevin! I do have a question though. How could I tell the program that I have two languages in a text? It seems to recognize one, but sometimes drops the other. The discussions happen in Spanish and English.
Hi Kevin. I'm a big fan! Why dont you post some more videos on Chat GPT and Open AI Playground....and further downstream application of them? I know there are many videos across the internet, but watching your videos are real informative. I loved the whisper one and immediately installed. Have a great day !
Thank you Kevin. This was exactly what I needed. It is saving me so much time. You are a great educator! I will continue watching your videos and learning from you. Blessings!
Hello, has anyone experienced a similar error code? ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts. llmx 0.0.15a0 requires cohere, which is not installed. llmx 0.0.15a0 requires openai, which is not installed.
I tried this for meeting 1 yesterday and it worked so beautifully. It picked up on different accents and punctuated everything beautifully. Trying a second meeting today, I just get "/bin/bash: whisper: command not found". I'm too unskilled to work out what is going wrong. Tried it a few times and event started from scratch. Update: It appears that I just had to reinstall whisper ... for some reason :) Phew!
@@dookishooz It just picked up the transcript but I could swiftly edit whilst listening back. And having sat in on the meeting, I kinda knew already. Weirdly on a recent meeting it included all the ums and ahs and all the repeated and superfluous words “um, you know, you know, um, er, blah blah” and yet in the first runs it edited those out perfectly.
Great! Thank you. Adding codes may not be the easiest way for regular users, and you didn't show us the alternative method to do it without codes. However, I tried it with Arabic short video and had hard time to finish the process. Could you please make a video to show us other software programs on PC to have auto-captions in various languages? By the way, I liked your Clipchamp video, but when it comes to auto-captions in Arabic, Clipchamp keeps freezing and doesn't finish the task.
I have same freezing issue with Finnish in Clipchamp. Short English clip worked well IIRC. I have very entry level HP PC that was 239 euros on sale (RRP 429 euros). Bought it about 2,5 years ago. Do you have more powerful PC than mine, because I'm suspecting that my PC not being efficient/powerful enough could be contributing factor to thi issue with Clipchamp?
@@senkinholkkaaja Clipchamp is all web based, so your PC shouldn't impact whether it's able to transcribe. You could shoot Clipchamp support a note telling them about the issue. Alternatively, this method should work well irrespective of your computer specs.
Such an awesome video.. I wish everybody explained stuff this way. To the spot, clear, patient and took us through every step so nothing is intimidating
Thank you very much, Kevin! This is such a big help to me. I don't know the precise accuracy I just experienced in using this Whisper AI method of transcription compared to my manual transcription I previously experienced, but if I were to guess I would give this something above 90% accuracy. Yes, there were some errors in the transcription. For example, the speaker would say, ". . . gonna," and Whisper AI transcribed it as, ". . . going to." Well, using the 80/20 rule, I think I can live with errors like that for the work I'm doing. Other errors were Whisper AI transcribed a word that either wasn't spoken by the speaker or did not transcribe a word spoken by the speaker. Still, 90%+ accuracy with accurate timestamps is a plus for me. What I like most is it saved me from spending $99 on a transcription subscription service. I've yet to find anything that can beat free! It was very easy to follow your instructions. You're clearly a gifted instructor. God bless you! Thank you for sharing!
this was a great video, thank you so much for it. i'll share two things that helped me, after the first attempt didn't work: 1) the file name alone didn't seem to work; i needed to include the type of file, thus 'file_name' was changed to 'file_name.m4a' and 2) i noticed that in your video you typed 'model medium.en' thus i did the same. Now, I don't have the expertise to know if it was one or the other or both that enabled the successful output but it at least worked successfully doing these two things. thank you again
Kevin, Thank you so much! Typically, I won't use a new tool if I am under time pressure because there is always a risk of something unexpected happening. This worked flawlessly and is immediately paying dividends.😀
Excepcional! Eu aqui lutando pra achar uma opção gratuita pra transcrever minhas gravações, até que cheguei aqui e meus problemas foram resolvidos. Segui o tutorial e deu certinho, funcionando de forma magnífica. Muito obrigado!
Love you man thank you so much. All these free websites were so scammy and were asking to pay to get the text from audio (my audio file was approx 1 hr) this worked like a charm ! thank you
This was a very helpful video, which not only solved my dilemma, but also opened up new doors for me, with regards to using Google Collaboratory. Thank you so much.
Thank you so so much for this video! It works amazingly! I have been looking for a simple and free tool to transcribe audio to text. I am very grateful to you. I am saving lots of time and getting very good-quality transcribes.
Kevin, I am a big fan of yours. You explain every bit to us noobs very well. whenever, I am not understanding a topic I search for your videos and you often have better explanation of that. You are my inspiration to start a RUclips channel like yours in my native language i.e Urdu. I just want to request that if you can teach us python like this that would be very helpful. Thanks Man !
The transcription jobs to which I applied notes that they pay by the hour of spoken text, not by the hour of work. I'm glad we have tools that can make such a transaction work out in favor of the worker. I currently teach, but would like to become a digital nomad at some point. My fingers are crossed that transcribing with digital tools works out for me.
Thanks Kevin, even though some of the things changed in a sudden and some minor software cannot be installed.. i still can use the free 50 mins to make my meeting summary... thanks for teaching us! Appreciated!
Thank you so much for this I had whisper in my computer with SE, but It broke and I was training to transcribe some interviews for my final disertation online. This was incredibly helpful.
After searching for so long time and so many programs, finely i found the way to convert video into text, for me to make it easier, i convert it to from mp4 video to mp3 audio then i used your way., THANK YOU
Run Whisper AI locally on your PC (includes additional instructions to transcribe multiple audio files and different languages and how to use different models): ruclips.net/video/ABFqbY_rmEk/видео.html
Article that walks through how to use Whisper AI in the cloud: kevinstratvert.com/2023/01/19/best-free-speech-to-text-ai-whisper-ai/
Thank you
Do I understand it correctly that it costs money if I use the website, but it is free if I use the software via Transscribe_Audio.ipynb?
@@stefanjordan3942 yes
@@stefanjordan3942 You can run this on Google Colab for free. Alternatively, you can install directly onto your PC, also for free.
@@KevinStratvert Outstanding video. I was looking to purchase a product for transcription. As I see it you may have just saved me almost $100.00. Thank You.
Thank you! I am doing some transcribing that I just calculated would take me over 200 hours because of all the background noise and quiet speech, and a billion other issues. You are a life saver and this was able to transcribe one of the files in 10 minutes. It took me 8 hours to transcribe the same file with about 2 words difference! Amazing!!!
Being able to quickly transcribe audio files makes a huge difference in my work writing magazine articles that are based on recorded oral interviews. Thank you SO MUCH for making this accessible! Legitimately game-changing. You're the best.
Hello Mary, I could help if there's extra work to be done. Thanks in advance.
frr
"unexpected indent" Just to point this out in case it happened to anyone else:
When entering that code in the description, it kept failing, getting this "unexpected indent" nonsense. It drove me crazy for a while, but then I figured out the solution: just before the, " !sudo apt update && sudo apt install ffmpeg" part, was a blank character. Once I deleted that blank character (right before " !sudo......"), everything worked fine.
Thanks Todd
Thanks a million
wow, thank you, it worked!
Thank you soooo much!!! I was about to go crazy... haha
Your Comment helps a lot to me. I was facing the same error. When I follow your comment instructions it worked me. Thank you
omg thank you
Kevin this is life saving. I have to transcribe dozens of interview for a masters dissertation. This is sooooo easy and soooo good!! This is the most powerful and high quality free tool on the internet, bar none. Thank you for sharing this!
Did you like your performance? Are you still using this? Do I need a paid tool?
@@oguzr Yes, I am transcribing hundreds of interviews for a masters degree. The performance is stunning. You do not need a paid tool.
@@francoisdavel1786 Thank you so much.
Thank you SO much. I processed a file on Google yesterday, manually made all the changes - which took about an hour and several re-listens to the audio. This morning I followed your excellent instructions and the model turned out a perfect transcription! I could have saved 1 hr yesterday. Bless you.
This is so helpful! Working in the qualitative research field for years...this just makes the entire process of transcribing so much easier. Thank you!
This is the most beneficial video I have watched so far. Thank you Kelvin for your selfless commitment. I just love the way you take time to explain things making it so easy to understand. God bless you and increase you more.
Selfless commitment, what? He makes around $44k a month from youtube on this channel. Nothing selfless at all Carol.
I was about to say the same. Amazing work!
This is GREAT. I LOVE it!! I have done over 30 transcripts - some way over a hour without an issue. I have not paid anything. I convert all my files to mp3 before transcribing. I use the Lage Model. THANKS!!!
This is the simplest and no nonsense tutorial I have seen. Thank you so much, Kevin!
Because of you, I'm able to create subtitles for multiple production houses and have been doing a business for myself using this! Excellent!!! I'll love to explore your channel and see what other magic tricks i can learn!
I was wondering though... I transcribed a large video and do have certain mistakes. Is there a way to reprompt it to make a revision?
I must admit, I was a bit intimidated by all the code at first but it was super simple to set up! As a next step I'm going to try to get the translation working between English and Chinese!
There are just two Linux commands lmao.
@@tdyrc Not afraid to admit I was still intimidated 😂
@@tdyrc I'm a developer so obviously I know it is, but 99.9% of the world hasn't a clue what Linux is, and that's perfectly fine. Frankly, it's job security for people like me. Admitting that you don't know what something is or that you don't know how to do something is respectable, mature, professional adult behavior.
@@JeffSu LOZZ- way to jump in!
@@PeacefulPariah I totally agree, because that is where the value of developers like you and me comes from.
Thanks! Clear and accurate ... just enough detail ... his instructions actually work! Well done! (The only thing I'd suggest, go into a little more detail into how to insert parms into the command line.)
Excellent Kevin, I was a bit daunted by the code etc, but by following your simple instructions step by step it worked. You're a great educator.
The fact that it outperforms most human transcribers and other speech-to-text tools in various environments is truly impressive. Thanks for sharing this valuable tutorial and shedding light!
your response is AI generated.
I just tried this tool to get lecture notes and it works very well. It didn't work with mp4 file so I had to convert it to an mp3 file. Thanks for introducing this useful tool.
Mp4 file worked for me and it is very accurate. Amazing!
@@franciskapola2542 maybe it's a new update. It have been a while since I used this.
A bit off topic, but when I was playing around a Speak and Spell IC in the 1980s, I noticed that whispering really helped when trying to encode speech for the IC. Speak and Spell used linear predictive coding to synthesize speech. My analysis code could not handle voiced segments, but worked well on whispering. Back then it would take many hours to encode a couple of seconds of speech on my pitiful 8080 PC. Gotta say I am really impressed with the way speech technology has advanced over the years.
Awesome work, Kevin. Thank you very much for taking the time to make it possible for all of us to follow in your footsteps. My plan for your excellent tutorial is to apply it to two immersive language learning courses I've struggled with. Immersive sounds important but it's meaningless. What it really means is that we'll find ourselves stuck in a language course where only the teacher was capable of understanding & translating the target language. The rest of us will have to guess meanings from the clues and contexts in the material, like a newborn infant. It takes thousands of incidental exposures to build a large vocab. And like me half the other students often only speak their native languages fluently. This is how languages should be taught. Bless you, Kevin.
who knows kelvin himself can be an AI avator
Joke nhi chala Tera chal nikal
😄
@@himanshuvlogz67 🤣
Agree
Or an AI aviator
Hey man, I came back to this video as I am jumping back into my yt journey and want to give you a huge shout for such an indispensable tool! You did an awesome job explaining everything in layman's terms and making this accessible for everyone. Thanks a million 💛
If you're getting the error "/bin/bash: line 1: whisper: command not found" when trying to use Whisper AI, here's a quick fix:
Make sure you have the Whisper AI library installed. Just paste this command into your console: pip install openai-whisper
Hope this helps!
Bro You saved my day man God sent indeed
@kevinstratvert - Maybe add this to your description and instructions? It saved me as well!
Thank you man, much appreciated for your comment (peace)
Thank you so much
TY! Kevin, thank you! You have just saved me hours of transcribing the interviews I have with some of the musicians I have spent time with recently! I cannot thank you enough for this. Your tutorials are easy to follow, and you never make me feel like an idiot as you walk through these things. TY!
Wow Kevin this is a game changer. You video explained so well even for non geeky people. I got the following message and I wonder if you can shed light onto it: Change to a standard runtime
You are connected to a GPU runtime, but not utilising the GPU. To avoid hitting GPU usage limits, switch to a standard runtime.
Absolutely brilliant! Especially after spending lots of time on sites claiming to transcribe audio -- only to find about the limitations applied to free options once my files were uploaded...
Hi again, I was wondering whether you would suggest new options nowadays to have (even live) audio transcribed faster with Whisper AI. Huggingface seems to offer some great ways to do that but I couldn't figure out what to do exactly.
Also, is there a way to "batch process" multiple (small) audio files using Google Colab by a slight modification of the code above? Thank you!
I just love how you thoroughly explained this thing and provided a super easy step-by-step tutorial of this. You're so greattttt!!!! I'm so glad it worked! I have been spending DAYS jotting down all the notes about our company meeting but some words are just hard to comprehend! Hence, I'm indeed thankful for this video! You just earned a new subscriber here! Looking forward to more helpful and practical videos of you in the future!
Excellent Kevin! I've been struggling to transcribe large interviews for our publication.I'm sure this will help solve the problem. You're a great educator! Kudos!
Hi,
How has it been using it so far? It isn't working currently for me and I can't figure out why yet.
Cheers
@@akinbiyishakir556 KEVIN?
Man, this is absolutely stunning! 😮
I've been working for years now trying to find a solution to accurately transcribe my voice. I suffer from muscular dystrophy, so my voice is really low. I've tried dozens of microphones, programs and interfaces but never found something as powerful and accurate as Whisper.
It's completely unbelievable how it can be so accurate! I've tried to transcribe my worst-quality recordings and it got almost 100% accuracy every time, especially when using the "large" model which gives even more impressive results.
Thank you very much for sharing and don't hesitate to give us more tips about speech recognition. it's getting better lately but there still is a lot of work to do.
By the way, do you know any way to obtain a file without automatic line breaks in the .txt file? It gives me a lot of work to do after transcribing to format the text.
(this whole paragraph has been transcribed using whisper ; 100% correct! And I'm french and speak english as a foreign language. And I've got a terrible sore throat killing my voice)
I don't know about your experience, but I have just used JS based SpeechRecognition API. My wife is Chinese and she was very impressed by the results.
Salut Johann, as-tu obtenu une transcription en français ? Je viens de faire un essai et il me le "met" en anglais...merci
I have the same condition so I feel your pain. My problem with the current speech recognition solutions like DragonDictate is that they struggle a bit with the sound of the ventilator. I think this sort of thing will make a massive difference to people with disabilities.
@@robertsleight8013 Wish that the technology was actually focused on helping people. Instead we got CIA run SEO and TikTok.
@@robertsleight8013 The sound of my ventilator often appears as "fart noise" in the transcription I get from MacWhisper 😂( I swear my breath is not that bad. My girlfriend would have told me!) Even though it's pretty funny my texts are generally better without it ; it is very easy to remove from the transcript as it appears on a separate line. I really face no issue issue to transcript anything with that program and I can write in French, English and Spanish again, which I have not been able to do for years now. What a relief!
You just saved me so much time and money for the fan project I'm working on (a tumblr of favorite quotes from podcasts). Thanks so much for this step by step video!
Very informative. However a few things have changed since making this video. When changing runtime type you have two options to chose from now not shown in the video. You have to chose either Python3 or R from the runtime options. Then you can chose between CPU, T4-GPU, A100-GPU and V100-GPU or TPU. So far I tried picking Python3 and t4-GPU and it didn't give me any results. Are you using this, Have you seen this and what do you chose to get your results?
did you find a solution about this pb , bcs I have the same issue rn
same problem@@kevinltt9172
@@kevinltt9172 Has anyone followed up on this? I'm running into the same issues
same problem 🥲
Yes, I faced the exact problem too.
Thank you, Kevin, for assisting me in understanding Whisper AI and installing it. The step-by-step video was enjoyable, and despite the fact that I have no IT expertise of my own, it was straightforward and easy to understand. I had to convert the M4A files MP3 which was also easy for me to understand.
Excellent Kevin, I was a bit daunted by the code etc, but by following your simple instructions step by step it worked. You're a great educator great Thanks!
Very grateful for the quality and clarity of your video. This helped me to transcribe files with all simplicity.
Looking forward to discovering more of your content !
I already knew about Whisper but I was using the base model coz of my GPU. Thanks to you now I can use Colab and the medium model 🔥🔥
Thank you so much! I have just started my Master's course in Japanese language and this was truly life saving tutorial for me. I am grateful for your efforts for teaching us!
Could you please show me the code you used ?I failed to make it transcribe to any language other than english
I just need to thank you, Kevin, for the large amount of patience, hardwork, kindness and empathy shown in this video.
You are a true artist of education.
You have become my new standard of a true passionate teacher.
I wish I could hug you because the amount of perfectly explained information has moved me so much I want to cry.
I hope one day you'll be able to understand how generous this is of you to give to mankind.
Srly tough, I think I love you. 👏👏👏👏👏
wish i can say the same
Same hereeeeee thank you so much for the videooo 🥲🥲🥲
Hmmm, perhaps/
Edit to my comment two months ago.
YES! Your process worked for my Russian video chats! Thank you! My friends in Russia are incredible and now I can read their words in English! THANK YOU!
I just realized that if OpenAI would combine the capabilities of whisper and chat GPT that would make the best assistant imaginable! Like a million times better than Google Assistant
I believe that’s the eventual intent 👍
Thanks for the tutorial, Kevin. Note that of the five models, the largest, called large, will not take and extension, like large.en. large.en will throw an error. You have to use just large, like so. : --model large
Huh?
Thank you very much Kevin for the video and the well-written article. Your instructions helped me a lot. As a result, now I can better understand the content of three online courses and their many video sessions. As a student, paying this to a transcription service would have been prohibitive. Keep up the good work! 👍
Ok, I'm speachless. This is gold!
J.R.Ward wasn't counting on that when she took too much to make Darius readable... And not just... Audiobookable.
Not going to wait until July. Next book is already here.
Thank you so very much
Thanks for this, Kevin. I was hoping there was a free option for speech to text, looks like this is it. Looking forward to trying it out.
Kevin is a wonderful communicator: pithy, simple and knowledgeable. Explains the subject matter concisely and accurately.
I love your videos, Kevin. They're informative and useful, as well as the step by step instructions and every day examples. This is very cool technology!
Thank you, Kevin! You are so resourceful and put out such helpful videos. Keep up the excellent work!
Thank you! 👍
I am absolutely gobsmacked! This video and the tools you have provided are AMAZING. I have been struggling for a week to find a free way using AI to transcribe lengthy interviews and have had nothing but failures, or options that are only free trials. The output is flawless. I will have to do some more work organizing who is saying what etc, which requires me to listen to the file and make edits, but the most time consuming heavy lifting work is done. Thank you and I can't wait to check out your other videos!
Charlie was as well... broke a tooth
Amazing video Kevin, always love watching your stuff. Your teaching skill is top notch, you actually make people want to learn!
Omg, as a student in sociology this is definitely going to save me hours of retranscription work... I need to try it out as soon as possible ! Thank you for your great video !
Fantastic Kevin! I've learned so much from your content.
Glad to hear it!
Hi Kevin - thank you! Excellent tutorial. I am not well-versed in code work, but I was able to follow your action steps because your visual navigation from one window to the next was fairly easy to follow. Well done! Look forward to seeing what else I can learn from you in the future.
Thank you so much, Kevin! I am writing a book and doing tons of interviews and was looking for softwares and apps that could help me with the transcription and wouldn't break the bank (I live in Brazil and right now 1 dollar is worth 5 reais) and this is not only free but the quality is so much better than the paid options I was trying out! This will save me dozens, even hundreds of hours! Thank you!
Glad it was helpful!
Hi Kevin..Thank you for such clear step-by-step instructions. The first couple of times I was able to do it but then when I tried a .m4p file (about 45 min long) I am getting a "/bin/bash: line 1: whisper: command not found". Can you please explain this in layman's terms? what do I do to fix this? Thank you
same problem here
This is such a great tutorial! And your voice is somewhat calming to someone(me) who's a total tech idiot! Thanks a lot!
Magnificent tool Kevin! Thanks a lot.
Suggestion: If you wanna make a transcript in another language I would use the following command
!whisper "NAME OF THE FILE.EXTENSION" --model medium --language es
This is an example for a Spanish file, besides you can see that all the other languages are in the list with the !whisper -h command that you explained. Greetings from Bolivia!
You can leave off the --language es and it should auto detect, but you'll save some compute if you just tell it the language with an argument. 👍
The content was fantastic and very easy to learn from. It was the best eight minutes and 21 seconds I've spent on learning something in a long while.
This is one of the best hands-on videos on showing you "how to do". Thank you so much for sharing this with us, Kevin. I have a quick question. What is the Whisper command syntax if I would like to translate the orginal mp3 (English) to German? Thank you.
I believe it currently only translates from other languages to English. Hopefully any language to any language is coming soon 👍
OMG! This is a game-changer for me. Finally something I can use for my business that will give me an incredible edge with my efficiency.
Hi Kevin, Can you please explain more about transcribing in different language, where do we need to type the language that we want to transcribe? Thanks in advance. Keep up good work.
!whisper "your file" --model medium --language xy
@@lrrrruleroftheplanetomicro6881 Thank you so much
I've completed the task easily. It was very helpful for my video project. Thanks for your continuous support. Wish you a good luck.
This is so helpful for my dissertation research. Thank you Kevin! I do have a question though. How could I tell the program that I have two languages in a text? It seems to recognize one, but sometimes drops the other. The discussions happen in Spanish and English.
Hi Kevin. I'm a big fan! Why dont you post some more videos on Chat GPT and Open AI Playground....and further downstream application of them? I know there are many videos across the internet, but watching your videos are real informative. I loved the whisper one and immediately installed. Have a great day !
Yah, why not?
Thank you Kevin. This was exactly what I needed. It is saving me so much time. You are a great educator! I will continue watching your videos and learning from you. Blessings!
Great way to generate captions! thank you for the video! Can you share a coding example for transcribing other language, for example Chinese? Thanks!
!whisper "your file" --model medium --language xy
you have no idea just how much you have made my workflow so much more easier with this video. Thanks bro.
Hello, has anyone experienced a similar error code? ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
llmx 0.0.15a0 requires cohere, which is not installed.
llmx 0.0.15a0 requires openai, which is not installed.
Thanks!
I tried this for meeting 1 yesterday and it worked so beautifully. It picked up on different accents and punctuated everything beautifully. Trying a second meeting today, I just get "/bin/bash: whisper: command not found". I'm too unskilled to work out what is going wrong. Tried it a few times and event started from scratch.
Update: It appears that I just had to reinstall whisper ... for some reason :) Phew!
did the transcript identify different speakers? or was it all one long transcript?
@@dookishooz It just picked up the transcript but I could swiftly edit whilst listening back. And having sat in on the meeting, I kinda knew already.
Weirdly on a recent meeting it included all the ums and ahs and all the repeated and superfluous words “um, you know, you know, um, er, blah blah” and yet in the first runs it edited those out perfectly.
@@dookishooz does it gives error message when there are two speakers?
@@raizenman no, it just transcribes the whole thing as if it was all one speaker
This is brilliant. Going to be useful for content creators. Great tutorial. 👌🏻
It worked, though took a long time to transcribe a 15-minute discussion. But really enjoyed the output. Thanks man for enlightening us.
Great! Thank you. Adding codes may not be the easiest way for regular users, and you didn't show us the alternative method to do it without codes. However, I tried it with Arabic short video and had hard time to finish the process. Could you please make a video to show us other software programs on PC to have auto-captions in various languages? By the way, I liked your Clipchamp video, but when it comes to auto-captions in Arabic, Clipchamp keeps freezing and doesn't finish the task.
I have same freezing issue with Finnish in Clipchamp. Short English clip worked well IIRC. I have very entry level HP PC that was 239 euros on sale (RRP 429 euros). Bought it about 2,5 years ago. Do you have more powerful PC than mine, because I'm suspecting that my PC not being efficient/powerful enough could be contributing factor to thi issue with Clipchamp?
@@senkinholkkaaja Clipchamp is all web based, so your PC shouldn't impact whether it's able to transcribe. You could shoot Clipchamp support a note telling them about the issue. Alternatively, this method should work well irrespective of your computer specs.
This is one of the best DIY videos i have ever watched. Simple, straight and accurate. You have definitely gained a subscriber
Thank you very much Kevin! ¿What tweaks in the prompts will I have to make if the language of the audio file is Spanish or any other foreign language?
do not right medium.ed Just medium
@@krishnafitpro thanks dude
@@krishnafitpro As you say. To force a language:
!whisper "your file" --model medium --language xy
What would be the code for MP3 in German to German text? Please help.
This is litereally the best video on this subject I've seen to date. Well done, sir.
tutorial is outdated, the runtimes are different now. Creator has not responded to other queries about this change from 4 months ago.
Such an awesome video.. I wish everybody explained stuff this way. To the spot, clear, patient and took us through every step so nothing is intimidating
Waooooooooooooo man, you have changed my life with this video, I love how you explain everything very clear and straight to the point. Thanks a lot!
Thank you very much, Kevin! This is such a big help to me. I don't know the precise accuracy I just experienced in using this Whisper AI method of transcription compared to my manual transcription I previously experienced, but if I were to guess I would give this something above 90% accuracy. Yes, there were some errors in the transcription. For example, the speaker would say, ". . . gonna," and Whisper AI transcribed it as, ". . . going to." Well, using the 80/20 rule, I think I can live with errors like that for the work I'm doing. Other errors were Whisper AI transcribed a word that either wasn't spoken by the speaker or did not transcribe a word spoken by the speaker. Still, 90%+ accuracy with accurate timestamps is a plus for me. What I like most is it saved me from spending $99 on a transcription subscription service. I've yet to find anything that can beat free! It was very easy to follow your instructions. You're clearly a gifted instructor. God bless you! Thank you for sharing!
Kevin, this video is really helpful. It cut my work hours in half. I appreciate your hard work.
Thank you very much, Kevin. I followed the instructions to the letter and applied them. It worked perfectly for me. Thanks again for your help.
this was a great video, thank you so much for it. i'll share two things that helped me, after the first attempt didn't work: 1) the file name alone didn't seem to work; i needed to include the type of file, thus 'file_name' was changed to 'file_name.m4a' and 2) i noticed that in your video you typed 'model medium.en' thus i did the same. Now, I don't have the expertise to know if it was one or the other or both that enabled the successful output but it at least worked successfully doing these two things. thank you again
Kevin, Thank you so much! Typically, I won't use a new tool if I am under time pressure because there is always a risk of something unexpected happening. This worked flawlessly and is immediately paying dividends.😀
Excepcional! Eu aqui lutando pra achar uma opção gratuita pra transcrever minhas gravações, até que cheguei aqui e meus problemas foram resolvidos. Segui o tutorial e deu certinho, funcionando de forma magnífica. Muito obrigado!
Love you man thank you so much. All these free websites were so scammy and were asking to pay to get the text from audio (my audio file was approx 1 hr)
this worked like a charm ! thank you
This was a very helpful video, which not only solved my dilemma, but also opened up new doors for me, with regards to using Google Collaboratory. Thank you so much.
Thank you so so much for this video! It works amazingly! I have been looking for a simple and free tool to transcribe audio to text. I am very grateful to you. I am saving lots of time and getting very good-quality transcribes.
Thank you so much Kevin!! You can't imagine how much you help me with your videos! Greetings from Argentina
Kevin, I am a big fan of yours. You explain every bit to us noobs very well. whenever, I am not understanding a topic I search for your videos and you often have better explanation of that. You are my inspiration to start a RUclips channel like yours in my native language i.e Urdu.
I just want to request that if you can teach us python like this that would be very helpful. Thanks Man !
Thanks
The transcription jobs to which I applied notes that they pay by the hour of spoken text, not by the hour of work. I'm glad we have tools that can make such a transaction work out in favor of the worker. I currently teach, but would like to become a digital nomad at some point. My fingers are crossed that transcribing with digital tools works out for me.
Could you tell me the company's name I would like to apply too
Bro, saved me time man. Thanks for the super generous and clear explanation
Excellent video! Whisper outperforms all the alternatives out there re. accuracy, speed, and simplicity. Thanks at stack!
I can't say that the process was easy but, it worked and the result is great. Thank you
Thank you very much, sir!
I can't thank you enough. I want to jump for joy, God bless you for sharing, this is so helpful.
Thanks Kevin, even though some of the things changed in a sudden and some minor software cannot be installed.. i still can use the free 50 mins to make my meeting summary... thanks for teaching us! Appreciated!
Thank you so much for this I had whisper in my computer with SE, but It broke and I was training to transcribe some interviews for my final disertation online. This was incredibly helpful.
This is awesome! Had no idea, it was this easy to transcribe audio files.
After searching for so long time and so many programs, finely i found the way to convert video into text, for me to make it easier, i convert it to from mp4 video to mp3 audio then i used your way., THANK YOU
Thanks alot ! It works well for me. I installed the API step by step following your instructions in the video.
man, your video is a blessed from God, I never knew this existed FREELY till now
Thanks Kevin! This tool literally saved me hours of work!