Had that hooked up to my stereo with the bass pounding when it came out.. Was awesome and still is
@zeyangzhou88796 күн бұрын
Is that directly analysis the audio or two steps that first turn to transcript, and than analysis the text. Like can I input a text let GPT to tell me how long is the audio and how many speakers in the audio?
@nuclear_geek5 күн бұрын
@@zeyangzhou8879 this is audio to audio with no text translation in between.
@ThisHandleIsTakenTooToo8 күн бұрын
audio sounds very AI
@nuclear_geek8 күн бұрын
Using OpenAI voices currently, Eleven Labs is a little better but exponentially more expensive.
@By_Bachelor9 күн бұрын
YESSSSSS
@xrookiefalcon13 күн бұрын
Scam this is just wikipedias featured article. You really think you’ll blow up form this 💀 bum
@nuclear_geek13 күн бұрын
That is in the video description. Thanks for watching!
@xrookiefalcon12 күн бұрын
@@nuclear_geekyou right 💀 mb
@nuclear_geek12 күн бұрын
@@xrookiefalcon all good. I mostly make these for me to get better at creating and editing. I dont like doing voice over so that is where I wrote code to create that part. If other people enjoy them then cool, if not then I take feedback and try to make them better. Just a fun experiment for more my benefit than others.
@almeidachannel712113 күн бұрын
Nice Video ! Sub & Like ! And Love Retro Donkey Games !
@gthomashart392615 күн бұрын
Great data, but the automatic subtitles have appalling grammar and spelling 🙄
@nuclear_geek15 күн бұрын
@@gthomashart3926 yeah, been working on making them better. I edited them more in the last couple videos. Will work on improving them more
@StaleBearFarts15 күн бұрын
BAD
@mycoolvsgame21 күн бұрын
It’s also Hudson Soft final Mario Party game.
@keon-r6l23 күн бұрын
Appreciate the detailed breakdown! I have a quick question: My OKX wallet holds some USDT, and I have the seed phrase. (alarm fetch churn bridge exercise tape speak race clerk couch crater letter). What's the best way to send them to Binance?
@TheRealBombyFromTPOT25 күн бұрын
Cool!😊
@abacoa121528 күн бұрын
you gotta add photos on these things man
@nuclear_geek28 күн бұрын
@@abacoa1215 i do have a script for that and could add them.
@abacoa121528 күн бұрын
@ even though it’s AI you can still be creative with it
@augustineleny8018Ай бұрын
Great job 👍 Can you please help with video tutorial on how you pulled this off? Thanks🙏.
@StupidInternetPeople1Ай бұрын
Amazing doucheFace thumbnail! Clearly intelligent people click stupid face thumbnails more because everyone knows being a lemming moron and doing exactly the same as every other DoucheTuber is a clear sign that your content must be amazing! 😂
@briancunning423Ай бұрын
Brilliant thanks! I was struggling to get this working. Please would you consider doing another tutorial on a continuous listening/speaking version where you don't have to click on the record and submit buttons?
@大名-o9rАй бұрын
厉害了
@frenches1995Ай бұрын
dunno if you feel comfortable with it, but why don't you share the python source code? and just from watching it's not entirely clear where to put the API-key.
@nuclear_geekАй бұрын
@@frenches1995 I did, it is in the video description in my github. There are instructions on the github page. Change the .env.sample file to .env and put your api key there.
@abelaltaye1084Ай бұрын
Thank you!!
@GeovannyVergaraАй бұрын
Thanks a lot for sharing. I was not aware of this new feature until you pointed it out. I will give it a try.
@nuclear_geekАй бұрын
Awesome! Yeah, I was surprised they didn't send out an email or notification on this. Even Microsoft Azure put out a notice. I feel like it still needs some work but still cool to play with.
@supra__editАй бұрын
😊😊😊
@jaredcluff5105Ай бұрын
Having watched your content on how you greet you are a far better presenter than your artificial intelligence. I think you would very well benefit from recording These yourself
@nuclear_geekАй бұрын
Hey I appreciate that! Trying to make more original videos too. I will try to make one tomorrow on the new OpenAI GPT-4o-audio-preview model available in the chat completions endpoint. I also like making these because some of them do better than my original videos and they take about as long to make as it takes me to make the thumbnail, also been a passion of mine to convert source material to something more consumable, not quite perfect but it is worlds better than where I started.
@ClipgathererАй бұрын
Cops sometimes “shoot first, ask questions later” - or never ask questions at all.
@TrigeminalNeuralgiaAIАй бұрын
Are u juish
@Lorens32258Ай бұрын
Look if you're in a struggle with money you should make it a paid service. But if you're not think of the impact you would have on the people who want to learn complex topics but can't because of attention span and the sheer focus needed to comprehensively understand a research paper. What I'm saying is that this tool could help the very people who want to learn, and I think that's the most important thing. My advice make it a github project and make it run completely locally. Give the people the best tool for free add a donation option and if the tool is really good as i think it is. People are gonna donate out of sheer gratitude.
@JoaoPaulo-qi6srАй бұрын
It would be great if something like notebooklm podcast feature could be recreated with openai realtime API. Realtime API isn't there yet whent it comes to automated podcast creation. Clipping the audio conversation is a big problem. Hopefully google will make notebooklm voices available through an API. Google TTS models available aren't as good.
@nuclear_geekАй бұрын
@@JoaoPaulo-qi6sr yeah that is the idea, slowly getting there. I have another version that uses eleven labs that is closer but still leaves some to be desired.
@doendoentheater297Ай бұрын
Great content mate
@nuclear_geekАй бұрын
Thanks for saying so! I appreciate it!
@kora5Ай бұрын
Does this use the new streaming capability of Gradio 5 yet?
@nuclear_geekАй бұрын
Great question! It does not as of now, I am building the whole response before returning it, but I would be happy to look into it and make an update video on it! Great point! I was reading about the new capabilities in gradio last night. I upgraded a couple of my apps but I am still using legacy code.
@kora5Ай бұрын
@@nuclear_geek Anyway, it's a great work, both code and video. Thanks!
@nuclear_geekАй бұрын
@@kora5 Thank you! I appreciate you saying that!
@DannyAlvarez2004Ай бұрын
I would love to try this for instructional design. How can I get access to test this out?
@mikefarls5660Ай бұрын
Nice work!! 👌
@nuclear_geekАй бұрын
Thanks! I appreciate the comment.
@richardprobst7885Ай бұрын
Bad choice are catching up with you Trump good
@realbeeceeАй бұрын
Can we be best friends? LOL
@clee1690Ай бұрын
TRUMP MONEY HAS MADE HIM THE WORST CRIMINAL IN THE WORLD, GRABBING THE CROUCH OF YOUR WIFE AND DAUGHTER IS THAT OK
@donaldfschiff1229Ай бұрын
Get rid of this AI bullcr@p narration!
@nuclear_geekАй бұрын
bruh, this is my most popular video in two months!
@rogerevans96662 ай бұрын
Retired general Edwin Walker was there supporting the segregationists. On April 10, 1963 Lee Harvey Oswald shot a bullet at Walker when he was home in Dallas., coincidentally where Oswald lived at the time. Oswald told his wife when she asked him where he had been that day, Oswald said that he had stopped another Hitler from coming to power.
@WilliamThomas20402 ай бұрын
Great that you built something, but I waited for the full video to see how - that's why people come to KZbin! No tools, no approach, no lessons learned, nothing.
@nuclear_geek2 ай бұрын
I certainly could but I just don’t have an audience for it and it would be handing my work to someone else for nothing. But I understand what you mean and will consider it if I ever get any traction. Mostly I am making this for my own portfolio.
@CryptoPensioner2 ай бұрын
What did you use for text 2 voice? And where's the gitgub link :)
@nuclear_geek2 ай бұрын
Right now I am using OpenAI but I might swap to Eleven Labs to see how it does. I might release the github eventually but right now I am just building my personal workflow with it
@CryptoPensioner2 ай бұрын
@@nuclear_geek You mean Whisper?
@nuclear_geek2 ай бұрын
@@CryptoPensioner Whisper is for speech to text. OpenAI tts-1 and tts-1-hd are the text-to-speech models.
@CryptoPensioner2 ай бұрын
@@nuclear_geek I thought it works both ways. Thanks for the tip!
@YANGTodd-k5w2 ай бұрын
May I have the trial version of this software for evaluation?
@MichaelChin19942 ай бұрын
Interesting way of making information more digestible
@NicksStuff2 ай бұрын
Cool job!
@wmacosx3 ай бұрын
the reason why it sometimes freezes and you have to press enter may be that you selected something in the terminal by mistake, I think that blocks the output and depending on how the app is implemented it may "pause" the underlying process until you press enter (which copies the selection and deselect it)
@nuclear_geek3 ай бұрын
@@wmacosx nah, it just happens from time to time and wanted to mention it. I know what bug you are talking about and if you turn off quick select mode then that will stop happening, but for me the console still hangs from time to time so I thought I would mention it in case it helps anyone else.
@unnreaaal3 ай бұрын
hi. i have this error, can you help? OSError: [WinError 126] The specified module could not be found. Error loading "C: euro\NuclearGeek-Flux-Capacitor\.venv\Lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies.
@nuclear_geek3 ай бұрын
Are you running it from the batch file? Looks like torch isnt installed. Try deleting the .env folder and run the start flux batch file again.
@unnreaaal3 ай бұрын
@@nuclear_geek yes, from batch file, i installed and uninstalled torch earlier, ill try it 1 more time now
@nuclear_geek3 ай бұрын
@@unnreaaal What OS are you using? You might need to edit the start_flux.bat file to get a compatible torch install for your environment
@unnreaaal3 ай бұрын
@@nuclear_geek i have same error in the end of batch file installing. Traceback (most recent call last): File "C: euro\NuclearGeek-Flux-Capacitor\main.py", line 8, in <module> import torch File "C: euro\NuclearGeek-Flux-Capacitor\.venv\Lib\site-packages\torch\__init__.py", line 148, in <module> raise err OSError: [WinError 126] The specified module could not be found. Error loading "C: euro\NuclearGeek-Flux-Capacitor\.venv\Lib\site-packages\torch\lib\fbgemm.dll" or one of its dependencies. Press any key to continue . . .
@unnreaaal3 ай бұрын
@@nuclear_geek Windows 11 Pro 23H2 12/05/2024 22631.3880
@CarlosElKik4 ай бұрын
Thank you so very much! (now to find some time to play with this!)
@nuclear_geek3 ай бұрын
Did you have any luck?
@CarlosElKik3 ай бұрын
@@nuclear_geek Haven't had the time yet 😢but i'll report back when done. Will try M3max128 & 4090. Thanks!
@ShaneZarechian4 ай бұрын
solid vid
@AmanDeep-jp4bm4 ай бұрын
Tried on 4090 and after all the installations the script seems to silently fail. Wondering if you have any idea why, this is what I see in terminal: Running transformer quantize Press any key to continue . . .
@nuclear_geek4 ай бұрын
@@AmanDeep-jp4bm weird it doesn’t give you an error. Can you manually activate the env and just run the script?
@nuclear_geek3 ай бұрын
Did you have any luck? I also had this happen if had been doing other things with my gpu, seems like quantize can close silently if it runs out of memory.
@AmanDeep-jp4bm3 ай бұрын
@@nuclear_geek Hey no I didn't. I just switched to comfy
@nigeldupaigel7 ай бұрын
Did you back test?
@nuclear_geek6 ай бұрын
This version just makes 36 predictions in the future and only predicts on price. My other ones for stock and crypto make historical predictions and use the two neural nets with the lowest MAE scores then use those to make the future predictions: stock-predict-next.vercel.app/
@nigeldupaigel6 ай бұрын
@@nuclear_geek Okay, but how do you validate the model for not capturing randomness? Do you statistics?
@nuclear_geek6 ай бұрын
@@nigeldupaigel Good question! For this example, I just pulled the off the shelf model for fun. For my more extensive stock and crypto price predictions, I use EDA to determine relationships and validate models with MAE. But I am learning and always trying to improve.
@someguy79 Жыл бұрын
portable that's cool. what's the charge last under continuous use? reminder: chatgpt is known to hallucinate
@johnhowetv2 жыл бұрын
What sort of keywords did you use to get that stylistic effect?
@_Roasted_Marshmallow2 жыл бұрын
Trippy af!
@Daofuq2 жыл бұрын
you trippin?
@Trini_Gaming5 жыл бұрын
I don’t have gta 5 on my pc ☹️ but great easy to follo tutorial