I Built a Personal Speech Recognition System for my AI Assistant
ฝัง
- เผยแพร่เมื่อ 21 ก.ค. 2024
- This video shows you how to build your own real time speech recognition system with Python and PyTorch. It walks you through the deep learning techniques that are effective when modeling speech problems, as well as code to build your own.
⭐ Play and Experiment With the Latest AI Technologies at grandline.ai ⭐
This video is the second episode of the series "How to build your own A.I. voice assistant with Pytorch"
• Build an AI Voice Assi...
Github:
github.com/LearnedVector/A-Ha...
Pre-Trained ASR Model:
drive.google.com/file/d/1jcNO... - วิทยาศาสตร์และเทคโนโลยี
this is michael reeves from the universe where he decided to do something useful with his life
LMFAOOOO DIDNT HAVE TO DO HIM LIKE THAT 😭☝️
6:57 yep
Wow, that's mean bro.
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
Imagine the greatest Michael colab. The two of them plus Michael Stevens (vsauce)
I've been looking forward to this part of the series the most! I've been trying to create/run a voice assistant locally, and could not figure out how to apply speech recognition without relying on Google's Python module (which i was trying to avoid for privacy reasons, defeating the purpose of making one) and the HMM basics in my Intro to AI course weren't enough to implement it. This is fantastic.
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
Hey man, I really like how you have written definitions in addition to your speaking, helps a lot.
This is so good, I remember seeing this video a year ago and wishing he'd continued the series.
After watching this video, I literally took off my hat as a mark of respect to this.
Cant thank you enough.
Thanks a million
Wow, just discovered your channel. Great work. I'm just starting my journey into Deep learning and speech recognition. Will be following your progress.
omg, thank you! every other video I look up on this subject is just an ad for a text-speech readers! thanks for going into such detail about your thought process, buut after looking at the rig you have vs the one I've got ... well. . . if it took you a handful of days, it'd take me a week or two LOL great video! thanks a lot!
Loved the high speed speech part!. Well done. Excellent production Mike!. TQ
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
This is excellent! (subscribed!)
I had to quickly brush up my skills for a project I'm working on (will be open sourcing it soon!) - and this video was short, sweet and to the point! Thanks
Brooo.I really dont know whether u coded this or just took reference from something....idrc u are AMMMMMAAAAZZZZIIINGGGGG.Hats off 2 u.U have a great talent man.......u could be the next ceo of any big fours too....
This is awesome! I wish there was more content from you
One Ai hacker to another, keep on going!
I'm building my own wearable AI assistant and this series is **exactly** what I was looking for! Great stuff!
Can you make tutorial
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
dude no way same
@@PonchoManOG does this tutorial still work?
@@morraza3307 yes
Wow...who knew you can make AI teaching so much fun....You should make more videos
we need programmers like you. For advance learning.
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
Loved the video. You really showed me the tools I could use to build my own speech recognition model!
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
When next?? Best Series ever!! Please post next!!
Keep going on and finish the project fast. I'm looking ahead for the project to be finished
Love your War Machine!. I build my first Pentium Pro Dual Proc decades ago. It had a special powersupply and I had to rig my Generic case to fit the Tyan motherboard!. It ran Linux then.
Wo hooo!
This thing for FREE?! And help for us how to make it ours?!
This data worth a HUGE amount of money, but you shared it! I'm so much surprised, in the good term!
Thanks, thanks, thanks for it!!
I really want to make an own Virtual Assistant, so big thanks for this video, for the data and for the help!
Be blessed!
Really dope video. Can't wait to see your next one.
This video is so high quality wish you uploaded more
Totally awesome. Understood about .5% (that's point 5%). Just got my headset set up in Win 10 and am loving it. You're awesome and I bow to your knowledge and expertise....thanks for the cool vid. It was not wasted on my limited knowledge, but it peaked my interest...thanks again...JT
Thanks for the information. Just goes to show that the idea that we evolved is just sheer madness.
Man your a genius man. I wish I could do this. I have some many ideas but dont know where to start.
Omg the video i was looking for thank for making one..... Full support❤
This was insane, gonna try to do similar now, thank you for the informations
This was a really good video dude. Can you tell me how to make the soundwave display that you had while testing the model
Bro you are freaking awesome!!! i love your content, helps a lot.
i think this is a very good video for me ,It can not only let me learn some knowledge, but also make me feel relaxed.thank you
A Cue Stick - used for playing billiards. Acoustic (a-COO-stick) - dealing with sound or audio energy.
this series is so cool! keep it up bro
Loved it Man , Great Video !
I'm enjoying discovering all these smaller ai channels
at the beginning of the video: Oh I see !
6:57 : Oh I ... oh ...
Hey Michele, your videos on AI is fantastic! I haven’t seen any videos lately and I am course what you are doing these days?
Thanks for this..I am going to thoroughly go through the speech recognition your code on Github
Planning on creating my own Jarvis, this video has given me an insight.
Same, I'm going to call mind Alfred and integrate it into my helmet
This is a helpful video. I have a question regarding to the audio augmentation. In my project, the test speaker is not in the train data, so my model performers pretty bad on test set,only 50% accuracy. I try to use the pitch shift to agument my train data but doesn't works well. How should I use audio augmentation for this dataset?
Which hardware accelerator are you going to use, if any? As far as I know the net would be quite slow on the Pi itself hence a hardware accelerator like the Intel NCS or Google Coral would be useful, wouldn't it?
Hey pal, your work's great I love it 🙌🙌
Cool video!!! This will help me a lot with my first NLP project. I wanted to get radio voice data and transcribe it. Any tips on that?
Btw you should come up with a more memorable outro! :D
Keep going bro.full support
thank you man... this was fun to watch
so it is possyble to make the virtual assistant write on another command prompt instead of talking, to use it with a custom text to speech AI? would love to see that
Could you show the code in detail or how I can use it with another model? Could I use a deepspeech model for testingit, too?
which ASR framework did you used? What is your thought on fairseq wav2vec for this purpose?
This is such a amazing video. Congrats! I am wondering about model deployment part. Are you going to share the coding part of ıweb interface? The sound wave and the text that occurs below the sound wave are awesome.
Hey did you found the code for deployment
Can you make a TTS using deep learning? :) I really want to see that.
Eagerly waiting
Thank you for sharing your knowledge !
Hi i work on a little project with (VOSK asr) and i have an issue , on the first its work good , but on timeline it will be slower on the translating words , im not talking about accuracy but on the time of responding
i thinking about loop of while (Python) mybe overload or somthing , can you help me please ? , thanks
What books/material would you recommend to someone wanting to learn to design models like the one you’re detailing around 7:36?
hi great work I just found your channel great job
This is really nice work dude
what a great video man, really inspiring! keep up the good work!
PS: you deserve a better t-shirt bro 😅
Wow I love it! I do have two questions tho. 1 how did you run it in ur Chrome browser. 2 how did u get the cool visual effects for while u were speaking?
You Michael Rock!!!! Way to teach!!!
bro which python version are you using in this project because in python 3.10.4 it is showing that cannot import textprocess from utils
This is awesome! thanks man.
This is what i was looking for, thanks
This is incredibly educational. Thx for sharing ur knowledge for free!
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
can we built it for multiple languages like if we speak in french or any other language it will also convert it in text?
I'm getting an error while installing torch. How to resolve it??
Excellent video, you helped me out tremendously
what's the different with speech_recognition library that we can use without training the data?
Why aren't you uploading more videos? I have already seen this video just came here to say... plzz upload it's been 7 months now!
how did you up-sampled data to create 50 hrs from 1 hr?
bro please explain ho did you make that wave with line of your voice
Can u make a AI voice assistant for Linux or Windows.
Which we can train
Hey, Can I use this to get details from my business software? Can this crawl any software and output the data from it?
THIS ONE WAS REALLY FUNNY gj love keep up the uploads :)
Wao, great video man. Really awesome stuff
Can you make a detailed video how did you manage to make it. I want to learn it.
This man is really a hero 👍💓
Did you have another tutorial as source or how did you do that?
can we use transformes type architecture for audio type data ?
Please post more videos about this!
Hello, I was fascinated with your Speech Recognition System. I wonder, could your system recognize sung musical notes?
that is, instead of words, musical notation.
Can I use deep learning to create a voice activated keyboard? If so, can you please provide some insight?
Great content dude. I haven't seen anything new for the last 7 months though. Hope you're well :)
You are totally awesome bro...👏👏👏
great job! i want to train a speaker identification project..any ideas where to start?
You are gold💛
Do you have any recommendations for how to learn AI? How long did this take to train?
Hello, when will there be a guide to creating your own speech synthesis?
(TTS)
This looks like exactly what I need!
Thanks for posting, I'm gunna follow along and watch tonight.
One question.. Why are you using the auto generated subs on this video 😂😁
Thank you for this video and the insight of how to design a voice recognition system independently from the ground up by an newly to AI. Most videos tell you to connect the internet and to a big server by google or someone else. The only question that I have is why use python and not C or C++, especially since you are running a raspberry pi with limited memory and slower CPU and the natural time restraints of real-time speech recognition?
*Curious;*
Isn't it possible (coding side) to train on TH-cam videos? They also got Subtitles (for verification), random noise and music (for more efficiency) and Unlimited videos to train from.
And does TH-cam legally stop me from doing so?
Good video but are you using Linix at 9:31 and Windows at 9:34? I haven't used Windows in a few years so I didn't know you could do this.
You can make a video about gait recognition biometrics in python
recognized you from your walk model
can you explain how to make authorized voice assistant it means voice assistant respond only owner of the device command by identifying his voice how to make that plz explain
Would this work with a language that is not pre-trained?
will it work if i take from Common Voice another language pack?
Sorry to inform you, but you pretty much rock, at the same time solved this so I don't have to.
what interesting presentation it is!!!
your videos are great bro! 🤝
i got this error when training the data
File "C:\Users\USER\anaconda3\envs\USER\lib\site-packages\pytorch_lightning\trainer\connectors\optimizer_connector.py", line 47, in update_learning_rates
monitor_key = lr_scheduler['monitor']
KeyError: 'monitor'
pytorch_lightning.utilities.exceptions.MisconfigurationException: ReduceLROnPlateau requires returning a dict from configure_optimizers with the keyword monitor=. For example:return {'optimizer': optimizer, 'lr_scheduler': scheduler, 'monitor': 'your_loss'}
Any suggestion on solving this?
Hey ! I just discovered your channel, nice content ! Your model seems overfitting, I think you should evaluate it on a test data (and not the validation). I would be curious to know how it would perform if you do hyperparameter tuning.
th-cam.com/video/iyl53zyz5zk/w-d-xo.html
Bro this is perfect wow thanks
Please upload more!