I am using a combination of "faster-whisper" and "whisper. cpp" offline. I will use "faster-whisper" for fast machines or servers with GPU and whisper in my project.cpp will be on a regular laptop on CPU. Thanks for sharing; your demo was crystal clear; keep it up. New subscriber
i have a problem i dont know why dont detect fine my voice with de base en model, i even scream and no all is detected. i using integrated microphone but is a good microphone. i will test with another microphone soon, but something tellme that not will work neither.
@@Dixon105 should be a problem with your microphone, try your settings and make sure your mic is well setup, that should work. Also I noticed some issues as well with airports, when I have them on it picks up sound not in the best way with them on.
thanks for you content for West Indies in the carribbeans. Guadeloupe :-) I am curious to know on what kind of machine you are working on ? Is there a big GPU ? I saw metal, normal apple laptop ?
What different with speech recognition library?? As i know speech recognition support engine like whisper,watsonx,and google speech, but for offline it use vosk by default
How do you get the make command to work on windows?, i got the make command but i just get error saying cc not found and someone said gcc=cc but i dont know how to do anything from there
Hi, noob here.. Trying to figure out how to get the `make` working from VSCode terminal, on windows so far I installed MSYS2 added C:\msys64\usr\bin and C:\msys64\mingw64\bin to PATH env variables but... still says command not recognized..
@@codewithbro95 I had to install Visual Studio and build the C code from there or something, but it didn't build the microphone one, and I don't know how to add it to the build step, so kinda gave up, also was trying to get my AMD GPU to work with ZLUDA which is a library that should make CUDA code work on AyyMD, but no luck there either even with AI helping with troubleshooting..
Your english is excellent. may i make a suggestion - python is not pronounced pie-ton but pie-thon - with the 'th' being the same as the 'th' in 'this'
can you put this offline whisper with a local llm model lets say phi3 to get reply based on whisper? i mean lets see how fast it can actually put out what the llm model will reply, this way you can make an offline ai assistant with no latency in responses and local 100 %
@@codewithbro95 cool nice job keep it up, can you also add a way to use phi3 llm with phidata as well for Local RAG and also options for reading csv , pdf ,word documents as well ? this will give you a lot of views also, we are talking for an actual use of an ai assistant with this abilities !!!
@@codewithbro95 ok. Thank you very much. I'm trying to install this at my place, but I can't manage it myself. I have never programmed, I don't know what c++ is, for example. It is very interesting, but unfortunately I haven't managed to do it yet. Do you perhaps run your own programming community, where I could find support :) ?
The guide to install and make it working was not clearly captured in this video. In between it was only voice and no screen record visible to us. I appreciate your effort, but you need to cover the content for wide audience from beginner to Advance in step by step procedure. The command ''make" still doesn't work. The problem with all these AI youtubers are not providing solution to an issue and keep moving to other AI tools with new content. Try to follow-up and provide solutions to your audience in order to get more followers.
@@contactmebaba I recon the screen went black at a point, sincere apologies for that. That was an editing error. Will try my best to do a better job at double checking before publishing.
@@snatvb definitely agree with you, inferencing with TTS is very bad at the moment, though I recently stumbled on a really promising project called ChatTTS apparently it’s being built specifically for this purpose, I haven’t tried it though, maybe I will and make a video on it.
@@codewithbro95 yep, I've seen recently. I tried "bark" from suno and it work pretty slow (I have rtx 3070) and sometimes it voices llm imagination text instad of I gave :D
If you have any questions please feel free to drop them below!
Please don't forget to like and subscribe for more interesting content like this🔥
hey bro, does it offer italian language?
@@maxxflyer I belive it does, you can check the repo
I am using a combination of "faster-whisper" and "whisper. cpp" offline. I will use "faster-whisper" for fast machines or servers with GPU and whisper in my project.cpp will be on a regular laptop on CPU. Thanks for sharing; your demo was crystal clear; keep it up. New subscriber
My pleasure, glad it helped;)
thank you for the amazing content!
Always a pleasure🎉
Great video bro. Keep it up 👍
Thanks, really appreciate 🙌🏾
i have a problem i dont know why dont detect fine my voice with de base en model, i even scream and no all is detected. i using integrated microphone but is a good microphone. i will test with another microphone soon, but something tellme that not will work neither.
@@Dixon105 should be a problem with your microphone, try your settings and make sure your mic is well setup, that should work. Also I noticed some issues as well with airports, when I have them on it picks up sound not in the best way with them on.
1K Subs. Congrats bro
@@edmondgoddy thanks man, really appreciate the support 🙌🏾🙌🏾
Why is there black screen in middle of the video?
Hey man, apologies for this, that should have been spotted before publishing.
Sorry!
It picks up sounds? weird... Doesn't it phone home?
haha, not yet!
incredible! Do you know if it only transcribes wave files? Can I also get mp3?
not sure but i believe you can convert to wav and transcribe from there!
thanks for you content for West Indies in the carribbeans. Guadeloupe :-) I am curious to know on what kind of machine you are working on ? Is there a big GPU ? I saw metal, normal apple laptop ?
apple silicon m1, with 8 core gpu I think
What different with speech recognition library?? As i know speech recognition support engine like whisper,watsonx,and google speech, but for offline it use vosk by default
This is more accurate in terms of recognition
@@codewithbro95 i will be force to use vosk becouse this dont work for me
Hey umm, can faster whisper detect sounds like that too or is it only Whisper.cpp?
@@Plash14 not sure what you mean
@codewithbro95 basically it can detect your keyboard typing sounds etc right? Was wondering if it can be done on faster_whisper as well
@@Plash14 I see, not so sure about that(haven’t tried it) however, if it’s based off of whisper then I believe it should be able to do that
@@codewithbro95 I see... thanks for the reply!
I need to install c++ to make it work on windows 10 ?
Yeah
@codewithbro95 what is the best source?
How do you get the make command to work on windows?, i got the make command but i just get error saying cc not found and someone said gcc=cc but i dont know how to do anything from there
@@RoarStaze not tried it yet on windows but from the error you got, I believe you have to install gcc on your windows machine
@@codewithbro95 i do have gcc someone said i need to make it gcc=cc but ive no idea how to do that
Hi, noob here.. Trying to figure out how to get the `make` working from VSCode terminal, on windows so far I installed MSYS2 added C:\msys64\usr\bin and C:\msys64\mingw64\bin to PATH env variables but... still says command not recognized..
same did u find a fix?
@@ToMooNoT does it work outside of vscode ? That’s the normal terminal
@@codewithbro95 I had to install Visual Studio and build the C code from there or something, but it didn't build the microphone one, and I don't know how to add it to the build step, so kinda gave up, also was trying to get my AMD GPU to work with ZLUDA which is a library that should make CUDA code work on AyyMD, but no luck there either even with AI helping with troubleshooting..
Your english is excellent. may i make a suggestion - python is not pronounced pie-ton but pie-thon - with the 'th' being the same as the 'th' in 'this'
Appreciate the correction!
Are you serious?
How do I implement this on website??
Please help.
Anybody knows??
There’s a section for that in the repo
can you put this offline whisper with a local llm model lets say phi3 to get reply based on whisper? i mean lets see how fast it can actually put out what the llm model will reply, this way you can make an offline ai assistant with no latency in responses and local 100 %
i am actually working on something like this, check out my recent videos on Jarvis. I am building Jarvis so you don't have to
@@codewithbro95 cool nice job keep it up, can you also add a way to use phi3 llm with phidata as well for Local RAG and also options for reading csv , pdf ,word documents as well ? this will give you a lot of views also, we are talking for an actual use of an ai assistant with this abilities !!!
@@gnosisdg8497 definitely something i am looking to work on, stay tuned!!!
love it !!!
Glad you love it... Please, don't forget to like and subscribe for more interesting content like this one🔥😎
Your screen went black at 6:10
yeah, editing mistake, my appologies
Wow amazing!
Question: how much faster is it estimated working faster then Python?
Thanks a lot!
No specific data on that but after trying both I’d say it’s just about 5x faster in transcription
Guys,
this is just for english language?
It does support other languages
@@codewithbro95 ok. Thank you very much.
I'm trying to install this at my place, but I can't manage it myself. I have never programmed, I don't know what c++ is, for example.
It is very interesting, but unfortunately I haven't managed to do it yet.
Do you perhaps run your own programming community, where I could find support :) ?
amazing! what gpu are you running? or it’s on cpu?
Running on macOS M1 chip with 8 core GPU, I believe whisper.cpp makes use of metal on mac
The guide to install and make it working was not clearly captured in this video. In between it was only voice and no screen record visible to us. I appreciate your effort, but you need to cover the content for wide audience from beginner to Advance in step by step procedure. The command ''make" still doesn't work. The problem with all these AI youtubers are not providing solution to an issue and keep moving to other AI tools with new content. Try to follow-up and provide solutions to your audience in order to get more followers.
@@contactmebaba I recon the screen went black at a point, sincere apologies for that. That was an editing error. Will try my best to do a better job at double checking before publishing.
I wait same speed TTS(text to speech), it would be great to have
Not sure i understand what you mean!
@@codewithbro95 we have option recognize speech to text in realtime, but text to speech is really slow now
@@snatvb definitely agree with you, inferencing with TTS is very bad at the moment, though I recently stumbled on a really promising project called ChatTTS apparently it’s being built specifically for this purpose, I haven’t tried it though, maybe I will and make a video on it.
@@codewithbro95 yep, I've seen recently. I tried "bark" from suno and it work pretty slow (I have rtx 3070) and sometimes it voices llm imagination text instad of I gave :D
Thanks your video goes blank in the middle for a little bit
@@siddharthchadha3930 really? Didn’t notice that. Apologies nonetheless
@@codewithbro95@ 06:13 onwards
Thanks for the info. But the screen is completely blacked out.