Not real benefit if you have the hardware to run it locally. I made this video for people that runs it on a raspberry pi or similar hardware, that is not powerfull enough to run the STT and TTS on real time.
Yes, but if you are a student, you can just use an academic email, and it will not require a credit card, if not, the credit card is just for identification since we are using the free version, it won't generate any charges to your credit card.
This specific service has a 5 hours monthly free usage tier, so it should be free for as long as you have an account active, check the list of free services on the link on my website
@@Kevin_VF Hi Kevin, the services remain free, but they do ask you to migrate the account to a pay as you go plan, so if in case you exceed the free tier limits, you'll be charged for the extra requests. But, if with your normal usage during the trial you didn't (they would have discounted it from the trial $ 200 they assigned to your account), you'll probably have nothing to worry about. Just be mindful of the free tier limits, so you don't exceed it.
WHAT'S THE POINT?? IT STILL SENDS PRIVATE DATA TO 3RD PARTY SERVER. ? People should stop feeding big corps their data! Yes if you control when audio is send than maybe it's ok, but if it listening all the time -than it's horror movie :D But thanks for video, usful anyway. PS only english is supported :DD what a crap Microsoft..BING sucks does find ANYTHING if it's non USA related the same with languages :D
Yes, you control when the audio is sent, the hot word detection runs locally (the always listening part), so only when it detects the trigget word, does it start recording the audio that is sent to the Azure Service.
The point is that in some countries electricity is expensive and it is not worth having powerful hardware just to recognize speech. A short recording of me saying "Turn on the kitchen light" is really not very private. Moreover, in some languages Whisper only performs well only on the large model, which requires powerful hardware for fast response times. I don't know what it means that only English is supported, because I just did STT in Polish.
Thank you, this was a very helpful tutorial!
Love your videos! keep up the good work! you have filled a really cool home assistant niche
What's the benefit using Azure instead of Home Assistant ?
Not real benefit if you have the hardware to run it locally. I made this video for people that runs it on a raspberry pi or similar hardware, that is not powerfull enough to run the STT and TTS on real time.
For the Azure account you need credit card?
Yes, but if you are a student, you can just use an academic email, and it will not require a credit card, if not, the credit card is just for identification since we are using the free version, it won't generate any charges to your credit card.
How long is this gonna be free?
This specific service has a 5 hours monthly free usage tier, so it should be free for as long as you have an account active, check the list of free services on the link on my website
@@fixtse. Actually, no. You will lose the voice resource once the 30-day period ends.
@@Kevin_VF Hi Kevin, the services remain free, but they do ask you to migrate the account to a pay as you go plan, so if in case you exceed the free tier limits, you'll be charged for the extra requests.
But, if with your normal usage during the trial you didn't (they would have discounted it from the trial $ 200 they assigned to your account), you'll probably have nothing to worry about. Just be mindful of the free tier limits, so you don't exceed it.
WHAT'S THE POINT?? IT STILL SENDS PRIVATE DATA TO 3RD PARTY SERVER. ? People should stop feeding big corps their data! Yes if you control when audio is send than maybe it's ok, but if it listening all the time -than it's horror movie :D But thanks for video, usful anyway.
PS only english is supported :DD what a crap Microsoft..BING sucks does find ANYTHING if it's non USA related the same with languages :D
Yes, you control when the audio is sent, the hot word detection runs locally (the always listening part), so only when it detects the trigget word, does it start recording the audio that is sent to the Azure Service.
The point is that in some countries electricity is expensive and it is not worth having powerful hardware just to recognize speech.
A short recording of me saying "Turn on the kitchen light" is really not very private.
Moreover, in some languages Whisper only performs well only on the large model, which requires powerful hardware for fast response times.
I don't know what it means that only English is supported, because I just did STT in Polish.