I don't know why people buy USB wifi adapter which is capable of packet injection and penetration testing..I did this with my in-built intel Dual Band Wireless AC-3165 in my laptop..I wonder why lot of people suggest to buy alfa or atheros etc.
heyyyyy user you know if you want to troll just once............. you can set your mac address to the same one as your neighbour's device (troll troll troll) then if they are IT they will be like O.O came from .. -checks mac- my PC? zOmgz I gotz a virus
this is a good way to get experience on what a jamming broadcast de-auth will look like, so if it ever happens in a professional setting you can identify it.. well not do much about it though
@user334283 uhm not sure how it works, but if you spoof your mac address to be that of a connected client, there's no way the router knows you're not that client except through packet sequence checks
@CausingUnderflowsEverywhere even if you were not around with your wifi router the radiation if radio towers would still be there..no change of situation
my regret that I signed up stack exchange with google account..I don't really like my username at all
@user334283 yeah but dont you know how transit works? the closer you are the more radiation is penetrating you, since it spams everything in a sphere shape, imagine pointing two lazer beams from a single point at a defined angle, the farther you go the more distance between the beams, the closer you are, you might have both beams shine on you
I'm writing a plugin for TeamSpeak 3.1 in C++11 to help a deaf person: whenever someone says something, it will encode it in FLAC on the fly and submit it to Google's Cloud Speech API to have it recognize the text... and print it out for the user to see... might also implement client-side text to speech, where the user can type something into chat and a TTS engine renders it to audio and "mics up" to say it
going to make it open source... my code is SO GOOD... how did I get this motivated (and how did I get this good at C++?)
2
I'm using Qt 5.6, libgrpc++, googleapis, QtAv, protocol buffers, the Teamspeak 3 Plugin SDK, all in unison
it's amazing... this is so much fun
and the speech recognition will iteratively improve over time as Google improves their training dataset to accommodate for noise, encoding loss, accents, etc.
@DavidPostill There's a tradeoff between "real-timeness", accuracy, and cost, unfortunately. Each time you call the Speech API, it rounds up your audio time to the nearest second, and charges you based on the number of seconds of audio you submit. Also, if you use their "Streaming" API, it can actually go back and correct the audio it previously recognized to be more precise, like (for example) Siri does. But that introduces complexity into the UI.
I have chosen the worst approach for latency but the best approach for cost and accuracy, by waiting until the user stops "keying up" their mic, then buffering all of that audio and submitting it.
and when they "key down" (stop talking) I wait 2 seconds to make sure they have nothing else to add on, because some people will press, say a few words, let go, think for 0.25 sec, then start talking again
I'll eat the cost at first. I'm expecting that the cost might go down or I might be able to convince Google to give me reduced fees or no fees if I explain the good intentions of my usage, but that's a gamble
I'm going to make the code fully open source, but I'm only giving out the API key to my account that lets the user charge me for it, to this one, specific user
might give it out to some other users who have hearing so they can evaluate / look at the software without having to create their own API key, but will advise them to uninstall after they try it for a few minutes
the folks I give it to will definitely be respectful of my requests to limit the usage
and I can set payment maximums on my Google account to prevent an accidental $1000 bill
I'll probably set the maximum to $50/month but will hope for it to be lower
also, there are literally no good client-side software libraries that recognize speech... speech to text is a much harder problem than text to speech
CMU Sphinx has a very good engine, but no good dataset... the value of Google Speech API is that they have an enormous, continuously improving speech dataset... this is a problem that is dramatically improved by big data and I just don't have the resources to attempt that myself
so yeah, I'm basically stuck having to call a cloud service, because the only speech recognition solutions that are actually pretty accurate are ones that have an enormous dataset to draw on, and building and pruning that dataset takes corporate resources, not individual resources
(I believe they use "OK Google" stuff as feedback to help train their dataset... neural networks etc)
I'm not worried about the cost at this point -- it's low enough that I think there are many different viable ways to manage the cost, either with donations or working something out with Google
and this will REALLY REALLY make this user happy, because she says she'd love to participate in our teamspeak, and feels left out all the time because she doesn't hear what's said in TS, she only sees what we type in-game
just got another idea... sigh it's going to add some more coding work, but it'll save quite a lot of bandwidth/money
@DavidPostill if I analyze the audio coming out of Teamspeak and have it aggressively remove silence (and set a fairly high gate for what's considered "silence" since people have background noise), it'll be a big savings
@DavidPostill Thanks... I'll be sure to link to the code in here once I get to the index.html stage (that's a reference to Apache's "It Works!" index.html that displays when you get your web server configured correctly)
currently learning QtAV's mostly undocumented API by reading the source code
At least I kinda thought that was the case from the time I've spent learning JS.
...just ordered two USB 3.0 Micro-B cables from Amazon.
I doubt the machine is compromised. The site is probably more phishing than directly malicious.
Scans are coming up clean so far.
I also have high confidence in Norton.
The domain is extremely obscure but seems to have been around for a few months already. If it was doing something actively malicious, chances are good it would have been stopped by the machine's security.
I've restarted the browser after flushing its cache.
Speaking of, I think I have two SSDs in my bag I forgot about
@allquicatic Hah, nice project. I might point out that's not the worst case for latency and best for cost. The worst case for latency would be to record all the speech, concatenate it and upload it to a Youtube video, and wait for its closed captioning to do its work :-P
@allquicatic Really? I'd thought there've been plenty for several decades.
For some reason, my laptop's USB 3.0 ports tend to drop connections during large, high-speed transfers. Is there an explanation for this? This is not specific to the device: it happens with any USB 3.0 device capable of moving more than 100 MB/s.
Unlike Google, which "learns" to a generic, best-fit model, Dragon (heck even IBM Via Voice) is calibrated to each specific user's speaking habits, accents, speeds, etc.
And I think even Google's speech recognition engine has some, if not total offline capability