Uh, did UA-cam eat my comment? Oh well. What I was saying is that only providing visual captchas is a bad idea. I use ffmpeg and a TTS to create an mp4 file that has the captcha in it and a voice reading the word aloud. Been doing a great job in my Discord server to keep the bots out. Blocking access to visually impaired users may even be illegal, depending on your area.
@@forgotabhi Not much to say, really. I used gtts and a huge list of animal names (I limited the list to animals that it can pronounce, and saved them all to mp3 files. Then I use a captcha program to generate images for them, ffmpeg to combine the two into a ~3 second mp4, then I upload that to Discord. The user is asked to type the name of the animal and, if they get it wrong three times or take too long, it resets and gives them a new word.
yeah, relying solely on visual captchas can be exclusionary. Your approach of using TTS to create an mp4 file with a spoken captcha is a creative solution that helps include visually impaired users. It's important to ensure accessibility for all users. Well done.
I agree that usual visual captcha can block access to visually impaired users, but there are also many users, who can't listen to audio. I'd suggest that image should be paired with audio telling what's on image, thus providing accessibility to more various people.
im guessing you can generate thousands or tens of thousands of images and their original prompts, then give that to an image recognition AI and get some kinda good result
Typically, you'd store something to identify the user's session (usually a cookie value you've sent them) along with the word required into a local database (SQLite, for example, or as I prefer PostgreSQL), then check them afterwards.
You can verify a captcha by comparing the user-entered text with the generated captcha text. If they match, the captcha is verified. If they don't match, the captcha is not verified. It's a simple text comparison process.
Uh, did UA-cam eat my comment?
Oh well. What I was saying is that only providing visual captchas is a bad idea. I use ffmpeg and a TTS to create an mp4 file that has the captcha in it and a voice reading the word aloud. Been doing a great job in my Discord server to keep the bots out.
Blocking access to visually impaired users may even be illegal, depending on your area.
very cool! can you tell me more about it?
@@forgotabhi Not much to say, really. I used gtts and a huge list of animal names (I limited the list to animals that it can pronounce, and saved them all to mp3 files.
Then I use a captcha program to generate images for them, ffmpeg to combine the two into a ~3 second mp4, then I upload that to Discord.
The user is asked to type the name of the animal and, if they get it wrong three times or take too long, it resets and gives them a new word.
yeah, relying solely on visual captchas can be exclusionary. Your approach of using TTS to create an mp4 file with a spoken captcha is a creative solution that helps include visually impaired users. It's important to ensure accessibility for all users. Well done.
I agree that usual visual captcha can block access to visually impaired users, but there are also many users, who can't listen to audio. I'd suggest that image should be paired with audio telling what's on image, thus providing accessibility to more various people.
All the peripheral python 3rd party packages never cease to amaze me.
Great tutorial 👌
Nice one!
How about the other way around. Decide the generated captcha. Ihis may require AI to warn the decode the various letters...
im guessing you can generate thousands or tens of thousands of images and their original prompts, then give that to an image recognition AI and get some kinda good result
Thank you very much Federico
Thanks for your efforts:
I obtain this Exception : "cannot open resource" when execute write function, Can you help me to solve it?
I got the same error, i fixed it by not using the font argument
How i can verify Captcha?
For example user enter respective text, that text match with Captcha or not?
Typically, you'd store something to identify the user's session (usually a cookie value you've sent them) along with the word required into a local database (SQLite, for example, or as I prefer PostgreSQL), then check them afterwards.
You can verify a captcha by comparing the user-entered text with the generated captcha text. If they match, the captcha is verified. If they don't match, the captcha is not verified. It's a simple text comparison process.
So how to match captcha make video