RobotChickens wrote:

MousePotato1234 wrote:

(#265)

scootyboy-2010 wrote:

nickeljorn wrote:

No support, this invades users' privacy.

No, it doesn't, because anything said can't be uploaded anywhere because variables only support numbers.

But you can encode a value, store it in a cloud variable, and then subsequently decode it, thus making what is effectively an alphabet-supporting cloud variable in Scratch.

What does speech to text have to do with this? People can do this without speech to text.

What I'm saying is that you could take the results of what the speech to text engine comes up with, encode it, send it over the cloud variable, and decode it on the other end.

thunderclwd wrote:

RobotChickens wrote:

han614698 wrote:

(#257)

RobotChickens wrote:

undeterminstic wrote:

(#245)
no support unless cloud variables are disabled during it

I can't see why this would be an issue! The only output from these blocks would be text. People would in no way have access to the sounds from the microphone. I don't think disabling Cloud Variables would be necessary. I think people are misunderstanding how this extension works.

1. You tell the extension to listen.
2. You say something.
3. That audio is sent to Google's servers for processing just like all other voice-enabled features on many platforms.
4. The output variable gives you the processed text version of what you said.

The only thing this extension can output is text. You can't do any more with this than what you could with a keyboard. No blocks are able to give you access to the un-processed audio recordings… It's only text.

It could possibly be listening when the user isn't aware.

That's what the huge red microphone icon is for. Go check it out on the github!

DangerPuppy10 wrote:

(#259)
No support, this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

This desn't recognize voices and who they're from. This just turns speech into text! Basically the reveres of text to speech.

That's not what I'm supporting.
I'm want the extension to check for the volume of voice. I made a topic similar but not exact to this on and I was told to come here, But I think I'm the only one who wants the extension to check for volume. There's a volume block for in-game noises, why not for Our speech?

Loudness in Sensing.

MousePotato1234 wrote:

(#281)

RobotChickens wrote:

MousePotato1234 wrote:

(#265)

scootyboy-2010 wrote:

nickeljorn wrote:

No support, this invades users' privacy.

No, it doesn't, because anything said can't be uploaded anywhere because variables only support numbers.

But you can encode a value, store it in a cloud variable, and then subsequently decode it, thus making what is effectively an alphabet-supporting cloud variable in Scratch.

What does speech to text have to do with this? People can do this without speech to text.

What I'm saying is that you could take the results of what the speech to text engine comes up with, encode it, send it over the cloud variable, and decode it on the other end.

The only result you can get from the block is plain text. I still don't think there's an issue. See #7721884

no support, because
invades users privacy; hacker can hack in and hear you,
can be used in cloud variables to show voice to public amd
this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

Complete support. It'd be awesome to have this implemented, voice assistants, graphic novels, etc. could benefit from this!

I don't really know if it's been mentioned before, but modern browsers have speech recognition (looking at you, firefox ) (see the SpeechRecognition API on MDN), so no extra work for the ST!

v I saw other posts about recording without the user knowing, but all modern browsers include a popup everytime microphone access is wanted by a site

(turbowarp.org wants: use your microphones, allow, block)

and a giant blinking recording red dot

leog20162 wrote:

invades users privacy; hacker can hack in and hear you,

Not really, some browsers use offline recognition while others use servers, but in general hacking into Google for just getting a 5 second audio isn't worth the effort.

leog20162 wrote:

can be used in cloud variables to show voice to public

Would only return text, not the audio. RobotChickens already mentioned this a thousand times.

leog20162 wrote:

this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

Speech recognition can't distinguish between voices.

Also, didn't a user post these exact concerns some pages ago?

I found this from 2019: https://scratch-mit-edu.ezproxyberklee.flo.org/discuss/post/3554085/ I wonder why they stopped and never added it…

Carpit999 wrote:

There should be a block (or anything else) that starts a voice recognition (on user command, they are prompted) using the microphone. There would also be another block to stop that (on user command, again they are prompted). You could set where the text that was recognized will go. You could also set what certain spoken commands do in your project. That's basically it.

#cmd-c+cmd-v

i made a dupe of this on accident so yeah support lol

leog20162 wrote:

no support, because
invades users privacy; hacker can hack in and hear you,
can be used in cloud variables to show voice to public amd
this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

You wouldn’t be able to detect a specific voice. Also cloud variables could be disabled

ThisIsTemp1 wrote:

(#288)

leog20162 wrote:

no support, because
invades users privacy; hacker can hack in and hear you,
can be used in cloud variables to show voice to public amd
this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

You wouldn’t be able to detect a specific voice. Also cloud variables could be disabled

Disabling cloud variables is not needed. See #7721884

RobotChickens wrote:

MousePotato1234 wrote:

(#281)

RobotChickens wrote:

MousePotato1234 wrote:

(#265)

scootyboy-2010 wrote:

nickeljorn wrote:

No support, this invades users' privacy.

No, it doesn't, because anything said can't be uploaded anywhere because variables only support numbers.

But you can encode a value, store it in a cloud variable, and then subsequently decode it, thus making what is effectively an alphabet-supporting cloud variable in Scratch.

What does speech to text have to do with this? People can do this without speech to text.

What I'm saying is that you could take the results of what the speech to text engine comes up with, encode it, send it over the cloud variable, and decode it on the other end.

The only result you can get from the block is plain text. I still don't think there's an issue. See #7721884

I think what is going on is some people want to allow online text transmission, and some don't. I was just proposing a way to allow such a feature in a Scratch project. I did not intend to express any privacy concerns, as I see no privacy-related issues concerning the transmission of text decoded from sound. The real issue I see here is there is no currently implemented method to always moderate what gets transmitted over Scratch cloud variables.

MousePotato1234 wrote:

(#290)

RobotChickens wrote:

-snippery snake -

The only result you can get from the block is plain text. I still don't think there's an issue. See #7721884

I think what is going on is some people want to allow online text transmission, and some don't. I was just proposing a way to allow such a feature in a Scratch project. I did not intend to express any privacy concerns, as I see no privacy-related issues concerning the transmission of text decoded from sound. The real issue I see here is there is no currently implemented method to always moderate what gets transmitted over Scratch cloud variables.

I agree! Cloud chat projects with user-generated text aren't allowed and should be reported. It doesn't matter if the text is made using a keyboard or extension.

bump

leog20162 wrote:

(#284)
no support, because
invades users privacy; hacker can hack in and hear you,
can be used in cloud variables to show voice to public amd
this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

It's not really going to work like that, because you can imagine a certain voice having like “fingerprint” and a unique ID just for the voice.

There can definitely be some safety precautions while the project is sampling voice.

scienceexplorer1 wrote:

(#293)

leog20162 wrote:

(#284)
no support, because
invades users privacy; hacker can hack in and hear you,
can be used in cloud variables to show voice to public amd
this could hide project features from others. For example, if you have a friend, and you're making a game. And you want that friend to be able to play the game, but NOBODY else. You can use voice recognition to hide it.

It's not really going to work like that, because you can imagine a certain voice having like “fingerprint” and a unique ID just for the voice.

There can definitely be some safety precautions while the project is sampling voice.

That is correct. This extension will have no ability to “fingerprint” voices. It is simply taking in audio from the mic and outputting plain-text.

bump because of doop

8to16 wrote:

(#295)
bump because of doop

speaking of that dupe, we might want to look at this post that was on it

Someone else said that it invades users' privacy, so there should be an icon when voice recognition is being used.

Support

this sounds like a scratch lab kinda thing

EEU1180Nathan wrote:

(#298)
this sounds like a scratch lab kinda thing

It would definitely get added to scratch labs first!

Yea it took a long time for siri to be created. But what if you could record a sound, and if it was simmiler it would detect it? I don't think you would need ai for that