A keyboard? How quaint: Logitech and Baidu link arms to make an AI-enabled, voice-transcribing mouse

Swiss computer peripherals manufacturer Logitech has teamed with Chinese AI and internet company Baidu to make a voice-dictation mouse. The Logitech Voice M380 wireless mouse looks and acts like a regular mouse but with a special button to initiate voice dictation. Baidu claimed recognition facilitates content creation at two …

[Steve K]

Really?

Baidu claimed recognition facilitates content creation at two to three times the speed of what one can type.

Really? Whenever I have tried voice recognition, you can indeed get lots of words down, but by the time you have corrected errors and punctuation it takes at least as long as typing it in the first place....

[Steve K]

Re: Really?

Just replying to my own post here...

I wonder if I am missing anything being a native English speaker?

I can imagine that there is something about typing in Chinese (for a given size of keyboard!) that is more cumbersome than English - maybe that's the factor I am ignoring here and that's how they are gathering this time saving metric?

[doublelayer]

Re: Really?

That is possible. The traditional ways to type in Chinese are drawing the character or typing a romanized equivalent and selecting the correct option from a list. Each spoken character is pretty fast. Depending on the user's speed at handwriting, speech recognition could be very helpful if it's effective.

[Allan George Dyer]

Re: Really?

There are a lot more options for Chinese input - both shape-based, phonetically-based, and hybrid systems. One reduces the number of keystrokes per character to 2, for stenographical use, so speed is possible with practice. Conversely, Chinese (of all variants) is a tonal language so speech recognition is a different challenge to many other languages. Not infrequently I've seen people write a character in the air or on their hand to clarify which of several similar-sounding characters they mean, usually for names where context is less informative. Try doing that with your speech input.

Overall, it's worth remembering that this is a marketing claim.

[Erix]

Re: Really?

Perhaps that is where the mouse comes into play?

[Shadow Systems]

Re: Really?

And for anyone using a Screen Reader it's a right pain in the arse.

You have to disable the SR so it doesn't echo the screen while the dictation software is running, otherwise you hear yourself say something, the SR echo it back as it appears on the screen, & then the dictation program regurgitate the echo as it thinks the SR voice was your own.

Disable the SR, dictate your stuff, disable the dictation software, enable the SR, have the SR read back the dictation so you can make the corrections, then repeat the entire cycle for the next bunch of dictation.

Or we could just use a keyboard & get ~10x as much work done in the same amount of time with ~1/10th the migraine.

All of which ignores the fact that, since a blind person can't see to aim a mouse (or any other pointing device), we often have no reason to even own a mouse in the first place. Maybe for those times when we have to "borrow someone's eyes" so they can try to figure out WTF is happening on a monitor we also have no use for, but in such cases we won't be using the dictation function either.

=-j

[Steve K]

Re: Really?

Of course - I didn't even consider your use-case.

1 .Dictate

2. Get screen reader to read it back to you

3. Dictate corrections (a bit like Jeremy Clarkson and that BMW voice activation trial he did waaay back)

4. Get screen reader to read it back to you

5...

Jump out of window after step 17...!

[Dan 55]

I see what you did with the title, El Reg

Scotty was 35 years too early.

[jake]

Re: I see what you did with the title, El Reg

35 years? Really? Remember when so-called "Multimedia PCs" hit the market in the very early 1990s? Each one came with a microphone and speakers (joy!). The resulting cacophony in offices world-wide demonstrated quite nicely why talking (and listening) to one's computer was for niche cases only ...

[Flightmode]

Re: I see what you did with the title, El Reg

Ooh, remember IconDoIt and IconHearIt? Windows add-ons that would allow "UI enhancements", such as fancier icons, making the mouse pointer automatically jump to OK buttons, speech synthesis and so on? They were fun for the first hour, but that wore thin pretty quickly!

I remember filling out some form of aptitude or IQ test or something, a task that I needed to sit and focus on for about an hour anyway on a PC with those programs installed. About forty-five minutes into the test, I was asked to choose synonyms to a list of words; one of which was "replica". As I clicked the radio button next to the word "copy", my plastic but surprisingly loud Creative speakers decided to shout "COPY!!!" at me. I jumped. :-)

[Chris G]

I have enough problems with windows and Android trying to correct my idiomatic English into something like American English, if I started shouting at a mouse, dog knows what the result would be.

[Shadow Systems]

At Chris G, re: English...

At least you're not Scottish trapped in a voice activated elevator... =-)p

*Hands you a pint*

Drink it or use it to smash the daft dictation mouse into pulp, either result is fine by me. =-D

[Chris G]

Re: At Chris G, re: English...

Can't waste good beer, so I'll down the pint and then use the mug to smash the bugger.

As a lad from South London, some decades back I doubt that speech recognition would have made much sense of my accent, I have improved somewhat since then.

[alain williams]

So where does the speech recognition happen ?

It can't be in the mouse.

It could be in the user's PC - some chunk of software needed, probably MS Windows only.

I suspect that it sends the voice up to Baidu's servers to use its Baidu Brain. So this means that everything that you want speech recognised goes to China. They might well give you some text back but they will keep a copy to give to Chinese spooks.

Anyone who does not live their life 100% in the open should aoid this like the plague. It is a security disaster. But I can see idiot managers blindly ordering these things :-(

But then some twats do share stuff on facebook - so maybe it will sell well!

Can you switch the chatting to Baidu servers off ?

[Anonymous Coward]

Re: So where does the speech recognition happen ?

Yes, my first thought was "Does that 'security layer' include notifying Chinese authorities if certain keywords are detected?"

[James O'Shea]

Re: So where does the speech recognition happen ?

A good way to find out would be to mention 'Tiananmen Square'.

You go first, I'll watch.

[stuartnz]

Re: So where does the speech recognition happen ?

I was thinking "Winnie the Pooh" - perhaps "Winnie the Pooh visits Tiananmen Square" would guarantee the police arrive double quick?

[Efer Brick]

?Hello, computer

who doesn't love a gif --- https://giphy.com/gifs/reactionseditor-reaction-3o7btVRbshbbaC8Ygg

[Richard149]

Hello Computer

https://www.youtube.com/watch?v=hShY6xZWVGE

[Anonymous Coward]

Mouse?

Why use a mouse to do this? Mice don't type, keyboards do.

How is this different than using software* that already exists?

*local and/or remote

[jake]

Re: Mouse?

Crossed my mind, too. As it did 30ish years ago .... why have a separate microphone to clutter your desk when you can easily and cheaply add one to a basic keyboard, along with a lockable momentary switch to activate it. So I built my own :-)

The result worked nicely, but I never actually used it for anything.

[Roland6]

Re: Mouse?

>Why use a mouse to do this?

Well if the mouse is in the form of a pointer mouse - as used by presenters, it sort of makes sense to include the mic in the mouse/pointer, but then a lapel or throat mic would give better voice pickup.

I suspect in the case of a typical desktop, putting a mic in the (wireless/USB) mouse would avoid the need for a dedicated mic. But then the majority of webcams include a mic. Which given Logitech are involved can't have passed them by, so I suspect the intent is to use both mic's to enhance the quality of the voice detection and signal processing.

[Allan George Dyer]

Re: Mouse?

Mice and keyboards both have this problem of people's hands clicking on the buttons, which is likely to cause a certain amount of noise on the mic. A lapel mic gets yanked out when the user walks away, forgeting to take it off. Built into the monitor, a mic can be right in front of the user's face, and, if the speakers are there, filtering can eliminate the problem of picking up screen reader output inadvertently.

[pip25]

So... this is a mouse with a microphone?

Because I somehow doubt all that AI magic takes place in the hardware.

[Pete 2]

Link down, brain fried.

> The mouse uses Baidu's AI open platform Baidu Brain speech technology.

Luckily, my keyboard does not require an internet connection to work. I think for the time being, I'll stick with that.

[Paul Herber]

Mr Tibbet has bought my cow.

[jake]

That's odd, because ...

... Mr Tibbet is a hog farmer and has a sow farrowing.

[Paul Herber]

Re: That's odd, because ...

I'll raise you a Dally Llama and Winston La Pooh.

[Shadow Systems]

Re: That's odd, because ...

If I contract with you for delivery of one such beast per 24 hours, each with a different knitted decoration, of a maternal disposition, & with a random psychological issue...

Would that be a daily Dalli doilie llama's mama's trauma drama?

*Runs away before the pun police can catch me*

[Neil Barnes]

I don't have one, so I don't know

but can you tell Alexa (other spies are available) "hey, transcribe the following please and send it to my email"?

That ought to be a doddle, given Google's resources.