Microsoft will soon let you clone your voice for Teams meetings | Hacker News Reader

Microsoft will soon let you clone your voice for Teams meetings

techcrunch.com

58 pointsthunderbong61 comments

This is actually great.

If we can get only AIs to take meetings we can save a lot of work in offices.

Zoom has the same dream: https://www.theverge.com/2024/6/3/24168733/zoom-ceo-ai-clone...

Eventually we can also decouple scheduling meetings and reading summaries from humans. And then we can all spend time working instead of in meetings.

The robots can archive all their summaries, and have follow up meetings to track status of action points they invented. As long as they do not bother humans it will be a great boon to productivity.

Or, you know, humans could just get their shit together and use asynchronous communication channels in the first place.

Using AI for this is like a trying to solve a self-created problem. If human meeting culture would have just used an issue tracker and e.g. emails for this, this all would not have been necessary.

Human meeting culture exists only because the tools for synchronizing states is so bad in its UX/UI that there is seemingly no better way to do it. If Teams would have had a better UX focussed on issue management combined with better overviews for different roles, they would not have needed AI bots/avatars for this problem.

Although due to Poe's law I'm unsure if GP was earnest, I elected to read it with a /s modifier.

Same here, because otherwise I’m afraid the disease has reached their brain, and they must be destroyed. /s

In all seriousness, I hope that was sarcasm, because otherwise that sounds like absolute hell and the biggest waste of resources since the performance review. Except instead of just wasting time, now it also wastes energy, fresh water, land, and rare earths too.

To add to this, anti-social assholes who kill team jive tend to hide from responsibility with e-mails. Even when it needs to be a meeting.

They need to be dragged out of their comfort zone kicking-and-screaming for public display for the pleasure of others.

In my organization the tools are great but the people don't put the work in to use them correctly. They rely on meetings and emails to get up to date information when the information can be captured on a ticketing system. My organization is very beaurocratic and has an older age group though.

Some of the best architectural work we've done at the company was implemented more like a research paper. Have a central document, write down the current knowledge about an outage, or the current proposal - and improve on that through comments, revisions and discussion.

And then you can still have a meeting or a call if a non-obvious decision needs to be made, or if clarification and explanation of something is needed.

It is gonna be great when everyone sends their AI avatars to these meetings. Just LLM models talking to each other. Would be interesting to see what kind of output such meetings would generate.

Isn't that essentially the stock market trading floor scene?

I think only only about 60% of stock trading is the machines trading with each other.

No one has time to attend meetings. Suddenly you're going to have time to watch LLMs meet...and you imagine yourself doing so with popcorn, I'm assuming?

This is something that should happen in a background process in a few seconds. Therefore eliminating the need to create avatars or clone voices.

Reminds me of the Zizek bit about taking a nice lady home and having stimulating conversation while each others' sex toys are plugged into each other, doing the dirty work.

I'm a bit hesitant to give my voice away to really any company. Is it really that much of a stretch to imagine the ways this could go wrong?

I do prefer a robot standing in instead of me to fulfill the little power trip calling attendance. There is little reason the updates cannot be async in text or via the JIRA tickets I already have to do. A robot could add in pull request data too.

But it doesn't need to be my voice. It in fact should be illegal for Microsoft to use my voice for training without my express consent, and even then they must delete all models and data about it when requested. And it should be illegal for an employer to request or expect it.

There's tech that can do a reasonable approximation using a sample a couple of seconds long.

Any control over that is going to be through social and legal mechanisms, it isn't going to work to try to prevent access to the needed data.

I would trust Microsoft with my voice more than some random startup.

Why? Microsoft is a proven bad actor while the startup might have good people of you're lucky.

There is a nonzero chance that said startup gets bought by Microsoft or worse.

> I would trust Microsoft with my voice more than some random startup.

I bet you haven't read their privacy policy. It starts with: "Your privacy is very important for us." /s

Honestly I would have thought that 5 years ago. But given the latest news on what they are doing, and how they treat security, I honestly have lost all faith that they can handle that data with care.

Laundry list of breaches: https://firewalltimes.com/microsoft-data-breach-timeline/

Profit over customer: https://www.propublica.org/article/microsoft-solarwinds-gold...

This will be an unpopular take as I know tech folks get excited by any new tech but I predict deviants will gain access to this capability and I can imagine the thousands of ways they will wreck someones personal and professional life, make them homeless and destitute. People will need to put entirely new safeguards in places they never imagined they would need and this may even spin off entirely new artificially required industries in the same way that anti-virus and anti-spam came about despite being entirely avoidable but that's a different topic. This is not limited to Microsoft's service. This will happen anywhere people are storing a sufficient sample of peoples voices.

One example mitigation I implemented at a local bank is to have them disable most aspects of internet access to my accounts. They are read-only from the internet and outbound wire transfers are blocked. They now require my physical presence and I make sure all the employees know me personally. I test them from time to time on the phone. My goal is to terminate business relationships with anyone I can not do this with. This will be too much for some people at first, having to treat every business relationship as hostile and I am not even sure I can get 100% completion of my goal

I do not know how to solve this on a global scale. I think the only way things start changing around this vulnerability is when powerful people experience shared pain. Lawmakers, executives, investors, etc... Such as businesses losing billions because some department thought they were talking to a real person and governments losing large amounts of tax revenue when more companies have this capability and it gets abused by deviants. Curious what happens when the fake POTUS calls the real Vladimir and Xi.

Imagine what happens when the company gets hacked and they start selling your voice online?

Sophisticated Phishing attacks, blackmail, destroying someones public image

The AI hype is so strong that they're rolling out a feature to forge people's voices BEFORE they roll out the feature to sign your own voice cryptographically.

People were naive enough to upload their personal details and pictures to Facebook/Tinder/Google over a decade ago without any cryptographic signing.

What makes you think they won't make the same mistake again with their voice? You just have to bait people with features of convenience in return for their data.

There's a reason history keeps repeating itself.

Reminds me of this: https://youtu.be/rYaZ57Bn4pQ

It was a joke video from the onion about outsourcing their own jobs to India. This is kinda the same but with AI.

I know it's only for translation now but how long until "copilot" joins meetings on my behalf?

> I know it's only for translation now but how long until "copilot" joins meetings on my behalf?

Well, they have my voice (Teams), they have my image and biometrics (Windows Hello for business), they know everything i do a the computer (Recall) so they can replace me with an AI.

Just think at the next headlines: Microsoft increases the price of Office 365 bundling a new feature: virtual employees.

>virtual employees

Then we hold virtual meetings where virtual managers create PowerPoint slides with AI and virtual employees "listen" and in the end AI summarizes everything into something no one will ever read? :D

> something no one will ever read?

No, they train the next LLM on it.

While grey goo is about matter, this appears to be a grey energy scenario: one where machines run amok eating all possible energy in an exponential feedback loop of uselessness, driving civilisations to oblivion.

While the civilisations are long-gone, they build up the Kardashev ladder for the sole purpose of producing more of nothing.

Finally, all cosmic resources being expanded, they shut down. Since nobody remained to observe, it is unknown whether in a brief flash of self-awareness, realising the meaninglessness futility of their existence, the machines commited suicide.

AGI? It's either a few days or a thousand year away.

If the meetig consists only of copilots then, I might be fine with it

Real life, but China instead of India: https://www.bbc.com/news/technology-21043693

Only tangentially related, but hustle culture meets remote work: https://www.reddit.com/r/overemployed

This is a great comment. Apart from a phishing tool, this is an outsourcing tool.

Remember that we need more people to learn how to code (the corporate nonsense that many developers here and elsewhere amplified in 2018)!

LOL great video!

Also FYI someone actually decided to lived by that principle and even wrote a NYT bestseller about it: https://en.wikipedia.org/wiki/The_4-Hour_Workweek

Microsoft keeps creating the best tools for phishing.

"My voice is my passport" (quote from some action movie). They claim that they do not store biometric data. For now. Wait until the next rug pull where they'll claim that your voice and all conversations belong to them.

Which raises the question: What happens on all the "free" communication platforms like Google Meet and so on? Are the conversations recorded and mined?

The data collection business Salesforce had an "AI" meeting summary feature already years ago.

The quote is from “sneakers” https://en.m.wikipedia.org/wiki/Sneakers_(1992_film) which is actually a fantastic film.

HSBC (a bank where I have some CD accounts) requires me to shout “my voice is my password” to authenticate whenever I call them.

"...to speak to others in different languages" is conspicuously missing from the title, so much so that it has pushed most of this thread into "why would I give up my own voice" territory

I guess one positive way to look at it is we're one small step closer to the Universal Translator

The biggest use of voice cloning will be callers from foreign countries taking money from grandparents.

Expect encryption like RCS and WhatsApp to go away by popular demand, because when phishing gets too good, the only solution will be a service like Google recording all telephone calls.

How does Google (or whoever) recording all calls solve the problem?

The telephone is becoming like an email inbox without protection from spam. By listening to every telephone call Google can alert users about phishing, voice cloning, etc. in real time.

Voice cloning will get to the point where it's so good that there's no tech-savvy users who can detect it the way they can for email. Users will demand for protection to be added to their calls because you really can't live without it.

Oh, no Phishing is getting too real, let's drop encryption. Said no one ever

Users don't put it in those terms. They'll complain about spam calls that no longer sound like a male voice with heavy foreign accent.

The logical solution will be, we need a service that scans all calls for phishing, just like we have for email. The service cannot be local for the same reason as for email.

> The logical solution will be, we need a service that scans all calls for phishing

Come on, we have been using encryption, one-time codes, recovery codes for over a decade. I don't see that as the logical next step. Just make addresses renewable tokens that prove that you are allowed to contact some underlying number.

tel:+56 1234 5678::[$SEED/TOKEN THAT SUPPORT K_TIMES_USE+EXPIRATION FROM A FIXED CALLER/SENDER]

My phone will double check the token and ignore things when the calling tokens were used, expired, abused somehow and maybe kindly let you know if no human gets to see your call/text/media or maybe reporting the sender for abuse in case it's something that may be prosecuted according to local law.

Want to support calls from unknown numbers? Sure, wire me crypto with some secret and call me using that. If it wasn't worth my time at least I'll enjoy a beer.

All middle management done by AI having meetings with itself and making and presenting slide decks to itself and tracking it all in spreadsheets only it every looks at. The people at the top still get all the money and the people at the bottom still do all the work. But the middle will be so damn efficient we won't know what hit us. Amazing times.

I’d much rather have meetings go full Discord. A chat room is easier to deal with when you’re busy and trying to multitask. Having to drop everything and interact in a rigidly structured audiovisual format feels very constraining and counterproductive these days. A properly trained AI could stand in for you as needed.

Certain managers don’t like this since they want the participants’ full attention, which is why we have Zoom video meetings in the first place.

Who will take accountability if the AI says something offensive or damaging

It’s a good question. I guess the same sort of constraints that prevent A.I. customer service chatbots from cursing or otherwise saying inappropriate stuff.

We are so close to tricking leadership into just writing the email to replace the meeting they choose to have insteaad. We just need to call it “prompting an AI to speak on your behalf during the meeting” + “record and summarize the meeting.”

Is this something that people are clamouring for?

How many of y'all would be using this?

Why even have meetings? Just post a meeting talking points to some magical GenAI driven system and get meeting transcript in seconds. Do not like outcome? Modify initial prompt and try again.

I would rather have the feature to hide window switches which cause differences in face illumination and/or keeping the eyes focused more naturally on the camera.

I wonder if this could help with pronouncing words in other languages. That's always been hard for me

Nice option to have, can't wait until they expand it to more commonly used languages of the world.

Would be very interested in trying an English-to-English mode to wash away my ESL accent.

I think this requires consideration of privacy and security issues.

They were so preoccupied with whether or not they could that they didn't stop to think if they should -- Dr. Ian Malcolm

> clone their voices so they can have their sound-alikes speak to others in meetings in different languages

Yeah, no. Automatic text translation is "ok" but automatic translation direct into a work meeting in your voice? Enjoy your HR meetings and lawsuit for unfair dismissal based on botched translation.

Or the other way around — enjoy swearing and be mean during the calls saying “it’s just a botched translation”.

Reply on news.ycombinator.com