Tired of typing in Slack? Use Webspeech recognition API to chat hands-free | Hacker News Reader

Tired of typing in Slack? Use Webspeech recognition API to chat hands-free

github.com

36 pointscrackleware32 comments

By the time I'm speaking to my computer so it can TTS into Slack so I can converse with my colleague sitting at the next desk I might as well rotate my head 90°...

(EDIT: I should also add that at any point in time my colleagues may be spread over Massachusetts, New York, California, Oxfordshire, London, Britanny, Victoria, Lower Saxony and that's when they're not travelling. But we also use it for conversations when we're in the same office.)

Now they just need a voice synthesizer on the other end to convert the text back into speech to the person doesn't have to look at it, but can just listen to it instead. Man, that would be convenient!

Yeah caus everyone (outside of America at least) loves speaking out loud to inanimate objects.

Why should this be specific to Slack? Can't this just use OS-level speech-to-text functionality?

(Just like other input devices, e.g. keyboard and mouse)

Because the JS targets Slack HTML to insert the recognized text from the webkitSpeechRecognition API.

Something something open plan something something noise. Neat hack, though!

I'm not tired of typing in Slack, I'm tired of the fact that most of my co-workers use it, which in turn forces me to type and waste time in it.

I don't understand the appeal of slack over IRC.

Its laziness for me. But, this comment has prompted me to get some info on setting up a good local IRC server.

Our team switched to Slack a few weeks ago (coming from IRC), and the most frequent reasons for liking Slack that I've overheard are:

- looks more polished

- /giphy is hilarious

There's also this: http://cultofthepartyparrot.com/

It's far more polished, the integrations are endless and easy to use, and setting it up for non-technical people is a breeze. Along with its easy availability on mobile.

Damn, do we have to go over this in every single thread about Slack, or instant messaging?

embedded media, searchable archive, ready-made webhook integration, synced history across machines, proper support for multiple logged-in clients (and lately, I'm really impressed with how well the voice chat works compared to, say, skype).

Yes. There's a solution for all of this with IRC, but some of it requires specialised clients, servers running somewhere and some of it requires you to write additional code.

We're in the business of writing code for our core product, not for our internal chat server.

I'm not entirely opposed to writing code for our infrastructure that's not completely product-related but if I can have something that works out of the box for a reasonable amount of money, then I will go that way.

OTOH, when the problem concerns some dependency of the product in any way, then I'm totally willing to go much greater lengths of doing the work on our end than most of other people.

Because every single piece of the puzzle you control fully is a piece you won't have to find a replacement for in a hurry sometime in the future.

But company-internal chat? If Slack breaks (or gets too expensive), there are ton of alternatives all of which would be more or less equal replacements.

IRC clients already exist that can do most of these things, like https://thelounge.github.io

/Shameless plug

It looks like I have to run and administer my own server, and I don't see any support for uploading media or searching previous conversations.

The demo is pretty, though.

Uploading media is best done by using something like ShareX pointed at an internal server though so that you don't send important information outside the company.

The number of times I see people write root passwords or financial info in Slack is scary.

The Lounge is nice (I use it for IRC), but it has no persistent history, no search, or any of the other features a company would need.

IRC works for a devops shop, but it's unsuitable otherwise.

As someone who almost lived on IRC in the 90s and now employs several people who use Slack almost all day, my take:

- it's an integrated system with a reliable client on iOS, Android, Mac, and PC that all work somewhat similarly and support all of the features

- dragging and dropping media just works

- the way URLs are expanded is nice, I've not seen any IRC client do this as well as Slack

- 'reactions' on chat items

- the search is great and it's integrated with every client on all those aforementioned platforms

- you can set quite complex sets of notifications for specific channels or conversations.. such as don't mobile notify me between certain hours, but then never mobile notify me for our #random junk channel.. stuff like that. Employees are very thankful for this feature.

- custom emojis, we use these more than you'd think.

- having our backend systems ping messages onto Slack is very easy, basically a single liner - I'd probably have to write some sort of HTTP to IRC bridge to replicate this.

- syncing of read/unread messages across clients

It all adds up.

Forget about whether IRC can do everything that Slack does. I can give you number of reasons (many already have in this thread and others) but it does not matter. Millions have decided that it is better for them. We can either decide they are all dumb or choose to throughly understand why they choose it.

I think that understanding why people prefer Slack over IRC is an excellent exercise for developers. I am highly doubtful that one can build software for non-developers without understanding this.

I'm also tried of reading in Slack. Please add TTS too.

A product with these two features already exists under the name of skype ;)

Technically, you'd just need this STT solution to write a log somewhere (pastebin?)

And before Skype, actual phone calls! :)

text-to-speech (WIP),

https://github.com/crackleware/slack-webspeech/commit/d63564...

I've been wondering how long it's going to be before we have enough computrons in our phones to do real-time full text voice recognition, with the aim of sending over a voice model to the other side (or not) and subsequently sending only text. You get huge compression for long conversations, making phone calls possible under stupidly bad bandwidth and packet loss constraints.

I wonder what modern AI could do with vocoded signals. That's super low bandwidth (and sounds it) but I'm sure some modern software could do a good reconstruction.

I use Slack because I'm tired of speaking.

I packaged it as Chrome extension:

https://github.com/crackleware/slack-webspeech/commit/afae06...

Featuring TTS (optional).

Nice! Better solution would be a chrome plugin

Definitively!

Reply on news.ycombinator.com