Unfortunately, Mycroft is not very open itself. Sure, most of the code is open and available, but I tried to contribute and found my PRs ignored for weeks. When they were finally ready to merge them, their poor response cause me to lose interest in the project. At that time, they did not seem interested in cultivating a strong developer community around their core technology components; they were doing their thing, and they wanted the community to implement “skills”. I got the impression that community could either get on board or stand aside and watch them work. For that reason alone, I feel fairly certain that this project will fail eventually as well, and their hardware will become yet another high-tech relic of a paperweight.
As a formerly enthusiastic kickstarter backer, I cannot recommend the Mycroft project as the basis for a product; you don’t own and can’t control the platform on any meaningful way (short of forking it). It might be a better choice than a closed platform, but not enough to make me want to put any money in it.
Most people only use the voice assistants for a few simple tasks, which is perfect for an open source project like mycroft. It is, however, very, very bad for Amazon and Google, because those tasks don't make them money. That's why they're all going so aggressive on "you asked for the time, but by the way here's a 5 minute speech on all the easily monetizable tasks I can do instead"
People like the idea of voice assistants, but by and large they don't like all the problems associated with a voice assistant run by Amazon, Google, and Microsoft.
I've actually pre-ordered the Mycroft Mark 2, although no chance to evaluate it yet.
I'm very interested in devices that can do this locally.
I'm not interested in Alexa/Google home anymore AT ALL - I've gone that route, they both work, but they want my dollars all the time, and it's become increasingly clear that if they can't get me making purchases through those devices - they will kill them off, or become ever more scummy in the attempt (Alexa is now including ads in the "did you know" section - "did you know" was already a fucking terrible decision to include, since it's going to marginally increase interaction at the expense of huge user dissatisfaction. But putting ads there has made me leave.)
So basically - I think if anything, we're seeing a speed run of 90s/2000s tech company boom/bust. A huge amount of money poured into the space with no real idea of how to sustainably profit, but the space itself doesn't feel like it's going anywhere.
It's really, really compelling to allow voice control in all sorts of interactions - but it needs to be very clearly working with me, and not trying to subvert my intent for profit. That might even mean it needs to fall back to something like "if this command, then that action" style usage. No more changing commands, no more bullshit ads, no more subversion of what I'm asking it to do.
It needs to obey me, not google or amazon. Otherwise it's a sales rep and not a digital assistant.
I got an echo (alexa) for free and use it for home assistant. It only works when I have an internet connection. So when my internet is out, I cannot turn my lights on/off with it. I understand why, but i too would REALLY like to just have all functionality dependencies for home automation to be local.
While an improvement I don't want a cloud-dependent system except where it's unavoidable: like having it read me news.
Can a raspberry pi handle the server functionality I wonder?
I personally don't think there is enough cybernetica to control with voice. At some point there may be, but right now, the internet is just one giant consumption stream with a few searches and purchases now and then.
That digital daemon experience taught me that I care more about physical intelligence than verbal intelligence when it comes to my technology. I'm verbally intelligent myself, I don't need an AI who can't even speak correctly, let alone understand me, be my verbal interface to the world.
There is no way that a device can meaningfully parse information from the internet and present it to you right now. At some point? Sure maybe. But it's definitely not now.
What I want, and what I was pretty clear about in my list of use cases is the ability to push a button (or run a function with parameters) with my voice.
My exact use cases
- Kitchen timer: StartTimerFor(duration, timerName?), HowMuchTimeOnTimer(timerName)
- Alarms: SetAlarmFor(time, alarmName?), WhatAlarmsAreSet(), CancelAlarm(alarmName | time)
- Lights/devices(ex: tvs, ACs, etc): TurnOff(DeviceName | RoomName | ALL), TurnOn(DeviceName | RoomName | ALL)
- Conversions: WhatIs(numberOfUnits, inSecondUnit)
- Current time: WhatTimeIsIt(inlocation?)
- Current weather (optional, I don't use this a ton and it requires an upstream source, although HomeAssistant already gets that info for me): WhatWeatherIsIt(atTime?)
---
I've found basically all of the voice assistants I interact with are actually really damn good about understanding roughly those patterns (Alexa was the best, but only for the first year or so [honestly - it was actually wonderful as a beta product] - it's gone markedly downhill over the last several years as they try to cram in more detection and more features).
They just insist on trying to sell me on other parts of the experience (check out this tv show, there's a sale on, use this product, did you know this? did you know that? etc). And I don't want it.
But I'm more than happy to pay a fair sum of money for a thing that will just reliably do those commands.
It's incredibly liberating to be able to do those things with my hands busy, or my eyes closed, or while lying down.
Tack on an HDMI port or a display so I can view a recipe and I will literally give you money right now for this thing (and I have - since I've tried most commercial voice assistants).
Long term - I'll probably end up just cobbling together my own version using Rhasspy/Mycroft or another text parser/STT engine, and HomeAssistant or OpenHab if I can't get what I want commercially. I just know that I'll end up spending more money and time on it that way (which is not the end of the world, I just have other hobbies at the moment, and a very young child).
This is a word-for-word description of how Siri originally functioned. "You asked for the top 5 romantic resturaunts nearby; here are the top results from Google Search:"
I'll certainly grant that... but the price point where Mycroft is, is certainly not near what I'd pay for doing those few simple tasks.
Apple is at the upper end of what I'd spend for such a device (the HomePod mini is $99) - and that's because I'm fairly invested into the Apple ecosystem and thus it can make use of the iTunes library, home automation, calendar items, etc...
If I wasn't invested in Apple, then none of the home assistants other than Amazon (because of the price point for the echo) would be particularly interesting.
I've got a echo show - because its a very nice simple clock/weather interface (that's got Alexa behind it) too (I really liked the Ambient 7 day weather clock when it was available). I've got an echo wall clock that is paired with the echo in the kitchen - it makes timers nicely visible (a sibling of mine has an echo wall clock because its an analog dial that doesn't have any sound with it).
The problems with Alexa of suggesting by the way ("Alexa, stop by the way" - give it a try and yes, it is routineable) are tolerable for how much I'm paying for them and the functionality that I use it for.
So cold comfort since it’s annoying as hell, but it slowly learns you don’t like it and will back off its frequency. Amazon unsurprisingly tracks “dissatisfaction” responses and adapts rate of things (globally and individually) so you do actually have to cuss out Alexa to change it. It’s slow because obviously it’s profitable but it does happen.
On top of that, the Echo products are just not well built, not well thought out, and have NOT been upgraded to make them better: They update, but with NO visible benefit to the owner. One example: The Echo Show 8 Cannot and will not keep its display off all night, even if you explicitly command "display off" before going to bed (yes, it does understand and temporarily obey this command!) But sometime during the night, something will wake it up, and the damn thing turns into a lighthouse in your bedroom, waking one of us up.
I'd really like to find Alexa more useful, but like most folks I know with one, it's mostly just useful as a glorified voice-controlled radio - I'd use it more to control lights and such, if I could get the damn thing to actually realize waht lights are in what rooms, and that dimmer switches and smart lights can indeed share a location that should be controlled together. (Yes, this is supposed to work, but it doesn't...)
I would pay $500 to outfit the house with a central voice recognition processor that would be capable of supporting a dozen or so very secure listeners on the local LAN. Mycroft isn't that solution.
I'd contend that it's absolutely not a solution in search of a problem; it's much more of an unsolved problem, and a big part of the "why" is
- voice recognition/assistance tech still maturing
- major players are insisting that the tech supports their walled gardens
- price points are still a problem
The last two creates a conundrum: a lot of times tech prices come down by selling expensive stuff to rich people until the hardware becomes commoditized. But for a good voice assistant, you need a lot of up-front investment at scale. Unfortunately, the companies that are able to do this are also controlling the hardware that can use it, which limits its ability to spread and be useful.
This is why I think Mycroft is important to support:
1. If you can make voice assistant software open-source and plug-and-play, then it frees people up to tinker with form factors
2. Part of Mycroft's pitch to businesses is that they can make custom solutions. There are probably a thousand big businesses that might want to get into this space but don't want to rely on Amazon because they want to control the experience and not give up their data. Maybe Target wants to stick virtual assistants around their stores, or maybe a hospital wants to give tools for surgeons.
I also think there's an opportunity for voice control in home stereo, where someone decouples the speakers from everything else. It's still annoying to work with Bluetooth in 2022, and Sonos is still pricey, and another walled garden. I'd love to have a simple controller that connects a dumb speaker to Wi-Fi and lets me voice-control it to play music from a library of my choosing. That's not a thing yet, right?
Especially interesting as voice recognition is much easier, cheaper and more efficient within the limited space of a specific usage.
Really? An Echo Dot is $25 on Amazon right now. Which, if you use it at all, is pretty reasonable. (To be sure, if I were using it for music to any degree, I'd probably get a model with better speakers.)
For music, I have an old phone connected to a stereo receiver. So it has voice control although I mostly pick a playlist or album manually.
Yup, price point is solved for the "just timers and some music" people, but now you're stuck in Amazon's orbit.
https://epic.org/wp-content/uploads/privacy/internet/ftc/EPI...
EDIT: other point I forgot - original article says they're bleeding money, so clearly $25 isn't sustainable right now anyways.
For what it's worth, Google Assistant does have an open API to create new devices. It's not open source, but you can certainly experiment with your own custom form factors. There's even a tutorial:
https://medium.com/google-cloud/how-to-build-your-own-smart-...
Maybe you meant it like, "the voice assistant space isn't going to generate huge profits, and thus giant corporations will lose interest".
But even that is absurd. They will still have to do it as a loss leader. Maybe not Amazon — because they just ship us our toilet paper and protein bars and shit. They don't have an "ecosystem" (although they gave it a halfhearted try a few times).
But the chance that in 2032 people just like... don't have voice assistants? It's literally zero, barring an actual WWIII cataclysm reversion-to-barbarism event.
> doesn't work as well as people were hoping
Nothing does, until it does...
> little monetization
Yep, that might be right. But it doesn't necessarily mean the space is "dying". Just that it might not be amenable to oligopolization.
I think the story that you read simply says that it’s hard to monetize. You are inferring more than what the story says.
Voice assistants are here to stay.
I eagerly await the day when I can simply say respond to this post then begin writing with my voice.
- controlling smart devices (thermostats, TVs, speakers)
- broadcasting messages
- reminders / tasks
- asking questions
However, none of these use cases generate any revenue for Google afaik.
Those companies are failing to profit because they don't understand that a digital assistant needs to be working with me, locally, and not subverting my intent.
It just needs to be my device, and not a sales rep for google/amazon. I use voice controls all the time at home - it's astoundingly useful in all sorts of situations, and I'm not even disabled (where it's literally life changing in some cases).
Apple isn't necessarily in the same boat. Siri isn't particularly good, but it does all those things well. Most importantly, it keeps people on iPhones and in the Apple ecosystem, which does make money.
You already have the phone and probably have a device that works with HomeKit so why not try it out. Next, you buy some new lights. Before you know it, you're controlling most of the lights in your house, streaming Apple Music and setting kitchen timers from your Apple Watch. Next time you need a new phone, you're not even going to think about anything else because if you change you won't be able to turn on your lights anymore.
Apple has a plan that works and Amazon doesn't.
Amazon is selling the devices at or below cost and hoping that Alexa will make the money (which it doesn’t)
Look at all those Korean novelas and shovelware on Netflix. There are cohorts of subscribers who remain highly loyal because they're into it. Netflix isn't necessarily swinging for the fences with high brow, popular content that competes with the best studios in the world. Instead, they pump out a wide variety of content that keeps the maximum number of people subscribed.
Apple is similar. They promote features - whether it's Siri, health, privacy, family sharing/controls, etc. - that will strongly appeal to some cohort and keep them on the platform. Then, they incrementally hook you into services until you're buying $1000 devices for the whole family and paying $30/mo for the services bundle. And once you're there, they have you because the switching cost involves turning your digital life upside down.
I'm sure enough people find some small use for the voice commands that it's a good feature to have on the phones.
Apple isn't trying to make money with Siri. It's using Siri to make its ecosystem of Apple Music and similar more valuable to its customers.
The limits that Apple puts on what it can do makes that cloud side computation less expensive.
---
Consider that bit - less expensive. Apple doesn't run its own cloud in the way that Google, Amazon, or Microsoft do. So what does Alexa cost? It costs for AWS cloud time. That's the expense that it's running. Those skills that people use run on AWS compute time rather than a phone's local cpu and battery.
A google search "costs" about 1 KJ of energy. Alexa has similar costs somewhere just for energy and other costs for the maintenance of the additional software and content. It costs something to maintain that joke database.
I can't even say "Turn on the lights and the fan". I have to present it all in bite-sized chunks. Come on how hard is that.
I have a feeling there is no AI in these supposedly intelligent assistants. Only scripted stuff.
It's not "Siri is smarter" but rather "Apple is working on minimizing cloud costs because it is entirely a cost for them."
Siri is scripted and limited to make it locally "smarter".
With Amazon and Alexa, everything goes to AWS because that's where the entirety of Alexa's processing happens. The hardware devices in the homes are "dumb" terminals for AWS with a voice interface. This allows Amazon to make use of AWS as much as it can and do something with "surplus" computing power that it has available on AWS.
https://www.amazon.science/blog/on-device-speech-processing-...
This is also part of the work that Apple has done (and is likely part of the making Siri cost less for cloud compute).
https://www.macrumors.com/how-to/use-siri-offline-ios/
> In iOS 15, Apple moved all Siri speech processing and personalization onto your device, making the virtual assistant more secure and faster at processing requests. This also means Siri can now handle a range of requests entirely offline.
> Once you're using iOS 15, you don't need to enable anything for Siri to work offline. The types of requests that it can handle without phoning home to Apple's servers include the following:
Create and disable timers and alarms.
Launch apps.
Control Apple Music and Podcasts audio playback.
Control system settings including accessibility features, volume, Low Power mode, Airplane mode, and so on.i just wanted to share my 2 cents, as it's not unique to Apple and the cloud can be costly even if you own it. Big tech has been investing in on-device for a while. besides voice commands, apple and google do transcription locally too. because now you can have local speech to text with cloud level accuracy and of all the reasons you shared - cost, privacy, latency etc. (but again, i'm biased)
https://www.androidauthority.com/voice-typing-opinion-322134... https://techcrunch.com/2022/05/17/apple-adds-live-captions-t...
When I give a double command Siri literally tells me she can't do two things at the same time and I have to present them as two separate commands.
Perhaps it's because I have it set to UK English? Perhaps it's smarter in US English. I'll have to try that.
Most companies treat you know English as European, and European customers as second in line for new features after America.
Or more specifically, treat American English first for voice control stuff.
“What’s it like outside right now?” => GetWeatherIntent arguments=[time=current location=device_location]
There’s no Omni intelligent being making API calls anywhere. It’s NLU with a decision tree(s) based on the intent and arguments
I'm saying, new business idea: give up on the old strategy of Alexa driving induced profit in other business units, instead charge more than the breakeven cost of building devices and running the service. This will likely be substantially more, and fewer Alexas will be sold. But the users that do actually get a lot of value from the device and service they are using will pay more for it.
The product/market fit to test is: How many customers would pay more for a device that's not trying to sell you things? Can you get a solid (but smaller) business just by charging more for the device? What features would you add to persuade the marginal user to pay more? Offline mode for privacy-conscious users? Lean in to home automation features? What about a true AI-powered personal assistant? What about per-user language training with a local model that gets refined by your voice samples, with that data not shared back to the cloud? Etc.
Simple startup-style product iteration stuff here. You had a customer hypothesis and a growth model, and it was proven unviable. So can you pivot to find a viable business?
Also importantly, Siri mostly runs off your iPhones processor, and apple doesn’t have to pay a big cloud bill for it, unlike Alexa.
That set of functionality alone makes them well worth the money for us.
Theres a lot of things its really not good for and people have tried them all I'm sure. But where it fails the hardest is being able to increase sales volume for Amazon or increase ad revenue for Google - the only path to monetization seems to be to force it in - and THAT is what is dying.
And who the heck wants it in a ring or glasses??
On the other hand, Mycrodt sounds like something people would actually want to use provided that it can operate locally and doesn't send any data outside the home.
Some of the comments below are part of the explanation why. It doesn't work as well as people were hoping, and it's a solution in search of a problem with limited application and it seems little monetization. The above article sums it up better from the big tech company's perspective.