I am shopping for an AI solution for my home. Since I totally new to this and assuming others have more experience in this, I want to know what are the advantages of choosing Picroft over Amazon’s Echo Dot?
Cost wise, I think the Dot is a winner. At $50, it makes no sense why should I spend on Raspberry Pi 3, SD card, power supply, casing and a comparable mic that will end up more costly. On top of that, I have to go through the trouble of assembling everything, install the software and troubleshooting the picroft in oppose to plug and play on the Dot.
Skills wise, I think the Dot is also a winner. From the list, I can see thousands of skills are there to install with a click of a button. As for Mycroft, the only place I can find the list of skills is at https://github.com/MycroftAI/mycroft-skills
Hardware wise, the Dot wins hand down. It is available in black or white and I believe the mic is far superior than any USB mic in the market unless you go for those enterprise level speakerphone use for tel-conference which is also costly.
Dependency on third party wise, Amazon is definitely more reputable and stronger than any startups.
The only advantages I can think of for Mycroft are…
- Customization - I can change the wake word and the voice
- Availability - Unlike Echo which is only available officially in US, anyone can build Picroft at anywhere.
Please correct me if I am wrong. I am still sitting on the fence on this. Is it worth to DIY my own AI in long run? Let me know your opinion. Thanks.
open source and privacy concerns make mycroft an obvious choice for me, plus you get to learn something and understand your product to the guts
From my understanding, both products are sending my voice command to some servers in the Internet for processing. How is that going to help you in privacy concern? Please enlighten me.
- mycroft intends to launch openSTT
- and you can use a local Kaldi server already for offline voice recog (haven tried it but push request is in github)
- and there is several options to choose from where the voice is processed, you can always add more as they appear (choose who to trust, revoke trust without changing product)
you can just proxy your mycroft connect to the server , make a new pariign code that didnt ever run on original connection, and you will be using a shared api even on googleSTT, they have your voiceprint sure, but less data to process it since its mixed with lots of user’s requests
and amazon is big, for profit, they WILL try and harvest more data for sure,
making money out of you and listening to your voice requests is a recipe for …
also, do you know when it is recording? is it open source? or does it just promisse to not record on demand by amazon?
mycroft CANT spy on you, you know the code, it listens when you order it only, guaranteed!
privacy by design, not by policy, thats how it helps me in my privacy concerns, hope this answer helped you making a choice
EDIT: i actually want to discuss this subject better, so i created a topic Privacy Concerns and Discussion
I dont think Mycroft vs Echo as a consumer device is a fair comparison at this point in Mycroft’s development life cycle / startup phase
For me in the South Pacfic region here my opinion - a few points well
- Mycroft is open source and can be customized to local regions skills with out waiting for Amazon to open up Alexa skill developers ecosystem
- Can run on commodity RPi hw which is cheap to experiment with and has a strong developer eco-system
- Can use a variety of the Third party APIs for weather , news etc . Eg. Watson News, Weather.io
Thanks for everyone’s input.
I can see the appeal of Mycroft and I always support the open source movement as an alternative to commercial products.
However, I feel that Mycroft still have rooms to grow if it wants to give real competition to the commercial counterparts.
For example, I found out that the delay in responding to my questions in Raspberry Pi 3 is too long to for my patience. Maybe it is hardware related and can be overcome by using more powerful devices.
This is an example of little things that makes a huge difference in user experiences.
So the real question is, is it worth it to sacrifice user experience in expense of privacy?
My biggest issue so far, having built multiple instances of MyCroft on different devices… has been the microphone
I can’t seem to find a good solution to having a compact, high quality microphone. There are large USB “studio” microphones (too big and expensive) small (rubbish) ones and a variety of random ones in between that are not good
For any type of IoT / connected-home use MyCroft will need a microphone compact enough to be mounted on a small device, yet good / powerful enough to work well when someone in the room speaks to it
The Echo has a great microphone and I’m sure it’s not a standard USB mic off eBay.
Anyone have a solution for a good quality, compact microphone that can be used with the Pi? I’m eyeballing my old iPhone at present
I’m using Jabra Speak410. It works pretty good. But the problem is it costs more than an Echo Dot. It really makes no sense to use this cost wise.
I have tried PS3 Eye Camera. The mic works pretty well too. You can consider that.
Anyway, back to the topic.
I got my first Echo Dot a couple of day ago. After playing with Picroft first, I must say I m impress with Alexa’s performance. It is years ahead of Mycroft. The hardware is beautiful and solid, the response is fast and the voice sounds so natural.
I believe Mycroft is heading towards the right direction but until it is fully grown, Alexa will temporary be my household AI. I will continue to tinker with Mycroft as a hobby on my free time and hopefully do my bits to push Mycroft as real competitor to commercial AI such as Alexa and Google Home.
reading up on respeaker https://www.seeedstudio.com/ReSpeaker-4-Mic-Array-for-Raspberry-Pi-p-2941.html
has 4 array mics and programmable LEDs haven’t tried it myself, they also have one in a pi zero factor but once again haven’t tried one myself so have no idea how easy it is to setup or it’s quality.
Better mic, lower price, screen, camera.
You want to support Mycroft because this technology is too important to the future of computing to remain the property of a few huge companies. You also want to support us because we are NOT building a smart speaker. We are building an AI that runs anywhere and interacts exactly like a person ( a strong AI ).
If you want a simple smart speaker and you are happy BEING the product, buy Amazon or Google. If you want to help advance the state of the art in open source AI and support an effort to open this technology to the world, use and support Mycroft.
Mark 2 seems interesting. Is it still based on Raspberry Pi?
We’re looking at that. We want to add some new features to the next version ( array mic with beam forming, FPGA based noise cancellation, touch screen, camera module ) and we’re working on the bill of materials now.
The version shown has a Pi3 in it. It is our second draft. Here is a pic of the first three drafts. The one on the right is the latest. Is is cloth covered ( leaving more options for driver placement ) the plastics on the top leave room for a camera module/rotary encoder.
Is it available for pre-order?
Pre-orders open January 22, 2018.
Until then we are heads down trying to improve the user experience and deliver the skills we promised in our Kickstarter.
Let me try to answer this in 3 steps.
1.mycroft uses offline speech engines like Precise, PocketSphinx.Same is not true for alexa vs google.
2. Command based interactions with mycroft is completely local. Eg create timer, Switch on light etc. Same is not true for Alexa/google/apple
3. For queries which need internet access (book a uber, play music from spotify) no history is maintained, but same is not true for other tech giants…
Checkout this link for how Mycroft ai works
also note, you can run mycroft without backend