Pretty sad developement for us, the users, I suppose:
Pretty sad developement for us, the users, I suppose:
Will Snips continue to be developed as an open source project following this acquisition?
This is very sad indeed! It is hard to believe that the new owner will pursue the open software model, but rather will make it less interoperable and more closed.
My experience with SONOS as a company has been very disappointing as they do not really share the vision of a modular, customizable and open product.
Sad to see snips going down that road…
EDIT: But where there is shadow, there is also light
I would REALLY like to see a statement from SONOS regarding what they plan to do. Seems kind of counterproductive for a commercial company to buy up a “privacy” assistant. But, maybe they will build it in to some of their systems that will allow integration in to our own setups and what not…
but… im wishful thinking
What Ben said: what does this mean for the direction of Snips? Will it continue to be available to makers? Will the promise to make Snips open source eventually be upheld?
I write you as helpful leader in this community!
I think the community is deeply unsettled as to how the commitment to snips will continue. Could you please comment on whether the many hours we have worked with snips continue to be appreciated? Or will they be worthless in the future?
I think the community has contributed its part (beta testing, feature requests, skill development) to Sonos buying Snips now.
It would only be fair, open and honest to talk to the community about how things will go in the future.
Thanks in advance!
Picovoice has just released a similar tool to snips for offline NLU (console.picovoice.ai). I have been using it for my own project and it has much better accuracy than snips and is super lightweight! The only downside is that their free tier has limitations on context size and vocabulary.
Snips has been quiet in here and with releases lately… which isn’t a new thing, but support for their own products has always been more focused on the community helping the community.
Snips marketed themselves from the start as an
open source project, yet bugger all is open source at all, and what is took forever to be released to the public.
Snips has not even bothered making a huge announcement of the sale… not here, not twitter… just a tiny banner at the top of their main webpage that they “joined forces” (really… selling the company is joining forces…hmm)
Snips Air… will that even be a thing now?
Will see how long before the console page just stops working, and we the
BETA testers are left with a product that can’t be used, or one that Sonos will have Makers pay $$ to continue using.
The fact that the open source Snips never happened over 5 years, will never happen. Time yet again to find a new product. I have already been playing with Picovoice and the wake word works far better than snips hotword engine in regard to false positives, but there still is nothing yet out there for ASR offline the works as good as Google etc
This also has me very worried as long as there is no statement on how the maker part is going to be supported or not in the future.
Apart from the fact that a lot of people here including me have invested a lot of time on building projects with snips there is also no real local voice platform alternative for us German speakers right now.
I hope the people of Snips or Sonos will make their plans for the community public soon.
This was exactly the fear I had when I stopped developing for the Snips ecosystem half a year ago.
Have a look at Rhasspy and its community: offline, 100% open source, multilingual, integration with Home Assistant and Node-RED, but still a bit rough around the edges. The basic architecture is there, but there’s no apps ecosystem yet. So there’s still a lot where you all can contribute
I will definitely set up a pi with rhasspy to test, looks very promising as I run my complete action handling in nodered anyway. Thanks a lot.
What do you think? I mean you invested a lot of time with Project Alice.
What I think? Let’s ask Alice… https://twitter.com/Psychokiller188/status/1198285177270808576?s=20
Snips would have known everything about whats going to happen at the time of signing with Sonos.
Since Rand himself does not answer the question outright when asked, you can beat that its not good news, else they would just say it to alleviate Makers minds about whats happening.
Just read between the lines of what has been publicly said
sometimes things aren’t as easy as they seem, and we need to change direction. It’s the nature of entrepreneurship!
Rand’s twitter tag
Entrepreneur & deep tech investor. Formerly CEO @snips
Snips is no more, no announcement to users as to what the sell means to them is not a good sign.
Never going to use a product again that states they are “open source” when none of it is open sourced.
Good buy () snips , it was so good to have fun with you, now i have to found another private solution
which one’s that???
From what I learned by members of the Snips team at IFA in Berlin they just thought of Snips Air as a case study. If they had any plans to make it happen, they stopped these some time ago. Their focus was on the software and any devices were supposed to be developed by partners like Bosch. Now that Sonos bought them, I guess the idea behind Snips Air will be revived and they will build their own Smart Speaker / Voice Assistant to integrate it in their products. At least it sounds like that is the plan:
I guess they plan to release an alternative to Google, Alexa and Siri that keeps the “on device” aspect as an USP. Everything else like the community and the brand could be gone sooner or later.
I wonder if we can turn the conversation to what other privacy-first voice platforms are available. I’ve heard/seen Picovoice and Rhasspy (sp?) mentioned. I’ve been trying to grade these platforms on 5 categories
- Wake Word Engine
- Speech to Intent Engine
- Text to Speech Engine
- Intelligent Response Engine (based on context of dialog it can create an appropriate response to send to the text to speech engine)
- Ease of developers to end intents and features
Anyone have suggestions and a 1-5 grade on each category? (1 being very bad, 5 being very good)
Alternatively, could the community take the current Snips and continue it as an open source project? What would it take to not be dependent on the Snips platform? I’m assuming it’s the training part and hosting the apps ecosystem?
No also snips always advertised it’s open source component most of it actually isn’t.