This is huge ! Well done ! Little precision taht snips has the best NLU performance to date ! What kind of platforms / devices are you going or looking to integrate ?
@benjmerritt thanks 🙌
There are two ways to use Snips today:
- as a maker, or for prototyping on a Raspberry Pi, you can use our web console and build your assistant yourself, for free
- as a commercial product maker, we can make a build that will run on your device, on any platform (Android, Linux, etc..)
The whole thing is written in Rust, so it can cross-compile and works super fast!
Report
Being able to create voice commands right on-device and not have to query a server is awesome.
What I like about other services like api.ai though is the ability to train an agent based on unmatched inputs, making NLU even better overtime. This is prob not possible with Snips, right?
One more thing: citing "Resilient to Cloud Outages" as an argument for going on-device seems weird to me as it's rarely something one should optimise for IMO.
@gflandre we actually have an auto-tagging feature that leverages the self learning abilities of our engine. We will also soon add a killer feature that enables you to automatically augment your dataset to get better performances out of the box!
For the resilience, it turns out to be important for devices such as cars and home appliances that must respond at all time
Report
@gflandre We will add ways for users who accept to share data to give queries and improve their NLU too. Many devices might not be connected to the net at all time, think cars, or when you are travelling in a remote area. But first and foremost we don't want the variable costs of cloud subscriptions and we want to keep data safe
Report
@randhindi@mael_p That's awesome if in the future the NLU can be improved!
Indeed I didn't have the "cars" use-case in mind but that makes a ton of sense. Of course in terms of privacy/performance/cost, this is great!
Report
Looks pretty neat. Did you pivot from the personal assistant product? I had the app but found it... not very useful and it used quite a bit of battery. Can you describe the pivot in more detail and how you came to create the current product?
@danr_4 Hey there Dan, the underlying technology has been developed from the app into our new Voice Assistant Platform. After the launch of the app, we had an overwhelming number of companies inquire how they could include our tech in their products, so rather than pursue the mobile app route, we launched the platform. Now makers, prototypers, and enterprises can integrate a voice assistant into their products with ease.
@robjama Thanks brother!
The Snips Voice Platform allows anyone to build an AI powered voice assistant with ease. This is the first end-to-end platform to offer on-device ASR and NLU, while protecting user privacy.
The Snips Assistant can be configured easily through our web console, which includes the ability to select from pre-built assistants or to create completely custom ones. The assistant can then be deployed to a RaspberryPi, to Linux or Android, with more platforms coming soon. Come give it a try!
www.snips.ai
@chevreuil totally, it's time we stop saying we have to chose between AI and Privacy!
Report
@randhindi@chevreuil I would never say we don't need more privacy, but I wouldn't say this is the first AI-powered voice assistant "with privacy". That's misleading. Siri already protects your privacy extremely well.
For instance, "When we do send information to a server, we protect your privacy by using anonymized rotating identifiers so that searches and locations can’t be traced to you personally." https://www.apple.com/privacy/ap...
FunApéro
Encrypted Image Filtering App
Encrypted Image Filtering App
Data Generation by Snips
Boom video
Data Generation by Snips
Data Generation by Snips
Encrypted Image Filtering App
Product Hunt
Data Generation by Snips
Encrypted Image Filtering App