Earlier quoted context omitted.
I've been meaning to take a look at Spotify through Home Assistant and what that is capable of (and potentially voice controlling it) But, if you wanted to write some python to directly play some pre-defined playlists (it currently can't do open speech transcription), you could definitely make that happen with Home Intent! It supports custom components, so any integration is technically possible: https://homeintent.i…
Cheers! The Spotify part I'm assuming I will figure out - was more curious as to how your experience with voice control has been, if it performs well enough to be practical.
Ask HN: Private Alternatives to Alexa?
141–150 of 166 posts
Re: Ask HN: Private Alternatives to Alexa?
#142Earlier quoted context omitted.
Yeah, they absolutely were not scanning photos and anyone with basic knowledge of how hashing works should know that. I was really disappointed with some of the reactions a saw coming from this community, which totally lacked any nuance or acknowledgment of how serious the issue of child abuse is.
How is running a detection algorithm on one’s photos not “scanning” them?
Since creating a hash from a file is a one way process that you cannot reproduce the original file from the hash output, I would not call this a scan. All you can do is match to known child porn files.
But yes, go ahead, make more pedantic excuses on why we should protect people that hold child porn.
Re: Ask HN: Private Alternatives to Alexa?
#143Earlier quoted context omitted.
Yeah, they absolutely were not scanning photos and anyone with basic knowledge of how hashing works should know that. I was really disappointed with some of the reactions a saw coming from this community, which totally lacked any nuance or acknowledgment of how serious the issue of child abuse is.
They are Not scanning the photos But if there is a positive scan of the hash of a photo then the photo will be send to a human to check. Now with a certain error rate it is sure that humans working at apple will see privat photos that don’t have child pornography in them .
Re: Ask HN: Private Alternatives to Alexa?
#144Earlier quoted context omitted.
This was changed as far as I know. Do you have evidence otherwise?
Claiming that whatever has not been proven false must be true is the exact opposite of how it works. We have evidence and even admittance of abuse / improper handling in the past, but I'm unable to find any evidence that things have changed. https://en.wikipedia.org/wiki/Argument_from_ignorance
https://www.cnbc.com/amp/2019/08/28/apple-apologizes-for-lis...
I was just asking a question, I wasn’t attempting some sort of gotcha. I really don’t care that much.
Re: Ask HN: Private Alternatives to Alexa?
#145Earlier quoted context omitted.
This was changed as far as I know. Do you have evidence otherwise?
Do you have evidence that they changed it?
This. Is there any reason I should think this is not accurate?
Re: Ask HN: Private Alternatives to Alexa?
#146Dumb question: Why does Alexa need to connect to the internet to turn lights on/off or run timers. What happens when it cannot connect to the Amazon mothership.
It does all voice recognition in the cloud, only the trigger word detection runs locally.
[1] https://community.home-assistant.io/t/local-voice-control/29...
Re: Ask HN: Private Alternatives to Alexa?
#147You could also keep using Alexa and add a do-it-yourself privacy enhancement called Alias: https://www.instructables.com/Project-Alias/
Re: Ask HN: Private Alternatives to Alexa?
#148If you can get comfortable with Siri (which does send audio to the cloud, though with iOS 15 there's more on-device capability), you should experiment with controlling Google Assistant and Amazon Alexa via Siri. I set up my phone so I can say "hey Siri ok google" and then it asks me what I want to tell google. I then say any command supported by the Google Assistant, and it passes the command through. Technically I t…
Re: Ask HN: Private Alternatives to Alexa?
#149Earlier quoted context omitted.
How is running a detection algorithm on one’s photos not “scanning” them?
It’s a bit of a stretch to say generating a hash from a file is scanning. Most computer science terms like this are metaphors, but when I create a scan with a ‘scanner’ I can create an exact copy of what was scanned. Since creating a hash from a file is a one way process that you cannot reproduce the original file from the hash output, I would not call this a scan. All you can do is match to known child porn files. B…
There are types of hashes that aren’t for finding exact matches. My thesis in school way back when was building something for image near-duplicate detection using a sort of hash made out of the image features detected. The output of a neural net run on an image is also a sort of hash in a (maybe) semantically meaningful vector space. And then you can look for similar or nearby vectors. This isn’t just useful for finding one type of image.
So, in case you didn’t realize, and you’re actually arguing in good faith, no one is arguing against this because they want to defend sex offenders, so you can leave that out in the future, unless you want to seem disingenuous. It’s because it makes it easy to create a population-wide dragnet for other things. It’s not hard to apply something similar to find photos taken at a protest, for example. Or pictures of a specific person you happened to meet.
Anyway, have a good one.
Re: Ask HN: Private Alternatives to Alexa?
#150It's unfortunate that another comment that links to Rhasspy has been downvoted (I assume because it lacked any other context) so I wanted to mention the project with some additional context: https://rhasspy.readthedocs.io/ While I've not used the entire Rhasspy project myself (but trying it out is on the long list of things to do :) ) I have used the offline Text-To-Speech sub-project Larynx... ...and it is amazing !…
https://ai-service-demos.go-aws.com/polly check the neural then british, Amy. the "smoothness" is uncanny. the samples are "almost" there with larynx you linked. good but take kathleen (glow_tts), there is "still" some robotic in there. is this something that can be improved by tweaking the training ? this sounds really cool to be used at home
And Larynx is game-changingly ahead of the other FLOSS+offline options.
(Probably the highest quality voice is the one that's used for the demo video narration--which when I first heard it I had to skip to the end of the video to confirm it wasn't a live human. :) )
My (mostly uninformed) impression is that there's room for training tweaking/improvement given how young the project is. And there's also multiple stages to the generation process so presumably there's opportunities at each stage.