Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mycroft – open source voice assistant

mycroft.ai

Post reply on HN
Print view View original

131–140 of 142 posts

Re: Mycroft – open source voice assistant

#131

Post by dilek »

shagie wrote: post #128
dilek wrote: post #125

Earlier quoted context omitted.

Amazon has been working on on-device voice for a while. Actually everyone is trying to do that. Running large speech models in the cloud is expensive, considering the number of devices, they probably need more than "surplus" :) https://www.amazon.science/blog/on-device-speech-processing-...

It could certainly rack up a big bill. 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 ran…

I didnt disclose as I was not gonna promote anything, but I work for a startup specializing in on-device voice recognition. I am 100% biased towards on-device voice processing :)

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...

Re: Mycroft – open source voice assistant

#132

Post by artificial »

shagie wrote: post #117
one-another-dev wrote: post #102

Earlier quoted context omitted.

> If you've got an iPhone... put it in to airplane mode so that it is local only. You'll note that Siri no longer works when you do this This is not true anymore. Latest iPhone models have offline Siri working to some extent

Fair 'nuff... though I was after a "even with an older model iPhone, and no net connection, there the ability to do speech to text (and even with some interesting transformations of "two by four"), it can be done locally." What's more... consider https://mycroft-ai.gitbook.io/docs/using-mycroft-ai/customiz... > In order to provide an additional layer of privacy for our users, we proxy all STT requests through Mycroft…

You may be interested in a little background on Siri on device with CoreAudio https://machinelearning.apple.com/research/hey-siri

Re: Mycroft – open source voice assistant

#133

Post by Zuiii »

rexreed wrote: post #26

The voice assistant space is dying: https://arstechnica.com/gadgets/2022/11/amazon-alexa-is-a-co... 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.

I always wanted a voice assistant but there's no way I'm having big tech listen in on me and my family 24/7 just to have one. Most non-tech family members I talked with about this share my opinion. THAT is why these assistants are failing.

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.

Re: Mycroft – open source voice assistant

#134

Post by shagie »

artificial wrote: post #132
shagie wrote: post #117

Earlier quoted context omitted.

Fair 'nuff... though I was after a "even with an older model iPhone, and no net connection, there the ability to do speech to text (and even with some interesting transformations of "two by four"), it can be done locally." What's more... consider https://mycroft-ai.gitbook.io/docs/using-mycroft-ai/customiz... > In order to provide an additional layer of privacy for our users, we proxy all STT requests through Mycroft…

You may be interested in a little background on Siri on device with CoreAudio https://machinelearning.apple.com/research/hey-siri

I didn't know the specifics of it, that has a lot more information and is is an interesting read.

One of the bits in there caught my eye...

> We created a language-specific phonetic specification of the "Hey Siri" phrase. In US English, we had two variants, with different first vowels in "Siri"—one as in "serious" and the other as in "Syria." We also tried to cope with a short break between the two words, especially as the phrase is often written with a comma: "Hey, Siri." Each phonetic symbol results in three speech sound classes (beginning, middle and end) each of which has its own output from the acoustic model.

And the British version getting false positives on wake up with world politics.

https://youtu.be/-W_uiPVsTVM

https://youtu.be/2YjxiBqV6Kc

The specifics of the wake up and that its done with a ML model rather than a low power wake word chip akin to https://www.syntiant.com/post/syntiant-low-power-wake-word-s... is also interesting - and impressive that they were able to get it to be that low power.

Re: Mycroft – open source voice assistant

#135

Post by lannisterstark »

ElijahLynn wrote: post #112
GekkePrutser wrote: post #98

Earlier quoted context omitted.

Those prices are really insane tbh. No way it will take off like that, it's just a non-starter. 200 was already the upper limit of what's doable. I think they're stretching too much to satisfy the original backers and it's commendable not giving up on them but if it's going to be similar to Alexa, Siri or Google it's just not good enough. If it were an actual assistant I could talk to, then yes. It would be worth it.…

I'm not sure that price is going to be a blocker. The audience may not think of price as the main thing here, they may think of openness and non-vendor lockin and privacy and security as pretty amazing things for that price. I myself am getting tired AF of Google Assistant and their devices. So so tired of saying "Hey Google". And I have a Lenovo device with Google on it that was decent, the Google pushed an update t…

>I'm not sure that price is going to be a blocker.

Eh, idk if I want to spend $350 on what is arguably a half baked, subpar Google-Assistnat/Alexa.

Re: Mycroft – open source voice assistant

#136

Post by arbol »

shagie wrote: post #93
arbol wrote: post #42

Earlier quoted context omitted.

The pi isn't really fast enough to process the speech in real time. deepspeech by mozilla was cited as an offline alternative to the Google speech API but it's difficult to set up with Mycroft and doesn't work very well (lack of data and lag - https://mycroft.ai/voice-mycroft-ai/ ). Because of this, Mozilla set up Common Voice ( https://commonvoice.mozilla.org/en ) to help build open datasets of voice recordings.

> The pi isn't really fast enough to process the speech in real time. If you've got an iPhone... put it in to airplane mode so that it is local only. You'll note that Siri no longer works when you do this. However... open up the notes app and tap the microphone. Do some interesting text... > Mister Smith said that he wanted a two by four and half of a pie. (if you don't have an iDevice, it transcribes this as: > Mr.…

Yeah, but this is the closed source Apple implementation of speech to text versus Mozilla's abandoned deepspeech. I'm sure its possible to get it working well on a pi but I don't have the time to create and maintain a personalised speech training set and then optimise the resultant models.

Re: Mycroft – open source voice assistant

#137

Post by nshm »

arbol wrote: post #42
wolczek wrote: post #34

They use Google cloud speech-to-text API, one of the key technology. https://mycroft-ai.gitbook.io/docs/mycroft-technologies/over... ¯\_(ツ)_/¯

The pi isn't really fast enough to process the speech in real time. deepspeech by mozilla was cited as an offline alternative to the Google speech API but it's difficult to set up with Mycroft and doesn't work very well (lack of data and lag - https://mycroft.ai/voice-mycroft-ai/ ). Because of this, Mozilla set up Common Voice ( https://commonvoice.mozilla.org/en ) to help build open datasets of voice recordings.

DeepSpeeech is very old software. Vosk works just fine https://github.com/alphacep/vosk-api. People even run tiny Whisper on Pi, though they have to wait ages.

Re: Mycroft – open source voice assistant

#138

Post by nshm »

theCrowing wrote: post #4

I use Mycroft for around 2-3 years on a raspberry pi with an microphone array and the quality is still not nowhere near the level were I would give it to my mom or granny.

Try Sepia https://sepia-framework.github.io, it is an open source assistant, works very good.

Re: Mycroft – open source voice assistant

#139

Post by joseda-hg »

srmarm wrote: post #9
incomingpain wrote: post #3

$350 for a 4.5" CRT looking thing? OR $90 for echo show 8" which does many more things. *Including government surveillance. I wonder how many people have preordered.

There was another post[0] on here today about Amazon losing $10B on Alexa this year. The only other big player is Google who I assume must also run the division at a big loss - at least on the hardware side of things as I've got loads of their devices dotted around the house most were given to me free or at a stupidly low cost (£20 each). Even the ones like this with a full colour screen I've only paid £49 for. It's…

Google to some degree can justify it better considering that Assistant is ALSO present in Android

They need to have some sort of version existing to compete with Siri, and can also shoehorn integrations that wouldn't quite work for Amazon

Were not for this[0] ever expanding page, Assistant seems like the more reliable option

[0] https://killedbygoogle.com/

Re: Mycroft – open source voice assistant

#140

Post by GekkePrutser »

shagie wrote: post #128
dilek wrote: post #125

Earlier quoted context omitted.

Amazon has been working on on-device voice for a while. Actually everyone is trying to do that. Running large speech models in the cloud is expensive, considering the number of devices, they probably need more than "surplus" :) https://www.amazon.science/blog/on-device-speech-processing-...

It could certainly rack up a big bill. 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 ran…

Interesting! Is that also the case for the homepod (and homepod mini which I have)?
Post reply on HN