Apple details Siri's machine learning upgrades for better listening on HomePod
Apple on Monday updated its Machine Learning Journal with a post by the company's Siri speech and audio software engineering teams, explaining how the company uses machine learning to help the HomePod hear people under tougher circumstances than iPhones and iPads.
Siri on the HomePod had to be upgraded to cope with loud music, ambient noise, and distant talkers, the journal entry notes. Accordingly the HomePod employs not just far-field microphones, but "mask-based multichannel filtering using deep learning" to strip out echo and background noise, and "unsupervised" learning to split up multiple sound sources and use only the one including "Hey Siri" as a trigger phrase.
The entry goes into considerable technical detail, being oriented mainly towards professionals in the machine learning field. It does however mention that multichannel signal processing happens "continuously" on the HomePod's A8 processor, even in low-power states, and is able to adjust to both changing environments and moving talkers.
Apple suggests that while "other state-of-the-art systems" use multi-microphone processing, they typically only focus on echo and noise reduction.
The HomePod is a relative late comer to the smartspeaker market, having launched just this February -- Amazon's Echo debuted in 2014, and the Google Home shipped in 2016. Apple though has taken a different tack than many vendors, concentrating on sound quality with technologies like beamforming. A HomePod will automatically tune itself to match its position in a room.
Siri though has been criticized as limited next to Amazon and Google's voice assistants, for instance only natively supporting Apple Music in streaming services. The HomePod hardware is also expensive at $349 -- Apple has been rumored as working on a cheaper model to be more competitive.
The journal update coincides with Apple's appearance at the 32nd Conference on Neural Information Processing Systems in Montreal, Canada. The company has tried to open itself more to the academic community, presumably to appeal to potential recruits and to appease the researchers it already has, who previously complained about Apple's paper publishing restrictions.
The Machine Learning Journal is another one of those compromises. It began in July 2017 with a paper discussing neural net training with collated images, and has since gone on to cover a variety of topics, such as face detection and "differential" privacy.
A year ago the company's s director of AI research, Ruslan Salakhutdinov, spoke about the company's self-driving car project, the ultimate purpose of which is still shrouded in mystery. The company may or may not be working on a self-designed car -- it at least briefly abandoned the idea in favor of pure platform development.
Siri on the HomePod had to be upgraded to cope with loud music, ambient noise, and distant talkers, the journal entry notes. Accordingly the HomePod employs not just far-field microphones, but "mask-based multichannel filtering using deep learning" to strip out echo and background noise, and "unsupervised" learning to split up multiple sound sources and use only the one including "Hey Siri" as a trigger phrase.
The entry goes into considerable technical detail, being oriented mainly towards professionals in the machine learning field. It does however mention that multichannel signal processing happens "continuously" on the HomePod's A8 processor, even in low-power states, and is able to adjust to both changing environments and moving talkers.
Apple suggests that while "other state-of-the-art systems" use multi-microphone processing, they typically only focus on echo and noise reduction.
The HomePod is a relative late comer to the smartspeaker market, having launched just this February -- Amazon's Echo debuted in 2014, and the Google Home shipped in 2016. Apple though has taken a different tack than many vendors, concentrating on sound quality with technologies like beamforming. A HomePod will automatically tune itself to match its position in a room.
Siri though has been criticized as limited next to Amazon and Google's voice assistants, for instance only natively supporting Apple Music in streaming services. The HomePod hardware is also expensive at $349 -- Apple has been rumored as working on a cheaper model to be more competitive.
The journal update coincides with Apple's appearance at the 32nd Conference on Neural Information Processing Systems in Montreal, Canada. The company has tried to open itself more to the academic community, presumably to appeal to potential recruits and to appease the researchers it already has, who previously complained about Apple's paper publishing restrictions.
The Machine Learning Journal is another one of those compromises. It began in July 2017 with a paper discussing neural net training with collated images, and has since gone on to cover a variety of topics, such as face detection and "differential" privacy.
A year ago the company's s director of AI research, Ruslan Salakhutdinov, spoke about the company's self-driving car project, the ultimate purpose of which is still shrouded in mystery. The company may or may not be working on a self-designed car -- it at least briefly abandoned the idea in favor of pure platform development.
Comments
For instance, I just told Siri to “play chill house music” and the reply was “I have a hunch you’ll like this tropical house”. Wrong and incorrect.
So, not only do I have to deal with siri’s obnoxious and repetitive “personality,” Siri as well as apple
music continues to be complete shit.
I crige every time I have to use HomePod for anything other than rote HomeKit commmads.
Even as an airplay speaker HomePod sucks. Sometimes it doesn’t respond at all. And when I use it as an airplay speaker it drops the connection to my iPhone all the time, so I usually have to resort to a separate Bluetooth speaker. I keep all my iOS products new and updated, including the latest OS, the latest iPhone, and the latest Apple TV, yet there’s always problems.
Continued, huge gaps in the Apple ecosystem. An ongoing slap in the face to Apple users. Over and over.
No wonder investors are skittish about AAPL. The only thing that is consistently reliable is the core hardware. Services continue to be almost fucking useless (except for App Store revenue, which of course exists only bc the iOS hardware is so fantastic).
Edit: forgot to mention that my HomePod likes to activate itself randomly when I’m watching tv (using the crappy built in tv sv speakers).
And I occasionally ask it to flip a coin.
Ever consider that your WiFi router is the problem? The only time Airplay has failed is when my WiFi network has bogged to a crawl for whatever reason (I miss my Airport Express).
Seriously? Do you even understand what Apple's services are?
iCloud is a godsend. My phone was stolen the other day and it took almost no time to get a new one set up with all of the same information again. I remember the days when I'd have to pray that I'd backed it up recently, and then spend a number of hours syncing everything over to it again (and having to keep it attached to my computer). Now it just happens automagically once I log in to my iCloud account.
Then there's iTunes Match, which works beautifully for getting my music collection on every device.
This is my only gripe with HomePod. But the random activations seem to be happening far less often these days, so I'm assuming Siri is getting better behind the scenes.
Using it for a shopping list is a great idea! I'll have to try that.
I can't find a comprehensive how-to source, so she's no longer any help with home automation. Millions of people don't seem to have that/those problems, but that's no help for me.
Siri works more reliably for me, and seems to have been upgraded, if slightly, over the last part of the week and/or weekend. Siri isn't nearly as bright as Alex but is more obedient. If BB has another $100 off deal, I'll give the HP a shot. I wish B&H would do the same.
For the time being, they both compliment each other as neither does everything I want. Both have flaws, such as activating when hearing something on TV that may or may not sound like 'Hey, Siri...' or 'Alexa...' to me. My living room Dot activates occasionally for no reason I can see, with no audio in the room. I just assume it's the NSA checking in.
Other than that, it's been a great addition. The sound is wonderful and combined with Apple Music, it suits our needs well.
Something else I use it for: spelling out words while I’m writing.
The HomePod can’t even stream radio stations. A smart speaker it isn’t.
My HomePod was bought on day 1 of release. By week 3 it had made a ring on my wooden table, been disconnected and put away in the hope that Apple would eventually add abilities that made it useful to non Apple Music users.
Would most of the complaints about Apple Music go away if it was cheaper?
I find it to be of tremendous value, but then again, I just pay the equivalent of ~$2 per month for a family subscription.