Teken Posted October 5, 2017 Posted October 5, 2017 It looks to be a very interesting time for the connected speaker system as seen by this article for Google Home Mini & Max; https://www.theverge.com/2017/10/4/16405218/new-google-home-mini-max-speakers-photos-video-hands-on Two things I really found interesting was the auto sound adjustment and the so called ultra listening ability. In the linked video the author states one of the Google Engineers had the Home Max cranked up. While the back ground was filled with loud sound he stated when the engineer muttered *Hey Google* that the speaker system heard him and waited for the next command? I've been involved in the sound field for many decades and if this video is true. Google has something truly incredible on their hands which can be applied in many different industries. The ability to hear & listen for something (key word) while the back ground is extremely loud has never been reliably and consistently accomplished in any system I know of. The military & spy agencies have made huge strides in this area but none of them have ever been able to isolate a faint sound over louder ambient noise, ever. The closest they have been able to fine tune these systems is being able to triangulate where a sound came from. In this case the military uses it to detect where a gun shot has originated. As impressive as this sound detecting platform is it still can't compensate for echo effect in a city dwelling scenario. In a open field, forest, etc it does well . . . I believe the first company that offers *Sonos* like linking, ease of use, low price, proactive notification, and the ability to send unique and custom sound files to said speakers will have a money maker.
LFMc Posted October 5, 2017 Posted October 5, 2017 In the linked video the author states one of the Google Engineers had the Home Max cranked up. While the back ground was filled with loud sound he stated when the engineer muttered *Hey Google* that the speaker system heard him and waited for the next command? The military & spy agencies have made huge strides in this area but none of them have ever been able to isolate a faint sound over louder ambient noise, ever. The closest they have been able to fine tune these systems is being able to triangulate where a sound came from. In this case the military uses it to detect where a gun shot has originated. As impressive as this sound detecting platform is it still can't compensate for echo effect in a city dwelling scenario. In a open field, forest, etc it does well . . . I believe the first company that offers *Sonos* like linking, ease of use, low price, proactive notification, and the ability to send unique and custom sound files to said speakers will have a money maker. I watched the link and as you said, it (the Goggle device) was playing loud music. Wouldn't it seem fairly obvious that if your listening device is also the source of most of the sound/noise, it will have a tremendous leg up on cancelling or filtering out the generated sound coming into the mic? Even any reflective sound is going to follow a very similar pattern. So if the noise was not generated by the G.Home device, it would be much tougher to filter. I am always amazed even at my Echo Dot how it picks up our trigger word even when the TV is cranked up, yet in the silence of the night and I tell it to turn off the lights I have a 50/50 shot it/she just ignores me, lol. I might prefer the Google device to be honest, but the price is a bit high for now. Also noise is never really random. I once worked on a project, like you mentioned, where we could see very distinct patterns in waves to determine what were "natural" waves and what were "unnatural" waves as well as their source. To say more might trigger a knock on my door.
Teken Posted October 5, 2017 Author Posted October 5, 2017 I watched the link and as you said, it (the Goggle device) was playing loud music. Wouldn't it seem fairly obvious that if your listening device is also the source of most of the sound/noise, it will have a tremendous leg up on cancelling or filtering out the generated sound coming into the mic? Even any reflective sound is going to follow a very similar pattern. So if the noise was not generated by the G.Home device, it would be much tougher to filter. I am always amazed even at my Echo Dot how it picks up our trigger word even when the TV is cranked up, yet in the silence of the night and I tell it to turn off the lights I have a 50/50 shot it/she just ignores me, lol. I might prefer the Google device to be honest, but the price is a bit high for now. Also noise is never really random. I once worked on a project, like you mentioned, where we could see very distinct patterns in waves to determine what were "natural" waves and what were "unnatural" waves as well as their source. To say more might trigger a knock on my door. LOL . . . We don't want the *Men in Black* to be at your door, Leon . . . I'll be quite honest until I am able to test this voice detection myself I don't believe it. I've been around multi-million dollar systems and none of what I have seen, used, tested, could ever detect a faint voice command over a loud ambient back ground. Having said that sound technology has advanced several fold in the last ten years. The fact you can literally focus a bean at a single person / a whole crowd and bring them to their knees is something awe inspiring to say the least. I will simply put this out there for who ever happens to stumble upon this thread. Please make a video of the speaker blasting away and while its doing so ~ mutter the wake up phrase on video and lets see what happens. I will wager nothing happens at a reasonable distance . . . In the video the author doesn't define the actual distance of the engineer vs the speaker. So for all we know he could have been mere inches from the hardware. Regardless, I am eager to see what other companies have to offer in the not too distant future as competition is good for everyone and for advancing different technologies.
Bumbershoot Posted October 6, 2017 Posted October 6, 2017 The Amazon Echo accomplishes something similar to this when the new Sonos skill is invoked. If you initiate music on Sonos devices from Alexa, and crank them up, when you invoke the wake word, all the Sonos will drop their volume considerably and wait for the command. I suspected, initially, that I'd have to stand over the Echo and shout, "Alexa!" over and over... I don't have to, which makes the Sonos skill useful and greatly helps with the WAF. We've never tried this during a noisy house party, however. Sonos Tureplay can accomplish some "tuning" to match room acoustics, but only using Apple iPhone and iPad devices (they say the consistency of the microphones in these devices is the reason for that). Nothing, however, in real-time. I don't know if real-time acoustical tuning would annoy me or not.
Teken Posted October 12, 2017 Author Posted October 12, 2017 Here is another perfect example of why you need to have complete control over these *Always Listening* devices: http://appleinsider.com/articles/17/10/11/google-home-mini-preview-hardware-defect-causes-near-constant-listening-uploads-to-servers
Recommended Posts
Archived
This topic is now archived and is closed to further replies.