Article: The Value Proposition In Audio: Voice Control For Audiophiles (You Can't Buy Too Soon - You Can Only Sell Too Late)

[bluesman]

View full article

[bluesman]

1 hour ago, The Computer Audiophile said:

This is fantastic @bluesman

 

I've been following josh.ai for a while now. I think josh is the company to watch in the high end space for sure. 

Josh is cool, for sure - but they seem to be using current methods and tools to achieve something for which current methods and tools are not ideally suited.  Unless they're the ones to come out of the garage or basement with the next big thing in AI coding and output modalities, they'll be looked on as crude when someone else finally succeeds.  I know of no current platform that integrates the various functions necessary to achieve accurate and efficient voice control over devices in disparate systems, including what I know of Josh (which admittedly isn't a lot at the design level).

 

Right now, there are too many data bouncing 'way too far over 'way too many jury-rigged networks to do this smoothly.  And platform integration is not in the cards for an industry that profits largely from differentiation, so there's not likely to be one approach shared by all. This is a lot like the world of electronic medical records.  The most they hope for is "interoperability" - and that has us unable to share data universally across all healthcare institutions plus payers and the scientific community.  All Epic users can share their data if they wish, as can users of several other major EMR platforms.  But these programs aren't written in the same languages and they run on different architectures.  So if your hospital is on Epic and the one where you ended up unconscious in the ER because you fell off the train is on Cerner, you're out of luck unless you carry your medical records around on a USB drive or a CD.

 

It's a lot like needing home hubs for Z-wave, Zigbee, Google Home, Samsung Smart Things, SmartLife, and Apple Home because you have a few devices that work on each platform.  There are several smart speakers that require you to control some audio functions from Alexa and some from Google Assistant - this is no way to run a railroad.  As I was just saying to my watch, "Siri, tell Alexa to open House Band; Siri, tell Alexa to tell House Band to tell JRiver to play music by Wayne Henderson in the master bedroom; Siri, tell Alexa to make the music louder;"

[bluesman]

39 minutes ago, jrobbins50 said:

Good article. I use five Harmony Hubs (only $70 each at Amazon) at home, each tied to a different Gmail and Amazon account.

Thanks!
 

Your willingness to use multiple accounts and devices to “integrate” functions says that you’re flexible and adventurous, like me.  But we’re in the minority by far - most people would think we’re a bit daft to go that far.......and we shouldn’t have to.  I believe it won’t be long before there are more universal platforms and approaches available to us.  But until then, let’s stretch the envelope to see how much it can hold 😁

[bluesman]

3 hours ago, blue2 said:

Very professional article! Piqued my interest so did a bit of Googling:

 

System	Platform/Hardware	Notes
aido	Robot with GUI	cameras, multiple CPU's and GPU's, not available yet, price?
athena		Open source software project written in Python
bixby	Samsung mobile devices	Samsung's Google assistant
hound	Automotive
jibo	Robot for healthcare and education
josh		interfaces in posh homes to Lutron lighting, Sonos, Crestron thermostat, home security, smart TV's, home theatre
mycroft	Runs on Raspberry Pi etc.	Open source software project (Python?), Mk I was $180 now sold out, Mk II coming soon
ubi ucic	Runs on Android and Linux	Ubi Kit is free for developers supports Google assistant and Alexa
viv.ai		Viv is an artificial intelligence platform - intelligent personal assistant software created by the developers of Siri, bought by Samsung. Now to be integrated into Bixby 2.0

Thanks for your time & comments!  There's so much potential here that I'm amazed the audio industry hasn't recognized how important VC & AI are for development and sales of future products.  We all experience mechanical controller failures in everything we use, from audio to cars to coffee machines.  Tiny touch screens, bubble switches, touch sensitive controls etc are only pseudoelectronic - they still have physical parts that fail too often.  We could eliminate most of those last century pieces and concepts by integrating excellent voice recognition and synthesis with AI. Imagine no more noisy pots, no cracked or dented bubble switches, no broken or lost knobs, minimal internal wiring, etc.

 

Then imagine being able to control and monitor every audio parameter of interest to us in real time using voice input and synthesized voice response.  Throw in AI's ability to monitor real time performance and identify impending failures by detecting as yet inaudible changes in everything from voltage & current stability at various points to distortion to early ID of asymmetry in channel outputs.  In addition to telling your system what you want to hear (and how and where and when...), you could ask for a status check and get a verbal response plus a downloadable log report.  You could set up spontaneous verbal warnings when voltage, temperature, and other metrics go out of spec.  You could alter or switch amplifier operating characteristics to A-B changes in SQ.

 

How about a status report when powering up, e.g.  "Good morning, Bob - your system is in perfect operating condition and ready to play"?  Get vocal alerts as needed - "THD in your left channel has increased to 105% of the right channel.  Diagnostics show early failure of V4 with no other abnormality.  Replace tube."

 

This is cool stuff!  I can't wait to play with it all as it develops.

[bluesman]

1 hour ago, audiobomber said:

The last sentence needs an ending.

 

Note that you don't need JRiver to use DLNA with Chromecast Audio. I use my CCA devices with two QNAP controllers (Music Station and QMusic), and with BubbleUPNP. I believe there are others too, maybe MConnect and Kazoo?). No voice control though, which I don't care about. 

 

Whoops!  Chris was having some problems with the formatting of the original document I sent (a conversion from odt to docx).  When I converted it to a pdf for him, I must have converted the wrong draft.  Here's what it should have said:

 

"If you want to have voice control over JRiver playing through a Google device as a zone, you’ll have to use Alexa. If she’s not sharing a device with the GA, you can link her from an Amazon device or a third party host using an app like Helea Smart."  [Chris, if you can drop this in, it will save others the irritation of the typo.]

 

I'm sorry if I gave the erroneous impression that you had to use JRMC in order to cast to CCAs in general.  My point was that if you use Google smart speakers but want to have voice control over JRMC playing to them, you have to use Alexa either from an Alexa-enabled device or with a 3rd party integration app. Google speakers will show up as DLNA zones in JRMC if you have BubbleUPnP etc running along with JRMC, so there's some functional integration there among JRMC, Alexa and the GA (albeit crude integration).  But it's not ideal, and it's one reason we chose Amazon / Alexa for our primary smart platform.

[bluesman]

7 minutes ago, The Computer Audiophile said:

Ahhhh. Done. 

Thanks!  Sorry about that - I should have caught it.

[bluesman]

1 hour ago, palpatine242 said:

For an audiophile system with voice control, can't you use a Bluesound Node 2i with digital output to an external DAC which feeds your stereo?  You can then use Google Assistant or Alexa to choose a song to play on the Node 2i.  I would assume that Tidal Connect to the Node 2i could be controlled with Google Assistant and Alexa as well.

Yes you can.  There are several streamers like this, but a comparison was far beyond the scope of this article. Yamaha makes 2 models with which I'm familiar, Denon has the Heos system and devices, etc.  Voice control in all of these is limited by what Alexa, GA etc can do.

 

From the Bluesound website, "...you can use voice commands to play saved playlists, select your favorite radio station, adjust volume levels, or even group Players together".  The Node 2i will respond to Alexa using the Bluesound skill and to Google Assistant using middleware called Blue Voice. As long as you've set up everything correctly from the DAC downstream, you can control the stated functions from the Node - but you can't control any other element in your system.  There are enough downsides to this to deter me from using it.

 

For example, your power amp gain control has to be set high enough to encompass the loudest playback you'll ever use, if Alexa's controlling your system volume at the streamer.  This leaves your speakers vulnerable to any transients generated in your front end but not attenuated by the variable gain stage, e.g. at turn-on and turn-off or when switching sources.  If you have an uncontrollable pop with any function, your voice controller can't cut the gain before executing it.

 

The BS Node is marketed to "...instantly [breathe] new life into your decades-old stereo equipment" by adding network and web streaming sources. The only analog inputs I see are in the combo 3.5mm optical / line level jack, which must be how one connects a turntable or CD player.  I can't tell if you can stream the output over the LAN or WLAN (I assume not) but the 2 way BT should let you drive BT speakers with any source.  The USB input is only for storage devices - you can't connect a USB turntable or other real time USB source.  There are many limitations to this approach to voice control, although it does work within the limits of system and technological constraints.  Stay tuned - it will get better!

[bluesman]

35 minutes ago, ShawnC said:

The other thing for me, it would seem weird to say things out loud while listening to music.  If I'm rocking out, I'd have to turn down the volume or yell into the room to change the song being played or add something to the queue. 

I’m amazed at how well and accurately Alexa hears me even with loud music or TV playing - and I don’t have to shout.  So with a bit more refinement in hardware and AI, I expect excellent responsiveness to be the norm.  
 

I have no problem playing a track selected by title, artist, version, edition etc with voice control of JRiver.  If I forget that I have multiple versions of the same tune by the same artist, the vox ex machina recites them and asks which I want to hear.
 

I also still prefer the buttons on our Comcast remote.  But my wife loves the voice control and uses it 100% of the time.  I guess she’s just used to voice control after 4+ decades of marriage 😝

[bluesman]

1 hour ago, nrcyclist said:

Personally, I do not need or want to control every button, knob or setting through voice control; unlikely that I would ever remember all the commands.

I walked into the library a few days ago to discover my wife in the middle of a major effort to get Alexa to turn on the lights.  Unfortunately, she forgot to say "Alexa" first 🤪

 

AI is the secret to fully functional voice control.  It's not hard for an intelligent computer to figure out that you want the lights on in the room when it's dark outside and you've asked for the same lights at the same time of day from the same spot 16 days in a row.  If your request is ambiguous, you can be presented with rational options, e.g. "Do you want all the lights on in this room or something else?", "Do you want the version of Kind of Blue that you play most often?" etc.

 

Then we'll be able to retire mechanical controls and failure-prone interfaces in favor of simple, unscripted voice interaction. I look forward to it!

[bluesman]

33 minutes ago, LarryMagoo said:

Roon with Voice control would just be icing on the music listening cake

The new crop of VC/AI programs will almost certainly be able to control Roon.  What's needed is the ability to send the same call to the processors in response to a voice command that's generated by clicking on a Roon icon. It's not as simple as that sounds, but it's doable today.  I've been playing with this using Braina, but I've not yet succeeded.

 

Siri is almost certainly capable of doing this now - there are many custom business applications out there resulting from licensed use of Siri technology.  But I suspect that Apple's not about to support any platforms at their own expense that don't augment their revenue stream beyond their costs.

[bluesman]

43 minutes ago, LarryMagoo said:

when I ask Siri a question how to "they" decide which device answers me..??

That's a great question!  One of these days, I'm going to start a conversation between Alexa and Siri to see where it goes.

[bluesman]

40 minutes ago, ASRMichael said:

it did work, long process to get just 10 albums, but was my favourite albums at the time

...and that's how progress is made.  It's easier now, and before long it'll be the status quo.  Check out the latest AI/VC efforts on line - they're really pretty cool, even though they're still add-ons.  Once some genius figures out how to integrate them into SoCs, we'll be in fat city!