[App_rpt-users] Voice control options

Pierre Martel petem001 at gmail.com
Mon Jan 7 00:35:14 UTC 2019


Thanks again for the explaination William. I do understand the principle
behind the porcupine software.. And as you just told, to test its
fonctionality someone just need a mic and speaker that run on the linux
sound system.

The part that I need to know is how did you pass the ptt info to the irlp
board? I will try to run porcupine in a small VM.

Next I will need to find how to pipe the audio to and from the
Analog_bridge software to Picovoice suite.

That will be on the dvswitch mailling list as it has not much to do with
allstarlink.

I think I am getting somewhere. The planning in such type of project is
what makes it fun or a pain and frustration magnet ;-)




Le dim. 6 janv. 2019 à 18:11, William Franzin <wfranzin at gmail.com> a écrit :

> IRLP boards have been around for about 20 years and it's just a DTMF
> decoder wired to the parallel port and interfacing for PTT, COS, PL etc.
> You don't need anything fancy here. Anything that works with alsa audio in
> Linux will work fine. Control the PTT using a script and your choice of
> GPIO. I would run this outside of Asterisk (on the side) and then use it to
> tell Asterisk which commands I want to execute. Asterisk audio required
> transcoding. So there was zero hardware work required to add this to the
> IRLP node. Same would apply anywhere you've got a way to control the PTT
> output.
>
> If you want to try something today use a Raspberry Pi or Linux desktop
> (Ubuntu etc) and connect a USB headset up. You'll be able to speak and
> detect. Because this doesn't require COS or PL detect its just listening
> for the wakeword then listening to speech & executing commands. Just need
> PTT output. An EchoLink USB adapter is probably the easiest way to do this
> as the audio hardware is just a USB sound card and they also emulate a
> serial port for transmitter/receiver interfacing. That would be just a few
> lines of code needed to turn the serial lines on/off for PTT control.
>
> When you're using a wakeword to activate any of these voice services it's
> continuously listening so doesn't depend on COS or CTCSS. Dan VE4DRK and I
> were talking about this and he suggested listening for signals without the
> PL might be a nice way to keep the voice commands away from the audio path
> when connected to links. Program the same repeater frequency twice into
> your radio and one has a PL for normal repeater use and the other no PL.
> Then wakeword still works and repeater keys up with PL encode and nobody
> hears the voice commands at all, completely silent to links,
>
>
>
> On Sun, Jan 6, 2019 at 4:42 PM petem001 <petem001 at gmail.com> wrote:
>
>> Can I ask how you interfaced the audio from and to porcupine into the
>> irlp node?
>>
>> I am not very aware of the irlp interface.
>>
>>
>>
>> Envoyé depuis mon téléphone intelligent Samsung Galaxy.
>>
>> -------- Message d'origine --------
>> De : William Franzin <wfranzin at gmail.com>
>> Date : 19-01-06 12:31 p.m. (GMT-05:00)
>> À : Users of Asterisk app_rpt <app_rpt-users at lists.allstarlink.org>
>> Objet : Re: [App_rpt-users] Voice control options
>>
>> Hi Pierre
>>
>> This was just a fresh IRLP install for testing and nothing more. Did the
>> updates. Added the SDK examples and then a few lines changed for
>> transmitter control. Anything with audio input and output should work fine,
>> so could be an IRLP box with sound card, EchoLink USB adapter, DMK USB
>> Radio Interface, NWDR UDRC/DRAWS, or transcoded input/output from AMBE. The
>> commands back to the system are being done via IFTTT HTTPS webhook request
>> right now and just a script that accepts authenticated commands.
>>
>> I think you'll find it's fairly easy to build this functionality into
>> anything. All of the real work happens back in Google/Amazon's massive data
>> centers and these smart speaker devices are really just voice terminals.
>> They're just a small embedded platform like a Raspberry Pi with a
>> microphone array and speaker(s). On the device they're running wakeword
>> detection "Hey Google" | "Alexa" and then start streaming the audio back to
>> the data center for processing. Depending on the product (speaker, TV,
>> coffee maker) there could be code running there to handle the understood
>> command(s).
>>
>> If you want to see some progress today, I'd start with
>> https://github.com/Picovoice/Porcupine and get Porcupine up and running.
>> It also comes with everything you need to get started including wakeword
>> files for "bumblebee" and other words you won't hear on a repeater in
>> common speech. Later you can generate your own custom wakeword files. When
>> I started with Picovoice all my code did was understand a wakeword and
>> respond with a beep tone over the air.
>>
>> There's quite the debate over Google/Internet services being tied to
>> radio. I think the Picovoice Rhino is probably the best area to focus on
>> for on-device repeater control. It includes the Porcupine library for
>> wakeword detection so "OK Repeater" is possible and the intents (turn on
>> the link etc) are all handled within the controller platform, no internet.
>> That's probably where I'll focus my effort on - it will end up being a
>> hybrid system where the repeater controller has basic commands and I can
>> still say "OK Repeater, ask Google ..." and then we're using Google's web
>> services as a secondary method.
>>
>> One of the first voice apps I've been working on is just a radio check.
>> But rather than asking other people on a repeater "how does this radio
>> sound" I can use a SDR on the repeater receiver and determine if the user
>> requesting a check is on frequency, what kind of deviation, DTMF deviation,
>> signal level in db and all that stuff you can do with a $10 RTL SDR.
>> Getting those voice services up and running will open up a pile of new
>> things you can do over the air with analog/digital radios.
>>
>> Best regards
>> William
>>
>> On Sun, Jan 6, 2019 at 10:08 AM Pierre Martel <petem001 at gmail.com> wrote:
>>
>>> Thanks William for the feedback.
>>>
>>> I find your implementation on the IRLP controler to be a piece of art..
>>> ;-)
>>>
>>> Can I ask you what method did you used for the irlp controler? was it
>>> the original irlp software or did you used the irlp driver in app_rpt?
>>>
>>> did you implement a way to prevent the AI to respond to request sent by
>>> a networked audio feed?
>>>
>>> I am hyper excited by this and cant wait to test this.
>>>
>>> Pierre
>>> VE2PF
>>>
>>>
>>>
>>>
>>>
>>> Le dim. 6 janv. 2019 à 08:53, William Franzin <wfranzin at gmail.com> a
>>> écrit :
>>>
>>>> Good morning,
>>>>
>>>> We're also keen AllStarLink/app_rpt users/experimenters here in
>>>> Winnipeg and Dan mentioned a discussion about the work I did on here, so
>>>> thought I'd fill in the blanks. Also joined this list.
>>>>
>>>> I started with Google Assistant and Amazon Alexa and really just wanted
>>>> to control devices using voice, the "what's the weather" was just an added
>>>> bonus, but all that requires internet.
>>>>
>>>> I also liked the speed of running the assistant code in real-time, as
>>>> you speak that's actually a stream to Google/Amazon so the results you get
>>>> are immediate vs rec/upload/download/play.
>>>>
>>>> For the majority of ham radio actual "needs" in voice control the
>>>> Picovoice Rhino project speech-to-intent would do a great job. If you
>>>> haven't check it out: https://github.com/Picovoice
>>>>
>>>> Also watch their video: https://www.youtube.com/watch?v=WadKhfLyqTQ
>>>> which I think is exactly what we'd want from a ham radio setup. Wakework
>>>> "Hey Repeater" "Turn on ..".
>>>>
>>>> All of that great capability runs entirely on the
>>>> device/server/reflector, no internet or web service call to Google/Amazon
>>>> is needed. And it produces this simple output too:
>>>>
>>>> {
>>>>   "type": "espresso",
>>>>   "size": "small",
>>>>   "numberOfShots": "2",
>>>>   "sugar": "a lot",
>>>>   "milk": "some"
>>>> }
>>>>
>>>>
>>>> We should do this, my 2010 Ford Escape has had Microsoft Sync voice
>>>> control and it's nothing new. Everything has voice control these days, and
>>>> it actually works really well now.
>>>>
>>>> The only challenge in ham is we need to come up with some standard
>>>> commands so if you're visiting here you know the repeater has this
>>>> capability and how to work it.
>>>>
>>>> If anything I'm just glad the videos I posted a while back got this
>>>> topic going again. Certainly not the first person to poke around with voice
>>>> - it's just really easy now to implement this.
>>>>
>>>> Thanks to all the folks working on AllStarLink/app_rpt too. I just
>>>> built a firmware image with AllStarLink for a Sierra Radio Systems
>>>> CommServer board and it was apt-get-easy
>>>> --let-me-drink-my-coffee-while-the-hard-work-gets-done ;)
>>>>
>>>> Thanks,
>>>> William VE4VR
>>>> _______________________________________________
>>>> App_rpt-users mailing list
>>>> App_rpt-users at lists.allstarlink.org
>>>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users
>>>>
>>>> To unsubscribe from this list please visit
>>>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users
>>>> and scroll down to the bottom of the page. Enter your email address and
>>>> press the "Unsubscribe or edit options button"
>>>> You do not need a password to unsubscribe, you can do it via email
>>>> confirmation. If you have trouble unsubscribing, please send a message to
>>>> the list detailing the problem.
>>>>
>>> _______________________________________________
>>> App_rpt-users mailing list
>>> App_rpt-users at lists.allstarlink.org
>>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users
>>>
>>> To unsubscribe from this list please visit
>>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users and
>>> scroll down to the bottom of the page. Enter your email address and press
>>> the "Unsubscribe or edit options button"
>>> You do not need a password to unsubscribe, you can do it via email
>>> confirmation. If you have trouble unsubscribing, please send a message to
>>> the list detailing the problem.
>>>
>> _______________________________________________
>> App_rpt-users mailing list
>> App_rpt-users at lists.allstarlink.org
>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users
>>
>> To unsubscribe from this list please visit
>> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users and
>> scroll down to the bottom of the page. Enter your email address and press
>> the "Unsubscribe or edit options button"
>> You do not need a password to unsubscribe, you can do it via email
>> confirmation. If you have trouble unsubscribing, please send a message to
>> the list detailing the problem.
>>
> _______________________________________________
> App_rpt-users mailing list
> App_rpt-users at lists.allstarlink.org
> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users
>
> To unsubscribe from this list please visit
> http://lists.allstarlink.org/cgi-bin/mailman/listinfo/app_rpt-users and
> scroll down to the bottom of the page. Enter your email address and press
> the "Unsubscribe or edit options button"
> You do not need a password to unsubscribe, you can do it via email
> confirmation. If you have trouble unsubscribing, please send a message to
> the list detailing the problem.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.keekles.org/pipermail/app_rpt-users/attachments/20190106/6268c1ff/attachment.html>


More information about the App_rpt-users mailing list