Author Topic: ESP8266 Speech Recognition System  (Read 3092 times)

0 Members and 1 Guest are viewing this topic.

Offline skillz21Topic starter

  • Frequent Contributor
  • **
  • Posts: 311
  • Country: au
ESP8266 Speech Recognition System
« on: July 02, 2018, 09:40:09 am »
I tend to use headphones quite a lot, because of this, I usually can't hear when my parents want something and call out my name. I am also usually in my room, which is a decent distance from the main living area of the house. I have this plan, and I just want to confirm whether it's plausible or not. My idea was to get an ESP8266, connect a small microphone to it, place it in the living area, then run that audio through some sort of speech recognition service (maybe google cloud). Then, once that service recognises my name, it would send that back, and I would get a notification on my computer. My initial idea was to just use my computer microphone, but the sound level was too low, and the noise too high once I turned up the gain.

Anyway, so I just wanted to ask whether this was possible. I can see two ways of doing this. One, getting the ESP to directly send the audio to a Google server then once a response has been received, alert my computer. Or, two, send the raw audio data directly to my computer, then have all the programming and stuff on my computer.

I'm asking because I have no previous experience with an ESP8266. (I picked that chip because it has network connectivity, and it is dirt cheap). Is it even possible to use this in such a way? Is it even possible to send audio like that to my computer? I personally like option number two, because the ESP programming seems really straightforward, (and also it's a lot easier to find support for programming on a computer)

Also, how do you program an ESP8266? Is it possible for me to just use the Arduino IDE or no? If not, I would probably prefer option two as well, since it's easier for me to use a programming language I'm comfortable with.

Looking back on this, it seems like a pretty stupid solution, but meh, I just wanna have some fun, and with the school holidays coming up next week this should keep me occupied for a while at least ;)
 

Offline skillz21Topic starter

  • Frequent Contributor
  • **
  • Posts: 311
  • Country: au
Re: ESP8266 Speech Recognition System
« Reply #1 on: July 02, 2018, 10:55:24 pm »
Can someone at least tell me whether this can work or not?

Sent from my SM-G900I using Tapatalk

 

Offline MK14

  • Super Contributor
  • ***
  • Posts: 4690
  • Country: gb
Re: ESP8266 Speech Recognition System
« Reply #2 on: July 03, 2018, 02:31:39 am »
One thing to bare in mind.
The google speech recognition service, is not really free, it just seems at a quick glance to be.

My understanding is that you always have to be signed in to a google account to use the service, and it limits the free part of its usage to something like, a maximum of one hour per month.
After that one hour (free), you have to pay a modest amount, which soon adds up to be a large amount of money.

https://cloud.google.com/speech-to-text/pricing

The $ seems small per 15 seconds, but if you turn it into $'s per hour, it becomes much more significant.

So "$0.006 USD / 15 seconds" = $1.44/hour, so for a month at 100 hours use (estimated) = around $144!
« Last Edit: July 03, 2018, 02:33:58 am by MK14 »
 

Offline MK14

  • Super Contributor
  • ***
  • Posts: 4690
  • Country: gb
Re: ESP8266 Speech Recognition System
« Reply #3 on: July 03, 2018, 02:46:02 am »
If I had, that problem to solve.
I'd probably buy a wireless doorbell and give the other party the front door button, to press when they want my attention. Ideally the type of wireless door bell which is very loud and/or has a big visual indication of when activated.

Alternatively, you could hook up something similar, by using a pair of your ESP8266, or similar parts.

You might be able to find some kind of free speech recognition system from somewhere, or maybe limit your usage to less than one hour per month, by only accessing the service, when loud shouting for someone, is occurring.
 

Offline MK14

  • Super Contributor
  • ***
  • Posts: 4690
  • Country: gb
Re: ESP8266 Speech Recognition System
« Reply #4 on: July 03, 2018, 02:51:27 am »
Also, google do this kit, which might give you a head start.

Google AIY Voice Kit
Product Code: 2408

You can probably buy it from other places. Here is one of them:

https://coolcomponents.co.uk/products/google-aiy-voice-kit?utm_medium=cpc&utm_source=googlepla&variant=864887734286&gclid=EAIaIQobChMImeSlzPWB3AIVop3tCh2SsQGrEAYYASABEgJ1LvD_BwE


 

Offline skillz21Topic starter

  • Frequent Contributor
  • **
  • Posts: 311
  • Country: au
Re: ESP8266 Speech Recognition System
« Reply #5 on: July 03, 2018, 03:25:08 am »
But what about my phone? So when I use Google assistant when I'm offline, it still functions... Are there any other offline APIs or something I could get as a software package?
 

Offline skillz21Topic starter

  • Frequent Contributor
  • **
  • Posts: 311
  • Country: au
Re: ESP8266 Speech Recognition System
« Reply #6 on: July 03, 2018, 03:28:54 am »
If I had, that problem to solve.
I'd probably buy a wireless doorbell and give the other party the front door button, to press when they want my attention. Ideally the type of wireless door bell which is very loud and/or has a big visual indication of when activated.

Alternatively, you could hook up something similar, by using a pair of your ESP8266, or similar parts.

You might be able to find some kind of free speech recognition system from somewhere, or maybe limit your usage to less than one hour per month, by only accessing the service, when loud shouting for someone, is occurring.
My intention was to have some fun while learning a bit about using an ESP8266. I could use a doorbell, and it would work, but that wasn't my intention.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf