I wish to build a simple headless device, powered by a 9 V battery that allows a person to speak into it upon touch. The device would then capture to allow "speech to text" functionality and, subsequently, have that text sent over Wi-Fi, to my server.

My background is in Android development and I can handle the necessary programming, but I have very little background in the hardware side. In short, I do know I need a microphone, Wi-Fi transmitter, possibly Bluetooth to allow the device to link to mobile device for Wi-Fi login information, and bare bones microcontroller (way less than an Arduino given the specific functionality).

I am very new to this. Can such products be purchased for around $20 total? Or perhaps a unit with all features outlined above aside from the speech-to-text logic? Where?

  • You want to use the Arduino for speech-to-text? Or are you hoping to be able to implement the Google Speech API on it? – Majenko Mar 20 '16 at 11:21
  • An ATmega based arduino is not really sufficient even for capturing samples to submit for remote processing. Orders of magnitude too little buffer memory, and likely insufficient ADC. – Chris Stratton Mar 20 '16 at 14:27
  • @Majenko yes, I was aiming to implement a speech to text api onto a microcontroller, that would then send the text to my server over WiFi. – Sauron Mar 20 '16 at 14:32
  • I am still unclear: you want to sample onto the Arduino, convert the audio into the right form for the API to receive, send it to the API server, get the response back, interpret it, then send it on to your server? Or do you want to sample on the Arduino, do the converting to text on the Arduino and then send the results only to your server? No third party involved? Either way it's not really practical... – Majenko Mar 20 '16 at 14:34
  • @Majenko yes, the arduino can do the speech to text recognition. Why is that impractical? I have viewed several examples on YouTube doing the exact same thing. ... – Sauron Mar 20 '16 at 14:36