Voice recognition-Artificial neural networks

Thread Starter


Joined Mar 2, 2015
How it can be improvedDoes anyone want to try to improve voice recognition algorithms?

I imagine I come and talk with the objects in the house. Android has voice recognition but not much used.

Do you think we can make voice recognition more accurate ?


I wnat-
1. Let us agree on the algorithm used in voice recognition
2. Let us agree on the algorithm used in voice recognition
3. How it can be improved

I want to get an algorithm to implement on a controller.
I know there is Voice recognition from MikroC, Microchip pic30f, hm2007

Thread Starter


Joined Mar 2, 2015
From My Mail- This is the answer from MikroC

Forward Mail

On Nov 27, 2013 7:14 PM.
could you please guide me, like give some steps to make like a circuit.

On Thu, Nov 28, 2013 at 3:06 PM
You start first small amphitheater with microphone.After shrill ( nose filter input) after frquency converter to bits or adc converter .after micropreccessur do bits samples ++++ after / for
average numbers.

On Thu, Nov 28, 2013 at 10:37 PM
You select one audio circuit ( sensitive circuit) after you add voice circuit out bass filter or you do bass out (voice circuit) .After you do measure voice wave with adc convertor Ic or my methot voice convert to bits.After you add bits 10 parts.Thats ok.

They mean that a 10-bit A / D converter is good, right?

Thread Starter


Joined Mar 2, 2015
Tue, Dec 2, 2014 at 3:37 PM

Source of the SpeakUp firmware is not open.
We basically made whole speech recognition library based on DTW algorithm.
Functionality of this algorithm is easy to understand but in order to write the whole firmware (library) from scratch
you need to have advanced knowledge about very speech recognition (and speech recognition algorithms).

With SpeakUp click we provided free PC application which can be used for easier manipulation of the SpeakUp click.

Best regards,

The bolding of the quote belongs to me.

Does anyone want to explain this algorithm?