DevHeads IoT Integration Server•3mo ago

How do I achieve a fully functional voice-controlled system with reliable command recognition

How do i achieve a fully functional voice-controlled system with reliable command recognition and LED control? Am done with my project on developing a home automation voice-controlled system using MicroPython and TinyML on an ESP32 microcontroller to recognize specific voice commands ("turn on" and "turn off") for controlling an LED. It runs smoothly as intended with no errors but sometimes doesn't respond to the voice command, how can i achieve consistent voice recognition and LED control with the ESP32, leveraging TinyML and MicroPython.

1 Reply

Enthernet Code•3mo ago

from machine import Pin, I2S
import time
import ustruct
import array
import tensorflow as tf

led = Pin(2, Pin.OUT)

i2s = I2S(0,
          sck=Pin(14),
          ws=Pin(15),
          sd=Pin(32),
          mode=I2S.RX,
          bits=16,
          format=I2S.MONO,
          rate=16000,
          ibuf=2000)

model = tf.lite.Interpreter(model_path="model.tflite")
model.allocate_tensors()

input_details = model.get_input_details()
output_details = model.get_output_details()

def read_audio():
    audio_data = array.array('h', i2s.read(1024))
    audio_samples = ustruct.unpack('<1024h', audio_data)
    return audio_samples

def final_process_audio():
    audio_samples = read_audio()
    audio_input = tf.convert_to_tensor(audio_samples, dtype=tf.float32)
    audio_input = tf.expand_dims(audio_input, axis=0)

    model.set_tensor(input_details[0]['index'], audio_input)
    model.invoke()

    output_data = model.get_tensor(output_details[0]['index'])
    command_index = int(output_data.argmax())

    commands = ["turn off", "turn on"]
    if command_index < len(commands):
        return commands[command_index]
    else:
        return "unknown"

while True:
    command = final_process_audio()
    print(f"Recognized command: {command}")
    
    if command == "turn on":
        led.on()
        print("LED successfully turned on.")
    elif command == "turn off":
        led.off()
        print("LED successfully turned off.")
    else:
        print("Unknown command received.")

    time.sleep(1)

from machine import Pin, I2S
import time
import ustruct
import array
import tensorflow as tf

led = Pin(2, Pin.OUT)

i2s = I2S(0,
          sck=Pin(14),
          ws=Pin(15),
          sd=Pin(32),
          mode=I2S.RX,
          bits=16,
          format=I2S.MONO,
          rate=16000,
          ibuf=2000)

model = tf.lite.Interpreter(model_path="model.tflite")
model.allocate_tensors()

input_details = model.get_input_details()
output_details = model.get_output_details()

def read_audio():
    audio_data = array.array('h', i2s.read(1024))
    audio_samples = ustruct.unpack('<1024h', audio_data)
    return audio_samples

def final_process_audio():
    audio_samples = read_audio()
    audio_input = tf.convert_to_tensor(audio_samples, dtype=tf.float32)
    audio_input = tf.expand_dims(audio_input, axis=0)

    model.set_tensor(input_details[0]['index'], audio_input)
    model.invoke()

    output_data = model.get_tensor(output_details[0]['index'])
    command_index = int(output_data.argmax())

    commands = ["turn off", "turn on"]
    if command_index < len(commands):
        return commands[command_index]
    else:
        return "unknown"

while True:
    command = final_process_audio()
    print(f"Recognized command: {command}")
    
    if command == "turn on":
        led.on()
        print("LED successfully turned on.")
    elif command == "turn off":
        led.off()
        print("LED successfully turned off.")
    else:
        print("Unknown command received.")

    time.sleep(1)

Gaming

Programming

How do I achieve a fully functional voice-controlled system with reliable command recognition