app.py

import streamlit as st
from pocketsphinx import LiveSpeech, get_model_path
import os
import sounddevice as sd
import numpy as np
import scipy.io.wavfile as wav

# Function to capture audio from the mic
def record_audio(filename='temp.wav', duration=5, fs=16000):
    st.text("Recording Audio...")
    with st.spinner(f'Recording for {duration} seconds...'):
        myrecording = sd.rec(int(duration * fs), samplerate=fs, channels=1)
        sd.wait()  # Waits until recording is finished
        wav.write(filename, fs, myrecording)  # Save as WAV file 

# Get the model path for pocketsphinx
model_path = get_model_path()
config = {
    'verbose': False,
    'hmm': os.path.join(model_path, 'en-us'),
    'lm': os.path.join(model_path, 'en-us.lm.bin'),
    'dict': os.path.join(model_path, 'cmudict-en-us.dict')
}

# Streamlit UI
st.title("Simple Speech Recognition with Streamlit and PocketSphinx")
button = st.button("Press to Speak")

# Store the state of the recording
if 'recording_done' not in st.session_state:
    st.session_state.recording_done = False

# When the button is pressed
if button:
    # Record the audio
    record_audio()
    st.session_state.recording_done = True

# If an audio was recorded, process it with PocketSphinx
if st.session_state.recording_done:
    audio = LiveSpeech(**config)

    st.text("Processing Audio...")
    with st.spinner('Recognizing...'):
        for phrase in audio:
            st.write(phrase)
            break  # We'll stop after the first phrase

    # Reset the state
    st.session_state.recording_done = False