app.py

import streamlit as st
from transformers import pipeline
import numpy as np
import soundfile as sf
from io import BytesIO
from stt_audiorec import stt_audiorec

# Load Whisper model
@st.cache_resource
def load_model():
    return pipeline("automatic-speech-recognition", model="openai/whisper-small")

st.title("Text Entry with Voice Input")
st.write("Enter text manually or use voice input:")

# Initialize session state
if 'combined_text' not in st.session_state:
    st.session_state.combined_text = ""

# Create columns layout
col1, col2 = st.columns(2)

with col1:
    # Text input
    text_input = st.text_area("Type your text here:", height=200)

with col2:
    # Audio input
    st.write("Record your voice:")
    audio_bytes = stt_audiorec()

# Process audio when recording is available
if audio_bytes:
    try:
        # Convert bytes to audio array
        with BytesIO(audio_bytes) as audio_file:
            audio_data, sample_rate = sf.read(audio_file)
            
        # Convert stereo to mono if needed
        if len(audio_data.shape) > 1:
            audio_data = np.mean(audio_data, axis=1)
            
        # Create input for Whisper
        audio_dict = {"raw": audio_data, "sampling_rate": sample_rate}
        
        # Transcribe audio
        whisper = load_model()
        transcribed_text = whisper(audio_dict)["text"]
        
        # Update session state
        st.session_state.combined_text = f"{text_input}\n{transcribed_text}".strip()
        
    except Exception as e:
        st.error(f"Error processing audio: {str(e)}")

# Combine inputs when button is clicked
if st.button("Submit"):
    if not text_input and not audio_bytes:
        st.warning("Please enter text or record audio")
    else:
        # Display combined text
        st.subheader("Combined Input:")
        st.write(st.session_state.combined_text)

        # Add download button
        st.download_button(
            label="Download Text",
            data=st.session_state.combined_text,
            file_name="combined_input.txt",
            mime="text/plain"
        )