app.py

import streamlit as st
import whisper
import tempfile

# Function to transcribe audio and detect language
def transcribe_and_detect_language(audio_file):
    model = whisper.load_model("base").to("cpu").float()  # Ensure model is in full precision

    # Load and process audio
    audio = whisper.load_audio(audio_file)
    audio = whisper.pad_or_trim(audio)

    # Convert to log-Mel spectrogram in full precision
    mel = whisper.log_mel_spectrogram(audio).to(model.device).float()  # Convert to float32

    # Detect the spoken language
    _, probs = model.detect_language(mel)
    detected_language = max(probs, key=probs.get)

    # Decode the audio
    options = whisper.DecodingOptions()
    result = whisper.decode(model, mel, options)

    return detected_language, result.text

# Streamlit UI
st.title("Speech to Text with Whisper")

# File uploader widget
uploaded_file = st.file_uploader("Upload an audio file", type=['wav', 'mp3'])

if uploaded_file is not None:
    with tempfile.NamedTemporaryFile(delete=False) as tmp_file:
        tmp_file.write(uploaded_file.getvalue())
        with st.spinner('Processing...'):
            language, transcribed_text = transcribe_and_detect_language(tmp_file.name)
        
        st.write(f"Detected language: {language}")
        st.text_area("Transcribed Text:", value=transcribed_text, height=300)