gradio_ui.py

import gradio as gr
from app.asr_model import load_model, transcribe_audio
from app.language_detection import detect_language_from_text
from app.history import save_to_history, export_history, get_history, save_transcript_as_txt
import pandas as pd

def process_audio(audio_path):
    if audio_path is None:
        return "No audio uploaded.", "Unknown", None

    print(f"\n--- New Request ---")
    print(f"Processing audio: {audio_path}")

    print("Transcribing...")
    transcript = transcribe_audio(audio_path)
    print(f"Transcription complete: {transcript[:80]}")

    print("Detecting language...")
    lang = detect_language_from_text(transcript)

    print("Saving to history...")
    save_to_history(audio_path, transcript, lang)

    # Create a downloadable .txt file for the transcript
    txt_path = save_transcript_as_txt(transcript)

    print("Done!\n")
    return transcript, lang, gr.update(value=txt_path, visible=True)

def export_history_wrapper():
    path = export_history("csv")
    if path:
        return gr.update(value=path, visible=True)
    return gr.update(visible=False)

def load_history_table():
    """Load history and return as a list of lists for gr.Dataframe."""
    history = get_history()
    if not history:
        return []
    rows = []
    for entry in history:
        rows.append([
            entry.get("timestamp", "")[:19].replace("T", " "),
            entry.get("audio_file", ""),
            entry.get("language", ""),
            entry.get("transcript", "")
        ])
    return pd.DataFrame(rows, columns=["Timestamp", "Audio File", "Language", "Transcript"])

def create_ui():
    with gr.Blocks(title="Multilingual ASR") as demo:
        gr.Markdown("# 🎙️ Multilingual Automatic Speech Recognition")

        with gr.Tabs():
            with gr.TabItem("📝 Transcribe"):
                gr.Markdown("Upload an audio file to get a text transcription using Wav2Vec.")

                with gr.Row():
                    with gr.Column():
                        audio_input = gr.Audio(type="filepath", label="Upload Audio")
                        transcribe_btn = gr.Button("Transcribe", variant="primary")

                    with gr.Column():
                        lang_output = gr.Textbox(label="Detected Language")
                        transcript_output = gr.Textbox(label="Transcription", lines=10)
                        download_txt = gr.File(label="⬇️ Download Transcript (.txt)", visible=False)

                transcribe_btn.click(
                    fn=process_audio,
                    inputs=audio_input,
                    outputs=[transcript_output, lang_output, download_txt]
                )

            with gr.TabItem("📋 History"):
                gr.Markdown("Your past transcriptions (newest first).")

                with gr.Row():
                    refresh_btn = gr.Button("🔄 Refresh History")
                    export_btn = gr.Button("⬇️ Export as CSV")

                history_table = gr.Dataframe(
                    headers=["Timestamp", "Audio File", "Language", "Transcript"],
                    datatype=["str", "str", "str", "str"],
                    value=load_history_table,
                    wrap=True,
                    label="Transcription History"
                )
                csv_file_output = gr.File(label="Download CSV", visible=False)

                refresh_btn.click(fn=load_history_table, outputs=history_table)
                export_btn.click(fn=export_history_wrapper, outputs=csv_file_output)

    return demo

if __name__ == "__main__":
    demo = create_ui()
    demo.launch()