app.py

import gradio as gr
import torch
import torchaudio
import numpy as np
from transformers import AutoProcessor, SeamlessM4Tv2Model
from datetime import datetime

class SeamlessTranslator:
    def __init__(self):
        self.model_name = "facebook/seamless-m4t-v2-large"
        print("Loading model...")
        self.processor = AutoProcessor.from_pretrained(self.model_name)
        self.model = SeamlessM4Tv2Model.from_pretrained(self.model_name)
        self.sample_rate = self.model.config.sampling_rate

        self.languages = {
            "🇺🇸 English": "eng",
            "🇪🇸 Spanish": "spa",
            "🇫🇷 French": "fra",
            "🇩🇪 German": "deu",
            "🇮🇹 Italian": "ita",
            "🇵🇹 Portuguese": "por",
            "🇷🇺 Russian": "rus",
            "🇨🇳 Chinese": "cmn",
            "🇯🇵 Japanese": "jpn",
            "🇰🇷 Korean": "kor"
        }

    def translate_text(self, text, src_lang, tgt_lang, progress=gr.Progress()):
        progress(0.3, desc="Processing input...")
        try:
            inputs = self.processor(text=text, src_lang=self.languages[src_lang], return_tensors="pt")
            progress(0.6, desc="Generating audio...")
            audio_array = self.model.generate(**inputs, tgt_lang=self.languages[tgt_lang])[0].cpu().numpy().squeeze()
            progress(1.0, desc="Done!")
            return (self.sample_rate, audio_array), f"✅ Translation completed: {src_lang} → {tgt_lang}"
        except Exception as e:
            raise gr.Error(f"❌ Translation failed: {str(e)}")

    def translate_audio(self, audio_path, tgt_lang, progress=gr.Progress()):
        if audio_path is None:
            raise gr.Error("❌ Please upload an audio file")
            
        progress(0.3, desc="Loading audio...")
        try:
            audio, orig_freq = torchaudio.load(audio_path)
            audio = torchaudio.functional.resample(audio, orig_freq=orig_freq, new_freq=16000)
            
            progress(0.6, desc="Translating...")
            inputs = self.processor(audios=audio, return_tensors="pt")
            audio_array = self.model.generate(**inputs, tgt_lang=self.languages[tgt_lang])[0].cpu().numpy().squeeze()
            progress(1.0, desc="Done!")
            return (self.sample_rate, audio_array), "✅ Audio translation completed"
        except Exception as e:
            raise gr.Error(f"❌ Translation failed: {str(e)}")

css = """
.gradio-container {
    max-width: 1200px !important;
    margin: auto !important;
}

.main-header {
    text-align: center;
    margin-bottom: 2rem;
    padding: 2rem;
    background: linear-gradient(135deg, #1e40af, #3b82f6);
    border-radius: 12px;
    color: white;
}

.main-title {
    font-size: 2.5rem;
    font-weight: bold;
    margin-bottom: 0.5rem;
}

.main-subtitle {
    font-size: 1.2rem;
    opacity: 0.9;
}

.container {
    padding: 1.5rem;
    border-radius: 12px;
    background: white;
    box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
    margin-bottom: 1.5rem;
}

.status-box {
    padding: 1rem;
    border-radius: 8px;
    background: #f0f9ff;
    border-left: 4px solid #3b82f6;
    margin-top: 1rem;
}

.footer {
    text-align: center;
    margin-top: 2rem;
    padding: 1rem;
    color: #666;
}
"""

def create_ui():
    translator = SeamlessTranslator()

    with gr.Blocks(css=css, title="A.R.I.S. Translator") as demo:
        gr.HTML(
            """
            <div class="main-header">
                <div class="main-title">A.R.I.S. Translator</div>
                <div class="main-subtitle">Advanced Real-time Interpretation System</div>
            </div>
            """
        )

        with gr.Tabs():
            # Text Translation Tab
            with gr.Tab("🔤 Text Translation"):
                with gr.Row():
                    with gr.Column():
                        text_input = gr.Textbox(
                            label="Text to Translate",
                            placeholder="Enter your text here...",
                            lines=5
                        )
                        with gr.Row():
                            src_lang = gr.Dropdown(
                                choices=list(translator.languages.keys()),
                                value="🇺🇸 English",
                                label="Source Language"
                            )
                            tgt_lang = gr.Dropdown(
                                choices=list(translator.languages.keys()),
                                value="🇪🇸 Spanish",
                                label="Target Language"
                            )
                        translate_btn = gr.Button("🔄 Translate", variant="primary")
                        status_text = gr.Textbox(
                            label="Status",
                            interactive=False
                        )

                    with gr.Column():
                        audio_output = gr.Audio(
                            label="Translation Output",
                            type="numpy"
                        )

            # Audio Translation Tab
            with gr.Tab("🎤 Audio Translation"):
                with gr.Row():
                    with gr.Column():
                        audio_input = gr.Audio(
                            label="Upload Audio",
                            type="filepath"
                        )
                        tgt_lang_audio = gr.Dropdown(
                            choices=list(translator.languages.keys()),
                            value="🇺🇸 English",
                            label="Target Language"
                        )
                        translate_audio_btn = gr.Button("🔄 Translate Audio", variant="primary")
                        status_text_audio = gr.Textbox(
                            label="Status",
                            interactive=False
                        )

                    with gr.Column():
                        audio_output_from_audio = gr.Audio(
                            label="Translation Output",
                            type="numpy"
                        )

        gr.HTML(
            """
            <div class="footer">
                Powered by Meta's SeamlessM4T model | Built with Gradio
            </div>
            """
        )

        # Event handlers
        translate_btn.click(
            fn=translator.translate_text,
            inputs=[text_input, src_lang, tgt_lang],
            outputs=[audio_output, status_text]
        )

        translate_audio_btn.click(
            fn=translator.translate_audio,
            inputs=[audio_input, tgt_lang_audio],
            outputs=[audio_output_from_audio, status_text_audio]
        )

    return demo

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