Create app.py

app.py

@@ -0,0 +1,61 @@
+import gradio as gr
+from transformers import pipeline
+import torch
+
+# Load lightweight models
+transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
+translator = pipeline("translation", model="facebook/m2m100_418M")
+tts = pipeline("text-to-speech", model="espnet/kan-bayashi_ljspeech_vits", device=torch.device('cpu'))
+
+# Function to handle transcription, translation, and TTS
+def process_audio(audio, target_lang):
+    if audio is None:
+        return "Please upload an audio file.", None, None
+
+    try:
+        # Transcribe
+        transcription = transcriber(audio["name"])['text']
+
+        # Translate (if target language is different from source)
+        if target_lang != "en":
+            translated_text = translator(transcription, src_lang="en", tgt_lang=target_lang)[0]['translation_text']
+        else:
+            translated_text = transcription
+
+        # Generate speech from translated text
+        audio_output = tts(translated_text)
+
+        # Create Braille-compatible text file
+        braille_file = "output.brf"
+        with open(braille_file, "w", encoding="utf-8") as f:
+            f.write(translated_text)
+
+        return translated_text, (audio_output, "output.wav"), braille_file
+
+    except Exception as e:
+        return f"Error: {str(e)}", None, None
+
+# Define UI
+inputs = [
+    gr.Audio(type="filepath"),
+    gr.Dropdown(choices=["hi", "te", "kn", "en", "hu"], label="Target Language")
+]
+
+outputs = [
+    gr.Text(label="Translated Text"),
+    gr.Audio(label="Generated Speech"),
+    gr.File(label="Download Braille-Compatible File")
+]
+
+# Create Gradio app
+app = gr.Interface(
+    fn=process_audio,
+    inputs=inputs,
+    outputs=outputs,
+    title="Multi-Language Voice Translator",
+    description="Transcribe, translate, and generate speech in multiple languages with accessibility features."
+)
+
+# Launch app
+if __name__ == "__main__":
+    app.launch()