Update app.py

app.py

@@ -27,33 +27,25 @@ def speech_to_speech_translation(audio_data, tgt_lang):
     torchaudio.save(output_file, waveform.unsqueeze(0), sample_rate=16000)
     return text, output_file
 
-# Gradio interfaces
-iface_s2t = gr.Interface(
-    fn=speech_to_text,
-    inputs=[
-        gr.Audio(label="Upload or Record Audio for Speech to Text"),
-        gr.Dropdown(list(languages.keys()), label="Select Target Language")
-    ],
-    outputs="text",
-    title="Speech to Text"
-)
-
-iface_s2st = gr.Interface(
-    fn=speech_to_speech_translation,
-    inputs=[
-        gr.Audio(label="Upload or Record Audio for Speech to Speech Translation"),
-        gr.Dropdown(list(languages.keys()), label="Select Target Language")
-    ],
-    outputs=["text", "audio"],
-    title="Speech to Speech Translation"
-)
-
-# Combine into an accordion interface
-accordion = gr.Accordion(
-    iface_s2t,
-    iface_s2st,
-    labels=["Speech to Text", "Speech to Speech Translation"]
-)
-
-# Launch the application
-accordion.launch()
+def create_interface():
+    with gr.Blocks(theme='ParityError/Anime') as interface:
+        # Dropdown for language selection
+        input_language = gr.Dropdown(list(languages.keys()), label="Select Target Language", value="English")
+
+        with gr.Accordion("Speech to Text", open=False) as stt_accordion:
+            audio_input_stt = gr.Audio(label="Upload or Record Audio")
+            text_output_stt = gr.Text(label="Transcribed Text")
+            stt_button = gr.Button("Transcribe")
+            stt_button.click(speech_to_text, inputs=[audio_input_stt, input_language], outputs=text_output_stt)
+
+        with gr.Accordion("Speech to Speech Translation", open=False) as s2st_accordion:
+            audio_input_s2st = gr.Audio(label="Upload or Record Audio")
+            text_output_s2st = gr.Text(label="Translated Text")
+            audio_output_s2st = gr.Audio(label="Translated Audio", type="filepath")
+            s2st_button = gr.Button("Translate")
+            s2st_button.click(speech_to_speech_translation, inputs=[audio_input_s2st, input_language], outputs=[text_output_s2st, audio_output_s2st])
+
+    return interface
+
+app = create_interface()
+app.launch(show_error=True, debug=True)