Pre translate audio. Code is writen but fails, so that i commented it

translatube.py

@@ -5,10 +5,16 @@ from pytube import YouTube
 import re
 import subprocess
 from lang_list import ORIGINAL_LANGUAGE_NAME_TO_CODE, S2ST_TARGET_ORIGINAL_LANGUAGE_NAMES
+import torch
+from seamless_communication.models.inference import Translator
+import time
 
 YOUTUBE = "youtube"
 TWITCH = "twitch"
 
+# Initialize a Translator object with a multitask model, vocoder on the GPU.
+# translator = Translator("seamlessM4T_large", vocoder_name_or_card="vocoder_36langs", device=torch.device("cuda:0"))
+
 def copy_url_from_clipboard():
     return pyperclip.paste()
 
@@ -30,7 +36,7 @@ def get_youtube_video_id(url):
 
 def is_valid_url(url):
     button = gr.Button(size="sm", value="translate", min_width="10px", scale=0, visible=True)
-    original_audio = gr.Audio(label="Original audio", elem_id="original_audio", visible=True)
+    original_audio = gr.Audio(label="Original audio", elem_id="original_audio", visible=True, interactive=False)
     translated_audio = gr.Audio(label="Translated audio", elem_id="translated_audio", visible=True)
     source_languaje = gr.Dropdown(visible=True, label="Source languaje", show_label=True, value="English", choices=ORIGINAL_LANGUAGE_NAME_TO_CODE, scale=1, interactive=True)
     target_languaje = gr.Dropdown(visible=True, label="Target languaje", show_label=True, value="Español", choices=S2ST_TARGET_ORIGINAL_LANGUAGE_NAMES, scale=1, interactive=True)
@@ -48,7 +54,7 @@ def is_valid_url(url):
                 )
     elif "twitch" in url.lower() or "twitch.tv" in url.lower():
         return (
-            gr.Image(value="assets/twitch.webp", visible=True), 
+            gr.Image(value="assets/twitch.webp", visible=True, show_download_button=False, container=False), 
             source_languaje,
             target_languaje,
             button, 
@@ -84,6 +90,13 @@ def get_audio_from_video(url, stream_page):
 
         return gr.Audio("audio.mkv", label="Original audio", elem_id="original_audio", visible=True)
 
+# def translate_audio(input_audio, target_languaje):
+#     print("Translating audio...")
+#     time.sleep(5)
+#     print("Translating audio...")
+#     _, wav, _ = translator.predict(input_audio, "s2st", target_languaje)
+#     return gr.Audio(wav, label="Translated audio", elem_id="translated_audio", visible=True)
+
 
 with gr.Blocks() as demo:
     with gr.Row(variant="panel"):
@@ -107,5 +120,6 @@ with gr.Blocks() as demo:
     translated_audio = gr.Audio(label="Translated audio", elem_id="translated_audio", visible=visible)
     url_textbox.change(fn=is_valid_url, inputs=url_textbox, outputs=[image, source_languaje, target_languaje, translate_button, stream_page, original_audio, translated_audio])
     translate_button.click(fn=get_audio_from_video, inputs=[url_textbox, stream_page], outputs=original_audio)
+    # original_audio.change(fn=translate_audio, inputs=[original_audio, target_languaje], outputs=translated_audio)
 
 demo.launch()