changes in ui

app.py

@@ -1,58 +1,47 @@
 from transformers import pipeline
 import gradio as gr
-from youtube_dl import YoutubeDL
+from pytube import YouTube
 
 
-# Function to download audio from YouTube link
-# def download_audio(youtube_link, output_path):
-#     ydl_opts = {
-#         "format": "bestaudio/best",
-#         "outtmpl": output_path,
-#         "postprocessors": [
-#             {
-#                 "key": "FFmpegExtractAudio",
-#                 "preferredcodec": "mp3",
-#                 "preferredquality": "192",
-#             }
-#         ],
-#     }
-#     with YoutubeDL(ydl_opts) as ydl:
-#         ydl.download([youtube_link])
+def dl_youtube_audio(youtube_link):
+    yt = YouTube(youtube_link)
+    audio_stream = yt.streams.filter(only_audio=True).first()
+    audio_path = f"youtube_audio_{yt.video_id}.mp3"
+    audio_stream.download(output_path=".", filename="youtube_audio")
+    return audio_path
 
 
-# def update_visibility(interface, input_type):
-#     if input_type == "audio":
-#         interface.set_visibility("audio_input", True)
-#         interface.set_visibility("youtube_input", False)
-#     elif input_type == "youtube":
-#         interface.set_visibility("audio_input", False)
-#         interface.set_visibility("youtube_input", True)
-def transcribe_audio(input_type, audio_path):
-    if input_type == "Audio":
-        model = pipeline(model="SofiaK/checkpoints")
-        return model(audio_path)["text"]
+def transcribe_audio(audio_path):
+    model = pipeline(model="SofiaK/checkpoints")
+    return model(audio_path)["text"]
 
 
 with gr.Blocks() as demo:
-    radio = gr.Radio(
-        choices=["Audio", "Youtube"], label="Choose your input type", value="Audio"
-    )
-    audio_input = gr.Audio(
-        sources=["upload", "microphone"],
-        type="filepath",
-        label="Upload Audio, or speak in the microphone",
-        visible=True,
-        interactive=True,
-    )
-    youtube_input = gr.Textbox(
-        value="https://www.youtube.com/",
-        label="Youtube Link",
-        visible=False,
-        interactive=True,
-    )
-    btn = gr.Button(
-        "Transcript",
-    )
+    with gr.Row():
+        with gr.Column():
+            radio = gr.Radio(
+                choices=["Audio", "Youtube"],
+                label="Choose your input type",
+                value="Audio",
+            )
+            audio_input = gr.Audio(
+                sources=["upload", "microphone"],
+                type="filepath",
+                label="Upload Audio, or speak in the microphone",
+                visible=True,
+                interactive=True,
+            )
+            youtube_input = gr.Textbox(
+                value="https://www.youtube.com/",
+                label="Youtube Link",
+                visible=False,
+                interactive=True,
+            )
+            btn = gr.Button(
+                "Transcript",
+            )
+        with gr.Column():
+            output = gr.Text(label="Model output")
 
     def make_visible(val):
         audio_visible = val == "Audio"
@@ -63,7 +52,15 @@ with gr.Blocks() as demo:
 
     radio.change(make_visible, inputs=radio, outputs=[audio_input, youtube_input])
 
-    output = gr.Text(label="Model output")
-    btn.click(fn=transcribe_audio, inputs=[radio, audio_input], outputs=output)
+    def on_button_click():
+        if radio.value == "Audio":
+            audio_path = audio_input.value
+        else:
+            audio_path = dl_youtube_audio(youtube_input.value)
+        output.value = transcribe_audio(audio_path)
+
+    btn.click(
+        fn=on_button_click, inputs=[radio, audio_input, youtube_input], outputs=output
+    )
 
 demo.launch()

requirements.txt

@@ -1,4 +1,5 @@
 transformers
 gradio
 torch
-youtube_dl
+youtube_dl
+pytube