check changelog for new update details

app.py

@@ -1,4 +1,4 @@
-import gradio as gr, glob, os, auditok, zipfile, wave, pytube.exceptions, librosa
+import gradio as gr, glob, os, auditok, zipfile, wave, pytube.exceptions, librosa, time
 from pytube import YouTube
 from moviepy.editor import VideoFileClip
 
@@ -146,6 +146,62 @@ def get_average_pitch(audio_file):
         mean_pitch = pitches.mean()
         return f"Average pitch: {mean_pitch:.2f} Hz"
 
+def all_in_one_inator(ytvideo, download_yt_video_as, min_duration, max_duration, name_for_outputted_split_files, progress=gr.Progress()):
+    if download_as == "mp3":
+        gr.Warning("MP3 is experimental, especially with this, so caution is advised.")
+    if not os.path.exists("output.wav") and os.path.exists("output.mp3"):
+        gr.Warning("Output.wav does not exist, but output.mp3 does. Splitting MP3 file instead...")
+    if not os.path.exists("output.mp3") and os.path.exists("output.wav"):
+        gr.Warning("Output.mp3 does not exist, but output.wav does. Splitting WAV file instead...")
+    if min_duration == max_duration:
+        raise gr.Error(f"Cannot split mindur={min_duration} and maxdur={max_duration}, min and max are the same number.")
+    elif min_duration > max_duration:
+        raise gr.Error(f"Cannot split mindur={mindur} and maxdur={maxdur}, mindur is higher than maxdur.")
+    elif name_for_outputted_split_files == None:
+        raise gr.Error("Split files name cannot be empty! This will be replaced with an alternative naming style in the future.")
+    else:
+        try:
+            progress(0, "Downloading video...")
+            yt = YouTube(ytvideo)
+        except pytube.exceptions.RegexMatchError:
+            raise gr.Error("URL no valid or was left empty! Please fix the link or enter one.")
+        video = yt.streams.get_highest_resolution()
+        video.download()
+        video_path = f"{video.default_filename}"
+        video_clip = VideoFileClip(video_path)
+        audio_clip = video_clip.audio
+        if download_yt_video_as == "wav":
+            audio_clip.write_audiofile("output.wav")
+        elif download_yt_video_as == "mp3":
+            audio_clip.write_audiofile("output.mp3")
+        audio_clip.close()
+        video_clip.close()
+        for removemp4 in glob.glob("*.mp4"):
+            os.remove(removemp4)
+        progress(0.5, "Video downloaded! Starting split process...")
+        audio_path = "output.wav" if not os.path.exists("output.mp3") else "output.mp3"
+        audio_regions = auditok.split(
+            audio_path,
+            min_dur=min_duration,
+            max_dur=max_duration,
+            max_silence=0.3,
+            energy_threshold=45
+        )
+        os.remove(audio_path)
+        for i, r in enumerate(audio_regions):
+            filename = r.save(f"{name_for_outputted_split_files}-{i+1}.wav")
+        for f in sorted(glob.glob("*.wav")):
+            audio_files = glob.glob("*.wav")
+            zip_file_name = "audio_files.zip"
+            with zipfile.ZipFile(zip_file_name, 'w') as zip_file:
+                for audio_file in audio_files:
+                    zip_file.write(audio_file, os.path.basename(audio_file))
+        for file2 in glob.glob("*.wav"):
+            os.remove(file2)
+        progress(1, "Done! Cleaning up...")
+        time.sleep(2)
+        return "Process done successfully! Check below for zipped files!", zip_file_name
+
 with gr.Blocks(theme='NoCrypt/miku', title="Global Dataset Maker") as app:
     gr.HTML(
         "<h1> Welcome to the GDMGS! (GlobalDatasetMaker Gradio Space) </h1>"
@@ -222,7 +278,24 @@ with gr.Blocks(theme='NoCrypt/miku', title="Global Dataset Maker") as app:
                         [upload],
                         [gr.Text(label="Result")]
                     )
+            with gr.Tab("All-in-one downloader and splitter"):
+                gr.Markdown("This is very experimental and may break or change in the future. This essentially combines both the first 2 tabs into an all-in-one script.")
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            ytvideo = gr.Textbox(label="URL")
+                            download_yt_video_as = gr.Radio(["wav", "mp3"], value="wav", label="Audio output format")
+                            min_duration = gr.Number(label="Min duration", minimum=1, maximum=10, value=1)
+                            max_duration = gr.Number(label="Max duration", minimum=1, maximum=10, value=5)
+                            name_for_outputted_split_files = gr.Textbox(label="Name for split files")
+                            download_and_split_btn = gr.Button("Download and split", variant='primary')
+                        download_and_split_btn.click(
+                            all_in_one_inator,
+                            [ytvideo, download_yt_video_as, min_duration, max_duration, name_for_outputted_split_files],
+                            [gr.Text(label="Result"), gr.File(label="Zipped files")]
+                        )
         with gr.TabItem("Changelog"):
+            gr.Markdown("v0.92 - Added all-in-one tab under Misc Tools.")
             gr.Markdown("v0.91 - Added mp3 file support for single file splitting, and also fixed bug if neither output.wav or output.mp3 exists.")
             gr.Markdown("v0.90a - Fixed bug that if 'show_amount_of_files_and_file_dur' was False, split wav files would not be deleted.")
             gr.Markdown("v0.90 - Added mp3 support for downloading a Youtube video.")