add file splitter in misc tools

app.py

@@ -1,4 +1,4 @@
-import gradio as gr, glob, os, auditok, random, zipfile, wave, pytube.exceptions
+import gradio as gr, glob, os, auditok, zipfile, wave, pytube.exceptions
 from pytube import YouTube
 from moviepy.editor import VideoFileClip
 
@@ -19,9 +19,9 @@ def download_video(url):
         os.remove(removalmp4)
     return "Finished downloading! Please proceed to final tab."
 
-def split_audio(mindur, maxdur, name_for_split_files, show_amount_of_files_and_file_dur):
+def split_audio_from_yt_video(mindur, maxdur, name_for_split_files, show_amount_of_files_and_file_dur):
     if show_amount_of_files_and_file_dur == True:
-        gr.Warning(f"show_amount_of_files_and_file_dur set to True. This feature may be inaccurate especially for WAV files, so dont rely too much on the count and duration.")
+        gr.Warning(f"show_amount_of_files_and_file_dur set to True. This will take longer if your audio file is long.")
     if not os.path.exists("output.wav"):
         raise gr.Error("Output.wav does not exist! Did you do the first tab correctly or at all?")
     if mindur == maxdur:
@@ -43,14 +43,13 @@ def split_audio(mindur, maxdur, name_for_split_files, show_amount_of_files_and_f
         for i, r in enumerate(audio_regions):
             filename = r.save(f"{name_for_split_files}-{i+1}.wav")
         for f in sorted(glob.glob("*.wav")):
-            set_name = name_for_split_files + "-" + str(random.randint(1, 91071988)) + ".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))
         if show_amount_of_files_and_file_dur == False:
-            return "Files split successfully!\n\nCheck below for zipped files.", zip_file_name
+            return "Files split successfully!\nCheck below for zipped files.", zip_file_name
         elif show_amount_of_files_and_file_dur == True:
             largest_file = ("", 0)
             total_files = 0
@@ -73,7 +72,7 @@ def split_audio(mindur, maxdur, name_for_split_files, show_amount_of_files_and_f
             length_mins = total_length / 60
             for file2 in glob.glob("*.wav"):
                 os.remove(file2)
-            return f"Files split successfully!\n\nCheck below for zipped files.\n\n{total_files} files created, {length_mins:.2f} minutes total.", zip_file_name
+            return f"Files split successfully!\nCheck below for zipped files.\n\n{total_files} files created, {length_mins:.2f} minutes total.", zip_file_name
 
 def analyze_audio(zip_file_path):
     with zipfile.ZipFile(zip_file_path, 'r') as zip_ref:
@@ -96,13 +95,41 @@ def analyze_audio(zip_file_path):
     else:
         return "No average sample rate could be found."
 
+def split_wav_file(audiofileuploader, mindur2, maxdur2, name_for_split_files2):
+    if audiofileuploader == None:
+        raise gr.Error("Audio file cannot be empty!")
+    if mindur2 == maxdur2:
+        raise gr.Error(f"Cannot split mindur={mindur} and maxdur={maxdur}, min and max are the same number.")
+    elif mindur2 > maxdur2:
+        raise gr.Error(f"Cannot split mindur={mindur} and maxdur={maxdur}, mindur is higher than maxdur.")
+    elif name_for_split_files2 == None:
+        raise gr.Error("Split files name cannot be empty! This will be replaced with an alternative naming style in the future.")
+    else:
+        audio_path = audiofileuploader
+        audio_regions = auditok.split(
+            audio_path,
+            min_dur=mindur2,
+            max_dur=maxdur2,
+            max_silence=0.3,
+            energy_threshold=45
+        )
+        os.remove(audio_path)
+        for i, r in enumerate(audio_regions):
+            filename = r.save(f"{name_for_split_files2}-{i+1}.wav")
+        for f in sorted(glob.glob("*.wav")):
+            audio_files = glob.glob("*.wav")
+            zip_file_name2 = "audio_files.zip"
+            with zipfile.ZipFile(zip_file_name2, "w") as zip_file:
+                for audio_file in audio_files:
+                    zip_file.write(audio_file, os.path.basename(audio_file))
+    return f"File split successfully!\nCheck below for zipped files.\nAmount created: {len(audio_files)}", zip_file_name2
 
 with gr.Blocks(theme=gr.themes.Monochrome(), title="Global Dataset Maker") as app:
     gr.HTML(
         "<h1> Welcome to the GDMGS! (GlobalDatasetMaker Gradio Space) </h1>"
     )
     gr.Markdown(
-        "This Space will create a dataset for you, all automatically. **Please be warned that due to not having a GPU on this Space, some steps might take longer to complete, so please be patient.**"
+        "This Space will create a dataset for you, all automatically. **Please be warned that due to not having a GPU on this Space, some steps might take longer to complete.**"
     )
     with gr.Tabs():
         with gr.TabItem("Download Video"):
@@ -126,7 +153,7 @@ with gr.Blocks(theme=gr.themes.Monochrome(), title="Global Dataset Maker") as ap
                         show_amount_of_files_and_file_dur = gr.Checkbox(False, label="Show total amount of files and duration?")
                     splitbtn = gr.Button("Split", variant='primary')
             splitbtn.click(
-                split_audio,
+                split_audio_from_yt_video,
                 inputs=[mindur, maxdur, name_for_split_files, show_amount_of_files_and_file_dur],
                 outputs=[gr.Text(label="Output"), gr.File(label="Zipped files")]
             )
@@ -143,5 +170,19 @@ with gr.Blocks(theme=gr.themes.Monochrome(), title="Global Dataset Maker") as ap
                         [zipuploader],
                         [gr.Text(label="Result")]
                     )
-
+            with gr.Tab("File splitter"):
+                gr.Markdown("If you would rather split a single WAV (mp3 support soon) audio file, use this method instead.")
+                with gr.Row():
+                    with gr.Column():
+                        with gr.Row():
+                            audiofileuploader = gr.File(file_count='single', file_types=[".wav"], label="WAV file")
+                            mindur2 = gr.Number(label="Min duration", minimum=1, maximum=10, value=1)
+                            maxdur2 = gr.Number(label="Max duration", minimum=1, maximum=10, value=5)
+                            name_for_split_files2 = gr.Textbox(label="Name for split files")
+                            audiofileuploadbtn = gr.Button("Split", variant='primary')
+                        audiofileuploadbtn.click(
+                            split_wav_file,
+                            [audiofileuploader, mindur2, maxdur2, name_for_split_files2],
+                            [gr.Text(label="Output"), gr.File(label="Zipped files")]
+                        )
 app.launch()