added check for filetype and file length

for preprocessing

Co-Authored-By: kalomaze <66376113+kalomaze@users.noreply.github.com>

my_utils.py

@@ -108,3 +108,28 @@ def load_audio(file, sr, DoFormant, Quefrency, Timbre):
         converted = False
     
     return np.frombuffer(out, np.float32).flatten()
+
+
+def check_audio_duration(file):
+    try:
+        # Strip whitespaces and unnecessary characters from the file name
+        file = file.strip(" ").strip('"').strip("\n").strip('"').strip(" ")
+
+        # Probe the audio file for information
+        probe = ffmpeg.probe(file)
+
+        # Extract the duration from the probe result
+        duration = float(probe['streams'][0]['duration'])
+
+        # If the duration is less than 0.75 seconds, print the message and exit the loop
+        if duration < 0.76:
+            print(
+                f"\n------------\n"
+                f"Audio file, {file.split('/')[-1]}, under ~0.76s detected - file is too short. Target at least 1-2s for best results."
+                f"\n------------\n\n"
+            )
+            return False
+
+        return True
+    except Exception as e:
+        raise RuntimeError(f"Failed to check audio duration: {e}")

tensorlowest.py

@@ -47,7 +47,7 @@ def main(model_name, save_freq, lastmdls):
     tfile = os.path.join(tensordir, latest_file)
     
     ea = event_accumulator.EventAccumulator(tfile,
-        size_guidance={ # see below regarding this argument
+        size_guidance={
         event_accumulator.COMPRESSED_HISTOGRAMS: 500,
         event_accumulator.IMAGES: 4,
         event_accumulator.AUDIO: 4,

trainset_preprocess_pipeline_print.py

@@ -14,7 +14,7 @@ from slicer2 import Slicer
 import librosa, traceback
 from scipy.io import wavfile
 import multiprocessing
-from my_utils import load_audio
+from my_utils import load_audio, check_audio_duration
 import tqdm
 
 DoFormant = False
@@ -80,28 +80,38 @@ class PreProcess:
         )
 
     def pipeline(self, path, idx0):
+        
+        file_extension = path.split('.')[-1]
+        supported_file_extensions = {'wav', 'mp3', 'flac', 'ogg', 'opus',
+                                    'm4a', 'mp4', 'aac', 'alac', 'wma',
+                                    'aiff', 'webm', 'ac3'}
+        
         try:
-            audio = load_audio(path, self.sr, DoFormant, Quefrency, Timbre)
-            # zero phased digital filter cause pre-ringing noise...
-            # audio = signal.filtfilt(self.bh, self.ah, audio)
-            audio = signal.lfilter(self.bh, self.ah, audio)
+            if file_extension in supported_file_extensions:
+                if not check_audio_duration(path): return
+                audio = load_audio(path, self.sr, DoFormant, Quefrency, Timbre)
+                # zero phased digital filter cause pre-ringing noise...
+                # audio = signal.filtfilt(self.bh, self.ah, audio)
+                audio = signal.lfilter(self.bh, self.ah, audio)
 
-            idx1 = 0
-            for audio in self.slicer.slice(audio):
-                i = 0
-                while 1:
-                    start = int(self.sr * (self.per - self.overlap) * i)
-                    i += 1
-                    if len(audio[start:]) > self.tail * self.sr:
-                        tmp_audio = audio[start : start + int(self.per * self.sr)]
-                        self.norm_write(tmp_audio, idx0, idx1)
-                        idx1 += 1
-                    else:
-                        tmp_audio = audio[start:]
-                        idx1 += 1
-                        break
-                self.norm_write(tmp_audio, idx0, idx1)
-            # println("%s->Suc." % path)
+                idx1 = 0
+                for audio in self.slicer.slice(audio):
+                    i = 0
+                    while 1:
+                        start = int(self.sr * (self.per - self.overlap) * i)
+                        i += 1
+                        if len(audio[start:]) > self.tail * self.sr:
+                            tmp_audio = audio[start : start + int(self.per * self.sr)]
+                            self.norm_write(tmp_audio, idx0, idx1)
+                            idx1 += 1
+                        else:
+                            tmp_audio = audio[start:]
+                            idx1 += 1
+                            break
+                    self.norm_write(tmp_audio, idx0, idx1)
+                # println("%s->Suc." % path)
+            else:
+                print(f"Unsupported audio format! - {path.split('/')[-1]}")
         except:
             println("%s->%s" % (path, traceback.format_exc()))