[TTS] Add format validation

Signed-off-by: Ryan <rlangman@nvidia.com>

scripts/dataset_processing/tts/preprocess_audio.py

@@ -99,7 +99,7 @@ def get_args():
     )
     parser.add_argument(
         "--output_format",
-        default=".flac",
+        default="flac",
         type=str,
         help="If provided, format output audio will be saved as. If not provided, will keep original format.",
     )
@@ -164,6 +164,7 @@ def _process_entry(
         if audio.size > 0:
             sf.write(file=output_path, data=audio, samplerate=sample_rate)
             output_duration = librosa.get_duration(y=audio, sr=sample_rate)
+            sf.available_formats()
         else:
             output_duration = 0.0
 
@@ -208,6 +209,11 @@ def main():
     else:
         audio_trimmer = None
 
+    if output_format:
+        if output_format.upper() not in sf.available_formats():
+            raise ValueError(f"Unsupported output audio format: {output_format}")
+        output_format = f".{output_format}"
+
     output_audio_dir.mkdir(exist_ok=True, parents=True)
 
     entries = read_manifest(input_manifest_path)