Add some docstrings

Author: tonybaloney

TransformersSharp.Tests/TransformerEnvironment.Test.cs

@@ -141,7 +141,7 @@ public void TextToAudioPipeline_Generate()
         public void AutomaticSpeechRecognitionPipeline_Transcribe()
         {
             var pipeline = AutomaticSpeechRecognitionPipeline.FromModel("openai/whisper-tiny");
-            var audioPath = "https://huggingface.co/datasets/Narsil/asr_dummy/resolve/main/1.flac"; // Replace with a valid audio path
+            var audioPath = "https://huggingface.co/datasets/Narsil/asr_dummy/resolve/main/1.flac";
             var result = pipeline.Transcribe(audioPath);
             Assert.NotNull(result);
             Assert.NotEmpty(result);

TransformersSharp/MEAI/SpeechToTextClient.cs

@@ -18,7 +18,7 @@ public static SpeechToTextClient FromModel(string model, TorchDtype? torchDtype
 
     public void Dispose()
     {
-        // Nothing to do right now
+        // Nothing to do right now. ASR pipeline object will be collected via next GC pass anyway.
     }
 
     public object? GetService(Type serviceType, object? serviceKey = null) =>

TransformersSharp/python/transformers_wrapper.py

@@ -157,14 +157,26 @@ def invoke_text_to_audio_pipeline(pipeline: Pipeline,
 
 def invoke_automatic_speech_recognition_pipeline(pipeline: Pipeline, audio: str) -> str:
     """
-    Invoke an automatic speech recognition pipeline.
+    Invoke an automatic speech recognition pipeline and return the result.
+
+    Args:
+        pipeline: The ASR pipeline object
+        audio: A local file path or a URL
+    Returns:
+        The text detected
     """
     r = pipeline(audio, return_timestamps=False)
     return r['text']
 
 def invoke_automatic_speech_recognition_pipeline_from_bytes(pipeline: Pipeline, audio: bytes) -> str:
     """
-    Invoke an automatic speech recognition pipeline.
+    Invoke an automatic speech recognition pipeline and return the result.
+    
+    Args:
+        pipeline: The ASR pipeline object
+        audio: The bytes of the audio file
+    Returns:
+        The text detected
     """
     r = pipeline(audio, return_timestamps=False)
     return r['text']