Microphone live inference catching up when inference is too slow (whisper). (#21219)

* Microphone live inference catching up when inference is too slow
(whisper).

* Adding copyright.

src/transformers/pipelines/audio_utils.py

+# Copyright 2023 The HuggingFace Team. All rights reserved.
+import datetime
 import platform
 import subprocess
 from typing import Optional, Tuple, Union
@@ -154,6 +156,8 @@ def ffmpeg_microphone_live(
 
     stride_left = int(round(sampling_rate * stride_length_s[0])) * size_of_sample
     stride_right = int(round(sampling_rate * stride_length_s[1])) * size_of_sample
+    audio_time = datetime.datetime.now()
+    delta = datetime.timedelta(seconds=chunk_s)
     for item in chunk_bytes_iter(microphone, chunk_len, stride=(stride_left, stride_right), stream=True):
         # Put everything back in numpy scale
         item["raw"] = np.frombuffer(item["raw"], dtype=dtype)
@@ -162,6 +166,10 @@ def ffmpeg_microphone_live(
             item["stride"][1] // size_of_sample,
         )
         item["sampling_rate"] = sampling_rate
+        audio_time += delta
+        if datetime.datetime.now() > audio_time + 10 * delta:
+            # We're late !! SKIP
+            continue
         yield item