transcribe: fix empty segments

Sometimes the whisper model generates segments with no words in them.
Skip these segments.

src/transcribble/transcribe.py

@@ -10,6 +10,8 @@ def transcribe(input_file, dev):
     segments = []
     for seg in result["segments"]:
         words = seg["words"]
+        if not words:
+            continue
         start = words[0]["start"]
         end = words[-1]["end"]
         segments.append(Segment(start, end, seg["text"].strip()))