clamp end_timestamp to prevent infinite loop

2025-07-01 18:17:27 -04:00 · 2023-04-11 20:15:37 +01:00
parent 48ed89834e
commit 6a72b61564
1 changed files with 5 additions and 1 deletions
--- a/whisperx/asr.py
+++ b/whisperx/asr.py
@ -269,6 +269,10 @@ def transcribe(
                    end_timestamp_pos = (
                        sliced_tokens[-1].item() - tokenizer.timestamp_begin
                    )
+                    
+                    # clamp end-time to at least be 1 frame after start-time
+                    end_timestamp_pos = max(end_timestamp_pos, start_timestamp_pos + time_precision)
+
                    current_segments.append(
                        new_segment(
                            start=time_offset + start_timestamp_pos * time_precision,