funasr/bin/asr_inference_launch.py
@@ -658,6 +658,10 @@ sorted_data = sorted(data_with_index, key=lambda x: x[0][1] - x[0][0]) results_sorted = [] batch_size_token_ms = batch_size_token*60 if speech2text.device == "cpu": batch_size_token_ms = 0 batch_size_token_ms = max(batch_size_token_ms, sorted_data[0][0][1] - sorted_data[0][0][0]) batch_size_token_ms_cum = 0 beg_idx = 0 for j, _ in enumerate(range(0, n)):
