保存 trascripts.json

examples/asr/asr_cache_aware_streaming/speech_to_text_cache_aware_streaming_infer.py

@@ -306,9 +306,8 @@ def perform_streaming(
 
         if debug_mode:
             current_texts = extract_transcriptions(transcribed_texts)
-            # 实时打印每个 chunk 的结果
-            print(f"\r[Streaming Step {step_num}] {current_texts[0]}", end='', flush=True)
-            # logging.info(f"Streaming transcriptions: {extract_transcriptions(transcribed_texts)}")
+            # 实时打印每个 chunk 的结果，每次换行
+            print(f"[Streaming Step {step_num}] {current_texts[0]}", flush=True)
 
     print() # 换行
     final_streaming_tran, final_streaming_timestamps = extract_transcriptions(transcribed_texts, return_timestamps=True)
@@ -481,6 +480,21 @@ def main(cfg: TranscriptionConfig):
                 for key, val in streaming_timestamps[0].items():
                     if key != 'timestep':
                         logging.info(f"Timestamps {key}: {normalize_timestamp_output(val)}")
+                        
+            # 将单文件结果写入到根目录 transcripts.json
+            output_json_path = os.path.join(os.getcwd(), "transcripts.json")
+            record = {
+                "audio_filepath": cfg.audio_file,
+                "pred_text": streaming_tran[0] if streaming_tran else "",
+            }
+            if streaming_timestamps and streaming_timestamps[0]:
+                for key, val in streaming_timestamps[0].items():
+                    if key != 'timestep':
+                        record[f'{key}'] = normalize_timestamp_output(val)
+                        
+            with open(output_json_path, "w", encoding='utf-8') as out_f:
+                out_f.write(json.dumps(record, ensure_ascii=False) + '\n')
+            logging.info(f"Results saved to {output_json_path}")
             
             # return early for single file mode
             return