diff --git a/api.py b/api.py
index 91294c8a..11c642bc 100644
--- a/api.py
+++ b/api.py
@@ -18,6 +18,7 @@
 `-p` - `绑定端口, 默认9880, 可在 config.py 中指定`
 `-fp` - `覆盖 config.py 使用全精度`
 `-hp` - `覆盖 config.py 使用半精度`
+`-sm` - `流式返回模式, 默认不启用, "close","c", "normal","n", "keepalive","k"`
 
 `-hb` - `cnhubert路径`
 `-b` - `bert路径`
@@ -304,6 +305,19 @@ def get_spepc(hps, filename):
     return spec
 
 
+def pack_audio(audio_bytes, data, rate):
+    with sf.SoundFile(audio_bytes, mode='w', samplerate=rate, channels=1, format='ogg') as audio_file:
+        audio_file.write(data)
+    return audio_bytes
+
+
+def read_clean_buffer(audio_bytes):
+    audio_chunk = audio_bytes.getvalue()
+    audio_bytes.truncate(0)
+    audio_bytes.seek(0)
+    return audio_bytes, audio_chunk
+
+
 def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language):
     t0 = ttime()
     prompt_text = prompt_text.strip("\n")
@@ -328,9 +342,10 @@ def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language)
     text_language = dict_language[text_language.lower()]
     phones1, bert1, norm_text1 = get_phones_and_bert(prompt_text, prompt_language)
     texts = text.split("\n")
-    audio_opt = []
+    audio_bytes = BytesIO()
 
     for text in texts:
+        audio_opt = []
         phones2, bert2, norm_text2 = get_phones_and_bert(text, text_language)
         bert = torch.cat([bert1, bert2], 1)
 
@@ -365,13 +380,15 @@ def get_tts_wav(ref_wav_path, prompt_text, prompt_language, text, text_language)
         audio_opt.append(audio)
         audio_opt.append(zero_wav)
         t4 = ttime()
+        audio_bytes = pack_audio(audio_bytes,(np.concatenate(audio_opt, 0) * 32768).astype(np.int16),hps.data.sampling_rate)
     # logger.info("%.3f\t%.3f\t%.3f\t%.3f" % (t1 - t0, t2 - t1, t3 - t2, t4 - t3))
-        ogg = BytesIO()
-        sf.write(ogg, (np.concatenate(audio_opt, 0) * 32768).astype(np.int16), hps.data.sampling_rate, format="ogg")
-        ogg.seek(0)
-        chunk = ogg.read()
-        yield chunk
-        audio_opt = []
+        if stream_mode == "normal":
+            audio_bytes, audio_chunk = read_clean_buffer(audio_bytes)
+            yield audio_chunk
+    
+    if not stream_mode == "normal": 
+        yield audio_bytes.getvalue()
+
 
 
 def handle_control(command):
@@ -465,6 +482,7 @@ parser.add_argument("-fp", "--full_precision", action="store_true", default=Fals
 parser.add_argument("-hp", "--half_precision", action="store_true", default=False, help="覆盖config.is_half为True, 使用半精度")
 # bool值的用法为 `python ./api.py -fp ...`
 # 此时 full_precision==True, half_precision==False
+parser.add_argument("-sm", "--stream_mode", type=str, default="close", help="流式返回模式, close / normal / keepalive")
 parser.add_argument("-hb", "--hubert_path", type=str, default=g_config.cnhubert_path, help="覆盖config.cnhubert_path")
 parser.add_argument("-b", "--bert_path", type=str, default=g_config.bert_path, help="覆盖config.bert_path")
 
@@ -507,6 +525,14 @@ if args.full_precision and args.half_precision:
     is_half = g_config.is_half  # 炒饭fallback
 logger.info(f"半精: {is_half}")
 
+# 流式返回模式
+if args.stream_mode.lower() in ["normal","n"]:
+    stream_mode = "normal"
+    logger.info("流式返回已开启")
+else:
+    stream_mode = "close"
+
+
 # 初始化模型
 cnhubert.cnhubert_base_path = cnhubert_base_path
 tokenizer = AutoTokenizer.from_pretrained(bert_path)