Merge pull request #1774 from lym0302/add_streaming_cli

[server] fix tts streaming server

demos/streaming_tts_server/test_client.sh

@@ -1,7 +1,7 @@
 #!/bin/bash
 
 # http client test
-paddlespeech_client tts --server_ip 127.0.0.1 --port 8092 --protocol http --input "您好，欢迎使用百度飞桨语音合成服务。" --output output.wav
+paddlespeech_client tts_online --server_ip 127.0.0.1 --port 8092 --protocol http --input "您好，欢迎使用百度飞桨语音合成服务。" --output output.wav
 
 # websocket client test
-#paddlespeech_client tts --server_ip 127.0.0.1 --port 8092 --protocol websocket --input "您好，欢迎使用百度飞桨语音合成服务。" --output output.wav
+#paddlespeech_client tts_online --server_ip 127.0.0.1 --port 8092 --protocol websocket --input "您好，欢迎使用百度飞桨语音合成服务。" --output output.wav

paddlespeech/server/utils/audio_process.py

@@ -157,7 +157,7 @@ def save_audio(bytes_data, audio_path, sample_rate: int=24000) -> bool:
 
     if audio_path.endswith("pcm"):
         with open(audio_path, "wb") as f:
-            f.write(bubytes_dataffer)
+            f.write(bytes_data)
     elif audio_path.endswith("wav"):
         with open("./tmp.pcm", "wb") as f:
             f.write(bytes_data)

paddlespeech/t2s/modules/predictor/length_regulator.py

@@ -49,7 +49,9 @@ class LengthRegulator(nn.Layer):
         encodings: (B, T, C)
         durations: (B, T)
         """
-        batch_size, t_enc = durations.shape
+        #batch_size, t_enc = durations.shape  # linux
+        batch_size = paddle.shape(durations)[0]  # windows and mac
+        t_enc = paddle.shape(durations)[1]  # windows and mac
         durations = durations.numpy()
         slens = np.sum(durations, -1)
         t_dec = np.max(slens)