remove audio duration limit, and fix Copyright writing problems.

paddlespeech/cli/whisper/infer.py

@@ -152,8 +152,7 @@ class WhisperExecutor(BaseExecutor):
         Init model and other resources from a specific path.
         """
         logger.debug("start to init the model")
-        # default max_len: unit:second
+
-        self.max_len = 50
         if hasattr(self, 'model'):
             logger.debug('Model had been initialized.')
             return
@@ -339,12 +338,6 @@ class WhisperExecutor(BaseExecutor):
         try:
             audio, audio_sample_rate = soundfile.read(
                 audio_file, dtype="int16", always_2d=True)
-            audio_duration = audio.shape[0] / audio_sample_rate
-            if audio_duration > self.max_len:
-                logger.error(
-                    f"Please input audio file less then {self.max_len} seconds.\n"
-                )
-                return False
         except Exception as e:
             logger.exception(e)
             logger.error(

paddlespeech/s2t/models/whisper/__init__.py

@@ -1,5 +1,5 @@
 # MIT License, Copyright (c) 2022 OpenAI.
-# Copyright (c) 2022 PaddlePaddle Authors and . All Rights Reserved.
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
 # 
 # Modified from OpenAI Whisper 2022 (https://github.com/openai/whisper/whisper/__init__.py)
 from paddlespeech.s2t.models.whisper.whipser import decode

paddlespeech/s2t/models/whisper/tokenizer.py

@@ -1,5 +1,5 @@
 # MIT License, Copyright (c) 2022 OpenAI.
-# Copyright (c) 2022 PaddlePaddle Authors and . All Rights Reserved.
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
 # 
 # Modified from OpenAI Whisper 2022 (https://github.com/openai/whisper/whisper/tokenizer.py)
 import os

paddlespeech/s2t/models/whisper/utils.py

@@ -1,5 +1,5 @@
 # MIT License, Copyright (c) 2022 OpenAI.
-# Copyright (c) 2022 PaddlePaddle Authors and . All Rights Reserved.
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
 # 
 # Modified from OpenAI Whisper 2022 (https://github.com/openai/whisper/whisper/utils.py)
 import zlib

paddlespeech/s2t/models/whisper/whipser.py

@@ -1,5 +1,5 @@
 # MIT License, Copyright (c) 2022 OpenAI.
-# Copyright (c) 2022 PaddlePaddle Authors and . All Rights Reserved.
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
 #
 # Modified from OpenAI Whisper 2022 (https://github.com/openai/whisper/whisper)
 import os
@@ -195,6 +195,7 @@ class AudioEncoder(nn.Layer):
         x : paddle.Tensor, shape = (batch_size, n_mels, n_ctx)
             the mel spectrogram of the audio
         """
+        print (x.shape)
         x = F.gelu(self.conv1(x))
         x = F.gelu(self.conv2(x))
         x = paddle.transpose(x, (0, 2, 1))