Closed WinderWL closed 3 years ago
请教大佬,windows10使用master分支代码执行本地预测报错如下:(使用DeepSpeech的模型,导出模型也正常,使用1.0的分支可以正常出结果)
(DeepSpeech) D:\PythonProjects\PaddlePaddle-DeepSpeech>python infer_path.py --wav_path=./dataset/test.wav ----------- Configuration Arguments ----------- alpha: 1.2 beam_size: 10 beta: 0.35 cutoff_prob: 1.0 cutoff_top_n: 40 decoding_method: ctc_greedy enable_mkldnn: False is_long_audio: False lang_model_path: ./lm/zh_giga.no_cna_cmn.prune01244.klm mean_std_path: ./dataset/mean_std.npz model_dir: ./models/infer/ to_an: True use_gpu: True use_tensorrt: False vocab_path: ./dataset/zh_vocab.txt wav_path: ./dataset/test.wav ------------------------------------------------ E0903 18:29:50.756563 324 analysis_config.cc:81] Please compile with gpu to EnableGpu() E0903 18:29:50.756563 324 analysis_config.cc:81] Please compile with gpu to EnableGpu() e[37m--- Fused 0 subgraphs into layer_norm op.e[0m Traceback (most recent call last): File "infer_path.py", line 38, in <module> use_tensorrt=args.use_tensorrt, enable_mkldnn=args.enable_mkldnn, to_an=args.to_an) File "D:\PythonProjects\PaddlePaddle-DeepSpeech\utils\predict.py", line 68, in __init__ self.predict(warmup_audio_path) File "D:\PythonProjects\PaddlePaddle-DeepSpeech\utils\predict.py", line 75, in predict audio_feature = self.audio_process.process_utterance(audio_path) File "D:\PythonProjects\PaddlePaddle-DeepSpeech\data_utils\audio_process.py", line 48, in process_utterance specgram = self._normalizer.apply(specgram) File "D:\PythonProjects\PaddlePaddle-DeepSpeech\data_utils\normalizer.py", line 55, in apply return (features - self._mean) / (self._std + eps) ValueError: operands could not be broadcast together with shapes (39,838) (161,1)
你应该是用旧的模型放在新的代码上面。如果你下载旧的模型代码就要用旧的代码。
我重新下载了模型,执行通过了,感谢!!
请教大佬,windows10使用master分支代码执行本地预测报错如下:(使用DeepSpeech的模型,导出模型也正常,使用1.0的分支可以正常出结果)