CODEWITHSUNDEEP
tensorflowspeech
xuchenxing
xuchenxing

Reputation: 61

TENSORFLOW speech-command: Error(Data too short when trying to read string) when decode the wav

thank you for your help.

i am learning the tuturial (speech-command) on the tensorflow, after i download the code and the dataset, i run the program, after several steps of training, an ERROR come out. 

InvalidArgumentError (see above for traceback): Data too short when trying to read string
     [[Node: DecodeWav = DecodeWav[desired_channels=1, desired_samples=16000, _device="/job:localhost/replica:0/task:0/device:CPU:0"](ReadFile)]]
     [[Node: DecodeWav/_21 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/device:GPU:0", send_device="/job:localhost/replica:0/task:0/device:CPU:0", send_device_incarnation=1, tensor_name="edge_4_DecodeWav", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:GPU:0"]()]]

it seems that the decode method is something wrong, but i can not figure out what is wrong. and i didnot change anycode after i download it from github. Can you help me. thanks.

Upvotes: 2

Views: 1941

Answers (2)

Георгий Черноусов
Георгий Черноусов

Reputation: 21

In my case I hadn't got any empty files. The function from the bottom of discussion helped me:

https://github.com/mozilla/DeepSpeech/issues/2048

import os
import wave
import pandas
import sys

def compare_header_and_size(wav_filename):
    with wave.open(wav_filename, 'r') as fin:
        header_fsize = (fin.getnframes() * fin.getnchannels() * fin.getsampwidth()) + 44
    file_fsize = os.path.getsize(wav_filename)
    return header_fsize != file_fsize

df = pandas.read_csv(sys.argv[1])
invalid = df.apply(lambda x: compare_header_and_size(x['wav_filename']), axis=1)
print('The following files are corrupted:')
print(df[invalid].values)

I found out that my audio files was different length while comparing 2 ways of measuring in the function presented.

The reason was that i added metadata to wav files while saving them in the Adobe Audition. That was a mistake

Upvotes: 1

xuchenxing
xuchenxing

Reputation: 61

Problem solved. the problem is one of the voice in the dataset is empty(0 kb), and the program use random to fetch the training data, if random to this empty training voice, it goes the ERROR in the question.

Upvotes: 4

Related Questions