0

On the client side, I am sending a blob audio (wav) file. On the server side, I am trying to convert the blob file to an audio wav file. I did the following:

blob = request.FILES['file']
name = "TEST.wav"
audio = wave.open(name, 'wb')
audio.setnchannels(1)
audio.writeframes(blob.read())

I thought that converting the blob would be similar to converting a blob image to a jpeg file, but was very incorrect in that assumption. That didn't work; I get an error - "Error: sample width not specified." I then used setsampwidth() and tossed in an arbitrary number between 1 and 4 (after looking at the wave.py source file...I don't know why the bytes have to be between 1 and 4). After that another error is thrown - "Error: sampling rate not specified." How do I specify the sampling rate?

What does the setnchannels(), setsampwidth() methods do? Is there an "easy" way I generate the wav file from the blob?

alienboy
  • 443
  • 1
  • 8
  • 18

2 Answers2

1

Previously, I never do it before.. but, in my test this script below is worked well for me.. (But the audio output isn't same like original file).

>>> nchannels = 2
>>> sampwidth = 2
>>> framerate = 8000
>>> nframes = 100
>>> 
>>> import wave
>>> 
>>> name = 'output.wav'
>>> audio = wave.open(name, 'wb')
>>> audio.setnchannels(nchannels)
>>> audio.setsampwidth(sampwidth)
>>> audio.setframerate(framerate)
>>> audio.setnframes(nframes)
>>> 
>>> blob = open("original.wav").read() # such as `blob.read()`
>>> audio.writeframes(blob)
>>>

I found this method at https://stackoverflow.com/a/3637480/6396981

Finally, by changing the value of nchannels and sampwidth with 1. and I got an audio that same with original file.

nchannels = 1
sampwidth = 1
framerate = 8000
nframes = 1

Tested under Python2, and got an error UnicodeDecodeError: 'utf-8' codec can't decode byte 0x95 in position 4: invalid start byte on Python3.

Community
  • 1
  • 1
binpy
  • 3,994
  • 3
  • 17
  • 54
0

I have encountered the same problem as well. My problem was having a low pitched output compared to the original. I manage to reverse engineer the original audio to get the nframes, samplerate, and sampwidth using getnframes(),getframerate(), and getsampwidth() respectively. At last, I managed to tweak the sample frequency/ frame rate to somehow bring the perfect tone.

The tweaking became perfect at a certain offset frequency than the original. Mine worked fine at an offset sum of the sixteenth of the original samplerate.

i.e.

OffsetFrequency = OriginalFrequency/16

Frequency = OriginalFrequency + OffsetFrequency

Dereje T
  • 1
  • 1