Python – Convert 32-bit Floating Points to 16-bit PCM range

audiopython

I have some data generated by the javascript HTML5 web audio api. It generates Float32Array, an array of 32-bit Floating Points, between -1 and 1. I stream the data to my server using a websocket.

I need to convert the 32-bit floating points to 16-bit PCM range between -32768 and +32767 (16-bit signed integer). This then allows the data to be used as a wav file.

I'm having trouble converting. I suspect the answer is to use the struct module, but I can't get the correct formatting.

Best Answer

Here's a sample Python 2.7 program that reads a file containing raw 32-bit floating point audio samples and creates a WAV file containing those samples converted 16-bit signed integer samples:

import sys
import array
import struct
import wave

def convert(fin, fout, chunk_size = 1024 * 1024):
    chunk_size *= 4    # convert from samples to bytes

    waveout = wave.open(fout, "wb")
    waveout.setparams((1, 2, 44100, 0, "NONE", ""))

    while True:
        raw_floats = fin.read(chunk_size)
        if raw_floats == "":
            return
        floats = array.array('f', raw_floats)
        samples = [sample * 32767
                   for sample in floats]
        raw_ints = struct.pack("<%dh" % len(samples), *samples)
        waveout.writeframes(raw_ints)

convert(open(sys.argv[1], "rb"), open(sys.argv[2], "wb"))

The code uses array.array to convert the 32-bit floating point samples to a Python floats because it should be a bit faster than struct.unpack. It also uses the native machine byte order, just like Float32Array does. It's not possible to use array.array to create the 16-bit integer samples because they need to use the little endian byte order regardless of the native machine order. The range conversion is handled by simple Python code.

Related Solutions

Python – Limiting floats to two decimal points

You are running into the old problem with floating point numbers that not all numbers can be represented exactly. The command line is just showing you the full floating point form from memory.

With floating point representation, your rounded version is the same number. Since computers are binary, they store floating point numbers as an integer and then divide it by a power of two so 13.95 will be represented in a similar fashion to 125650429603636838/(2**53).

Double precision numbers have 53 bits (16 digits) of precision and regular floats have 24 bits (8 digits) of precision. The floating point type in Python uses double precision to store the values.

For example,

>>> 125650429603636838/(2**53)
13.949999999999999

>>> 234042163/(2**24)
13.949999988079071

>>> a = 13.946
>>> print(a)
13.946
>>> print("%.2f" % a)
13.95
>>> round(a,2)
13.949999999999999
>>> print("%.2f" % round(a, 2))
13.95
>>> print("{:.2f}".format(a))
13.95
>>> print("{:.2f}".format(round(a, 2)))
13.95
>>> print("{:.15f}".format(round(a, 2)))
13.949999999999999

If you are after only two decimal places (to display a currency value, for example), then you have a couple of better choices:

Use integers and store values in cents, not dollars and then divide by 100 to convert to dollars.
Or use a fixed point number like decimal.

Python – Convert bytes to a string

You need to decode the bytes object to produce a string:

>>> b"abcde"
b'abcde'

# utf-8 is used here because it is a very common encoding, but you
# need to use the encoding your data is actually in.
>>> b"abcde".decode("utf-8") 
'abcde'

Best Answer

Related Solutions

Python – Limiting floats to two decimal points

Python – Convert bytes to a string

Related Topic