Converting .mcf files to .wav or .mp3 is not a common task. For me was an ugly task that had to be to done fast, without any support or documentation.

Because I had to find a method to convert like 100 files fast I have build a little convoluted quick and dirty hack. The program and the algorithm could be improved but I wont waste more time, I hope that I do not need it any more. I am not a python programmer, the script does not reflect my programming style.

The trip

I have found a reference about the format of the .mcf file in the AudioCapture.h file from the Oreka GitHub repository:

class DLL_IMPORT_EXPORT_ORKBASE AudioChunkDetails
{
public:
    AudioChunkDetails();
    void Clear();

    int m_marker;
    AudioEncodingEnum m_encoding;
    unsigned int m_numBytes;
    unsigned int m_timestamp;        // usually relative timestamp measured in samples
    unsigned int m_arrivalTimestamp; // usually unix timestamp of arrival
    unsigned int m_sequenceNumber;
    unsigned int m_sampleRate;
    char m_rtpPayloadType;           // -1 if none
    unsigned char m_channel;         // 0 if mono, 1 or 2 if stereo, 100 if we have separated multiple channels
};

Using Visual Studio Code and the ms-vscode.hexeditor extension I have analyzed some of the files and observed that all had 2 channels encoded in alaw format with a 8000 Hz sample rate.

Based on these I have built a python script for python 2.7 but probably will run with python 3 also.

How it works

First, the python script creates a convertor.sh file.

For every .mcf file in a specified directory:

• extracts the two streams from the filename.mcf file in two files, filename1.alaw and filename2.alaw
• write commands in convertor.sh to convert .alaw files to temporary .wav files
• write a command in convertor.sh to convert the two temporary .wav files to a stereo .mp3 file

After the python script finishes, launch the convertor.sh file.

Conversion commands

A raw alaw encoded file can be converted to .wav like this (8000 being the sample rate):

ffmpeg -f alaw -ar 8000 -i alaw_file_name output.wav

The .wav files can be converted to a stereo .mp3 file like this:

ffmpeg -i ch1.wav -i ch2.wav -filter_complex "[0:a][1:a]join=inputs=2:channel_layout=stereo[a]" -map "[a]" output.mp3

The script

Here is the ugly python script:

import os
from struct import *

scriptfile = open("convertor.sh", "wt")
str = "#!/bin/bash\n\n"
scriptfile.write(str)

def process_file(inputFileName):
    fileComp = os.path.splitext(inputFileName)
    fileName = fileComp[0]
    fileExt = fileComp[1]
    if fileExt != '.mcf':
        return

    filei = open(fileName + ".mcf", "rb")
    file1 = open(fileName + "1.alaw", "wb")
    file2 = open(fileName + "2.alaw", "wb")

    print("Processing " + inputFileName)
    while True :           
        header = filei.read(32)
        if not header :
            break
        if len(header) != 32 :
            print("\tERROR header length: " + str(len(header)))
            break

        hMarker, hEncoding, hNumBytes, \
            hTimestamp, hArrivalTimestamp, \
            hSequenceNumber, hSampleRate, hRtpPayloadType, hChannel, hDumb \
            = unpack('<IIIIIIIBBH', header)
        
        if hMarker != 0x2A2A2A2A :
            print("\tERROR hMarker: " + hex(hMarker))
            print(hex(filei.tell()))
            break

        if hNumBytes == 0 :
            print("\tERROR hNumBytes: " + str(hNumBytes))
            break

        data = filei.read(hNumBytes)
        if len(data) != hNumBytes :
            print("\tERROR data length: " + str(hNumBytes))
            break

        if (hChannel == 1) :
            file1.write(data)
        if (hChannel == 2) :
            file2.write(data)

    file2.close()
    file1.close()
    filei.close()

    str = "ffmpeg -f alaw -ar 8000 -i " + fileName + "1.alaw " + fileName + "1.wav\n"
    scriptfile.write(str)
    str = "ffmpeg -f alaw -ar 8000 -i " + fileName + "2.alaw " + fileName + "2.wav\n"
    scriptfile.write(str)
    str = "ffmpeg -i " + fileName + "1.wav -i " + fileName + "2.wav -filter_complex \"[0:a][1:a]join=inputs=2:channel_layout=stereo[a]\" -map \"[a]\" " + os.path.basename(fileName) + ".mp3\n"
    scriptfile.write(str)

srcDir = "10"
print("Searching for files in " + srcDir + " directory")
for root, subdirs, files in os.walk(srcDir):
    for filename in files:
        file_path = os.path.join(root, filename)
        process_file(file_path)

scriptfile.close()
print("Now run convertor.sh")

Easy peasy, right ? :)

Want to share this post? Click to copy the link.