gives bogus values

[Andrew Kelley]

Here's an example file: http://superjoe.s3.amazonaws.com/temp/test-rg.mp3

When I run the "mp3gain" cli tool on it, I get this:

Recommended "Track" dB change: -0.730000
Recommended "Track" mp3 gain change: 0
Max PCM sample at current gain: 7739.570901
Max mp3 global gain field: 150
Min mp3 global gain field: 128

Seems reasonable.

But when I use my own program to feed data into libreplaygain I get bogus results. I looked into this further and created a simple test case that fails.

Here's the test file, converted to 44100 Hz, mono, 64-bit float raw file:
http://superjoe.s3.amazonaws.com/temp/test-rg-44100-64bit-mono.raw

Here's some C code relying on libreplaygain to parse the file:

#include <stdio.h>
#include <replaygain/gain_analysis.h>

int main (int argc, char **argv)
{
    FILE *file = fopen(argv[1], "rb");
    double buf[50000];
    size_t bytes_read = fread(buf, 1, 400000, file);
    gain_init_analysis(44100);
    gain_analyze_samples(buf, NULL, bytes_read / 8, 1);
    fclose(file);
    printf("analyzed %lu samples\n", bytes_read / 8);
    printf("track gain: %f\n", gain_get_title());
}

When I run this, I see:
analyzed 47232 samples
track gain: 0.000000

Am I missing something? Is there anywhere that a working replaygain example exists, perhaps a bit simpler than the source to mp3gain?

[Andrew Kelley (andrewrk)]

In my opinion this module should be pulled in favor of https://github.com/jiixyj/libebur128