MP3 bitrate detection through frequency spectrum analysis

Question

Is there a program that detects MP3 bitrate? I'm not talking about the effective bitrate in which the file was encoded, but about the real bitrate, which can only be calculated using frequency-spectral analysis.

For example, if I have an MP3 encoded in 128 kbps with a size of 1 MB, and then I will encode this MP3 to 320 kbps, which will be 3 MB in size, I will have the same identical audio track, but with different sizes .

If I have an MP3 of 320 kbps and transcode it to 128 kbps, I will lose some quality and therefore some file size.

But still I don’t have the opportunity to verify that 320 kbps is my “real” MP3 bitrate.

Details are explained in this article: http://www.fileden.com/files/2009/2/14/2321055/My%20Documents/MP3%20Bit%20Rate%20Quality%20Detection%20through%20Frequency.pdf

Answer 1

Firstly, https://www.google.com/?q=mp3+cutoff+frequency can be very useful.

Secondly, almost all MP3s are encoded using presets with quite standard multiphase low-pass filters. Since it is actually impossible to achieve lossless compression with MP3s, the higher harmonics of the base frequencies are actually lost (see FFT, DCT, wavelet transforms, etc.); the filter is applied in such a way that the results of a later analysis of the Fourier spectrum are more consistent with the range of a person’s hearing (i.e., non-analyzed / masked frequencies are generally excluded from the analysis). In fact, it is impossible to achieve high compression without trimming / severe distortion of higher frequencies, since they occupy most of the space in the bitstream.

Of course, without clipping, the limitation in the frequency domain will be less accurate, but it will happen anyway. Cutoff is used, among other reasons, so that compression artifacts are generated outside the range of psychoacoustic hearing.

For reference, do a spectral analysis of the stream (a real-time SA with the Winamp clone is enough, if the higher frequency ranges are quite saturated, you can also just make a spectrogram if you have tools) and find the cut-off point. In the example below, clipping occurs at 15 kHz, which tells me that the stream was originally compressed at 128 kbit / s; I would even go so far as to say that you can actually distinguish between & lt; = 128 kbit / s streams by ear with many types of music (drum and bass and other genres of electronic music with many treble come to mind).

The most common slices: (note that they are “hard” in CBR and “soft” in ABR / VBR)

• 128 kbps: 15-16 kHz ( very audible on rock / electronic music! The effect of "loss of space")
• 192 kbps: ~ 19 kHz (in most cases, barely audible, most considered transparent)
• 256-320 kbps:> ~ 20 kHz (inaudible)

Yes, I know that some people can hear above 20 kHz, but the masking effect that appears in music, plus the actual response time of the speakers, means that in real music, 20 kHz cutoff is not related to sound quality.

_{Source: Own search as a sound engineer plus https://web.archive.org/web/20150313010213/http://www.whatinterviewprep.com/prepare-for-the-interview/spectral-analysis/ as an additional link}

Answer 2

The cutoff frequency and bit rate are independent. Yes, most people use presets, so there is a correlation between them, but it is not deterministic.

Answer 3

The only thing you can easily determine from the analysis of the frequency spectrum is the sampling frequency of the input mp3 file.

For example, if your mp3 is selected at 44100, you will not have sound above 22050 Hz, and this will be clearly visible on the spectrum graph.

As you move into the transcendental realm here, try the following:

• encode mp3 up to 128 kbps
• transcode it to say 320kbps
• try rar-ing or 7z-ing the resulting file and the source file. Observe compression ratios.

Their "entropy", i.e. randomness will vary, and perhaps this number will tell you something about how much information is “fabricated” in the bitrate extension during transcoding.

Answer 4

You can open the file "Adobe Audition" or "Cool Edit" and open the frequency analysis window. If the frequency lines up to 20 kHz are larger than the mp3 bitrate, perhaps 320 kbps, if the frequency lines are reduced less than 20 kHz, this is not 320 kbps.