I have received the attached image for my recording. What does it actually denote? What is the x-axis and y-axis for it?
1 Like
There’s an article How does Chromaprint work? by its author that might help you here.
1 Like
Thanks,I already went through it. That’s why I got this doubt. What exactly is denoted at x-axis and y-axis. Is it intensity vs time or something else?
1 Like
Well, I just read the Chromaprint explanation, and don’t know anything about it beyond that. But I can try to answer your question.
The Chromaprint image appears to denote in the y-axis a blend of 16 different small numbers that characterise the relationships between high, medium, and low frequencies over time for a brief time period of the music. It appears to denote in the x-axis time, or rather the starting times of several brief time periods of the music.
The algorithm for generating the 16 small numbers is described in the Chromaprint explanation, and I won’t try to summarise that. I think the explanation there is pretty clear. In a sense, the details of the small numbers and what relationships they describe don’t matter much.
What matters is the claim that the same audio recording processed with different bit rates and different compression will reliably come up with pretty much the same small numbers, and that two different audio recordings will reliably come up with different small numbers. By extension, if two Chromaprint images are pretty much the same, then the corresponding audio files are likely from the same source. If different, the audio files are likely from different sources.
5 Likes
Thanks a lot. This really helped!
Time would be the x-axis as they’re presented on the explanation page, but the y-axis in the original post or as they are displayed in acoustid, wouldn’t it? (Whichever of the two axes is longer is generally going to represent time.)
2 Likes
Sorry I could not understand what you are trying to say for y-axis. So let me summarize what I understood the longer axis is time and the shorter one will be frequency relationship based on the bits. Right?
I didn’t say “longer axis” and “shorter”. The Chromaprint explanation uses the terms “x-axis” and “y-axis”.
One axis is a a blend of 16 different small numbers that characterise the relationships between high, medium, and low frequencies over time for a brief time period of the music. This axis will be the same size no matter how long the piece of music is which is fingerprinted.
The other axis denotes the starting times of several brief time periods of the music. This axis will be longer for a fingerprint of a longer piece of music, and shorter for a fingerprint of a shorter piece of music. If the piece of music is short enough, then this axis might end up shorter than the first axis.
It is more correct to say the first axis is “bits based on the frequency relationships”, instead of “frequency relationship based on the bits”.
All this is based on my understanding reading that Chromaprint explanation quickly. I am not an expert by any means.
2 Likes
Thanks a lot for your explanation, helped a lot.
Sorry if I confused the issue. All I was trying to say is that the x-axis as shown on the Chromaprint explanation that Mineo linked is equivalent (or so I believe) to the y-axis as rendered on an acoustid fingerprint page .
1 Like
Y axis is time for sure as it is how we compare them by shifting up or down.
X axis, I guess it is low key sound on the left and high pitch on the right. But not sure at all.