Has anyone noticed an accuracy difference between .mp3 and higher resolution files when analyzing and detecting note and Key for songs?

 

seems like wav and mp3 both work. if for some reason your song has something in the high end that gets chopped by mp3 it wont get detected but thats rare.

try scaler, ez keys, ez bass detectors too. they have a pretty unique algo that seems to average what they detect. for more granular try neural note or melodyne

 

Didn't tested.

But with low quality MP3 (like 128k), there is a distinct "wobbling" effect, changing pitch.
There is less stereo separation too,missing freqs ...
But only on low quality.

 

Most of my .mp3 files are 256 kbp.

 

So it should not a make a difference

It is easy to test : take an original WAV, do the detection.
Convert it to MP3 256 and do the same detection.

Problem with most MP3 is you never know the source.
It is more related to MP3 quality itself than MP3 compression.

 

EDITED：
I deleted it because it seems to be a little different from this thread.
excuse me.

 

Last edited: Feb 9, 2025

Not a lick of a difference for note detection.

 

what a stupid reasoning, most common (<~8th octave) notes' harmonics operate on too low frequencies for normal (>112kb/s, >32kHz samplerate) audio quality factor to be even considered as error
all audio software is converting mp3 to wav due to speed

 

Of course .wav is better sound quality, but does it matter for note detection?

 

I hate to be this guy but... You can use your keyboard as a midi keyboard and then use your ear to find the key in a song. The tiny bit of "work"(and it pains me to use that word when there exists real ear training) is going to prodive you MASSIVE gains in your ability as a musician.

 

I do a lot of transcribing and use mp3 as there isn't any difference. Of course, there is differences in the quality of the data of the file but the way it is used has made no difference. The difference is more in line with the program used as well as updates. If I go back to an older file with an update say 6 months later the transcription is more accurate. Same file same data same mp3. There also remains a problem with string and organ sounds due to the vibrato in the timbre of the sounds and is interpreted as pitch changes. It is beginning to improve slightly.

 

.wav is NOT always better, than .mp3

it can be, it depends on the source and encoder settings.

read up on the discovery of mp3 encoding, they used 10's of thousand of hours of listening by grad students to tune the
alogrithms to human pyscho acoustics.

its basically a method of throwing away sound that can't be perceieved by humans. developed by having humans listen to music and stripping away various bits, and giving it a go again, over and over, until they had a very good understanding of what was going on.

it can be used recursivly, which done to its ultimate extremity, will produce a file of zero bytes, that is total silence.
an mp3 encoder is a tool to strip away unperceived audio, to reduce file size, if you push it past optimal parameters, it will strip away perceived audio (but make yet even smaller file size)-

320kps is basically equivalent to the .wav original it was produced from unless you have a perfect source, a perfect listening environment, perfect ears, and perfect silence to listen to it in, and even then the difference will be terribly small.

.mp3 is a audio file size compression tool, like any tool you have know how to use it and you have to know what you want the tool to do (the final result, ie how small do need the file to be)

 

Not enough unfortunately, or else you would've stopped yourself from posting this.

 

Last edited: Mar 22, 2025