Sample Sound Quality and Methods

83dude wrote:Quick tip: listening twice to the same mp3 never produces the same results, since psychoacoustical applications take place. They are not discrete in terms of amplitude and time. If possible, ALWAYS use WAV. mp3 is a so-called consumer format sacrificing quality providing usability: It‘s compact and transports well across devices. Think of it as cash: bills do transport much better, but only coins give you enough „detail“ to make a difference between 0.39 and 0.41 instead of compromising somewhere else.

hyena wrote:Sorry to be "that guy" but no. psychoacoustic process happens just once: its the encoder that does it, when the mp3 is created it will always sound the same. think how anti-economic a shapeshifting algorhitmic sound file format might be...
the encoder just chops out data based on psychoacoustic knowledge (which was confirmed by double blind tests on real people), then once the mp3 is done is done, it will always sound the same. that said, i also prefer to never use them, especially if i have to process that sound.

83dude wrote:

hyena wrote:Sorry to be "that guy" but no. psychoacoustic process happens just once: its the encoder that does it, when the mp3 is created it will always sound the same. think how anti-economic a shapeshifting algorhitmic sound file format might be...
the encoder just chops out data based on psychoacoustic knowledge (which was confirmed by double blind tests on real people), then once the mp3 is done is done, it will always sound the same. that said, i also prefer to never use them, especially if i have to process that sound.

You are right there, but you seem to forget that for all encoding there's a decoder required, too, and that can differ from programme to programme depending on the codec licensed for it. There are (arguably mostly negligible) shifts in timing or volume. Depending on the decoder used (Fraunhofer, ACM, LAME, etc.) by your machine or tool the process can differ. I might have exxagerated saying they "never" sound the same, but they do vary playbackwise depending how your computational processes are used.

This is obvious with long mp3 files or when cutting a video to a mp3 file. Even if both programmes are compatible, video cuts will be off depending on the decoder licensed for the software. The principles are the same, but the algorithms are not. Long mp3 files may produce silence at the end of the track, if e.g. a Fraunhofer-encoded file gets decoded by ACM.

This is strange indeed, but noticable anyways.

Terrico wrote:Where can I find WAV files to sample?