[music-dsp] Linearity of compression algorithms on more than one sound component
Hi all, Just a thought I share, because of associations I won't bother you with, suppose you take some form of audio compression, say Fmp3(wav) which transforms wav to an mp3 form, with some encoding parameters. Now we consider the linearity of the transform, most people will know this: Fmp3(Lambda * wav) ^= Lambda * Fmp3(wav) Fmp3(wav1 + wav2) ^= Fmp3(wav1) + Fmp3(wav2) Meaning, if I take independent transforms for the left and the right channel of an mp3 encoding, can I presume the stereo image is perfect, for instance, which isn't the case, different signal levels might change the encoding, and more spectral components (left and right combined of an actual stereo signal) added together take more encoding bandwidth. So it could be interesting to take some nice encoding (for all I care mp3, or aac or ac3), and use it for making a cheap sampler, where the samples of an instrument are encoded, and decoded when playing them back. I don't know, maybe for a nice cheap little toy sampler with bigger piano samples than fit through a SDCard bandwidth than when using wav files. T. -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp
Re: [music-dsp] Linearity of compression algorithms on more than one sound component
It's lossy. Definitely not linear. On Thu, Feb 12, 2015 at 4:33 PM, robert bristow-johnson r...@audioimagination.com wrote: On 2/12/15 3:02 PM, Theo Verelst wrote: Hi all, Just a thought I share, because of associations I won't bother you with, suppose you take some form of audio compression, say Fmp3(wav) which transforms wav to an mp3 form, with some encoding parameters. Now we consider the linearity of the transform, most people will know this: Fmp3(Lambda * wav) ^= Lambda * Fmp3(wav) Fmp3(wav1 + wav2) ^= Fmp3(wav1) + Fmp3(wav2) i don't think mp3 encoding is linear. i'm almost certain it is not. -- r b-j r...@audioimagination.com Imagination is more important than knowledge. -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp
Re: [music-dsp] Linearity of compression algorithms on more than one sound component
On 2/12/15 3:02 PM, Theo Verelst wrote: Hi all, Just a thought I share, because of associations I won't bother you with, suppose you take some form of audio compression, say Fmp3(wav) which transforms wav to an mp3 form, with some encoding parameters. Now we consider the linearity of the transform, most people will know this: Fmp3(Lambda * wav) ^= Lambda * Fmp3(wav) Fmp3(wav1 + wav2) ^= Fmp3(wav1) + Fmp3(wav2) i don't think mp3 encoding is linear. i'm almost certain it is not. -- r b-j r...@audioimagination.com Imagination is more important than knowledge. -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp
Re: [music-dsp] Dither video and articles
On 12/02/2015, gwenhwyfaer gwenhwyf...@gmail.com wrote: On 11/02/2015, Andrew Simper a...@cytomic.com replied to me: ... I made 7 sawtooth waves with random (static) phases and one straightforward sawtooth wave, with all partials in phase. I just listened to it again, to check my memory. On a half-decent pair of headphones, the difference between the all-partials-in-phase sawtooth and the random-phase ones is readily audible, but it was rather harder to tell the difference between the various random-phase waves; they all kind of sounded pulse-wavey. On a pair of speakers through the same amp and soundcard, though, I can still *jst about* pick out the in-phase sawtooth - but I couldn't confidently tell the difference between the 7 other waves. Which I'm guessing has something to do with the difference between the fairly one-dimensional travel of sound from headphone to ear, vs the bouncing-in-from-all-kinds-of-directions speaker-ear journey. Have you considered that headphones don't have crossovers? Nope. Good point. Indeed, it does seem to be a bit easier to pick out the in-phase sawtooth on the hideous tinny laptop piezo-buzzers I've got in front of me... but I'm not randomising the order of them or anything, and I really should be doing that, so interpret my report as subject to confirmation bias. Crest factor? I can't easily find out, but a visual inspection shows that all the waves are hitting one rail or the other. Which makes me think I normalised each wave individually, which means I introduced RMS differences as a means of distinguishing them... OK, forget I said anything. *pipes down* -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp
Re: [music-dsp] Dither video and articles
On 11/02/2015, Andrew Simper a...@cytomic.com replied to me: ... I made 7 sawtooth waves with random (static) phases and one straightforward sawtooth wave, with all partials in phase. I just listened to it again, to check my memory. On a half-decent pair of headphones, the difference between the all-partials-in-phase sawtooth and the random-phase ones is readily audible, but it was rather harder to tell the difference between the various random-phase waves; they all kind of sounded pulse-wavey. On a pair of speakers through the same amp and soundcard, though, I can still *jst about* pick out the in-phase sawtooth - but I couldn't confidently tell the difference between the 7 other waves. Which I'm guessing has something to do with the difference between the fairly one-dimensional travel of sound from headphone to ear, vs the bouncing-in-from-all-kinds-of-directions speaker-ear journey. Have you considered that headphones don't have crossovers? Nope. Good point. -- dupswapdrop -- the music-dsp mailing list and website: subscription info, FAQ, source code archive, list archive, book reviews, dsp links http://music.columbia.edu/cmc/music-dsp http://music.columbia.edu/mailman/listinfo/music-dsp