HydrogenAudio

Lossy Audio Compression => Other Lossy Codecs => Topic started by: Neuron on 2012-12-29 01:15:13

Title: First perceptual lossy codec?
Post by: Neuron on 2012-12-29 01:15:13
Just out of curiosity and interest in history, what was the first lossy compression method that was based upon perceptual/psychoacoustic principles (= not waveform prediction like ADPCM or companding like u-law or NICAM)? Was it MP1, ATRAC or were there codecs like that before?
Title: First perceptual lossy codec?
Post by: C.R.Helmrich on 2012-12-29 13:35:32
In the scientific literature, the earliest experiments I can recall are by Atal and Schroeder, based on linear predictive coding (LPC):

Predictive Coding of Speech Signals and Subjective Error Criteria (IEEE Trans. ASSP, 1979. Conference paper is from 1978) (http://148.204.64.201/paginas%20anexas/voz/articulos%20interesantes/F/7.pdf)

For music, developments started only in the mid 80s, I think. There's a paper by Johnston (Woodinville (http://www.hydrogenaudio.org/forums/index.php?showuser=18957) or JJ here on HA):

Transform Coding of Audio Signals Using Perceptual Noise Criteria (IEEE Journal Sel. Areas in Comm., 1988. Paper manuscript is from 1987) (http://www.ee.columbia.edu/~dpwe/papers/Johns88-audiocoding.pdf)

So yes, MPEG 1 Layer 1 with a perceptual encoder according to JJ's paper is what I consider the first perceptual codec. There might also be a CELP speech codec from that time with the Atal/Schroeder perceptual model.

Chris