EHARNDEN@AUVM.BITNET (Eric Harnden) (02/11/90)
ok, i think i'm a little confused here... a lot of synths these days take 'pcm' data in one form or another. recently i did a dump from my waveframe, and looked at the data. the tech in boulder had said something about it being 'pcm' (in one conversation), and '16 bit linear' (in another). it looked like 2's complement to me. now, isn't linear normally what you call it when the zero is just offset, so the most negative value is near numeric zero, the most positive is up at the top, and the relative zero is in the middle of the numeric range? was i just having a fuzzy conversation, or is linear becoming confused with 2's complement these days? and what's 'pcm' anyway? the data is clearly just simple sample points in the old style... each word represents an amplitude level. if this is pulse code modulation, then what is the pulse that is being modulated? i see the code changing form word to word, but that doesn't sufficiently describe the term. also, 'pcm' didn't come to my attention as a term until it was used to describe the technique for encoding digital sound on linear tape. now i know that what's going on that tape is some kind of modulated carrier, and i presumed some time ago that the 'pulse code' in pcm was sort of like that in SMPTE sync tone... a data word that controlled the modulation of the carrier. but i never really bothered to find out. still, i get the distinct impression that the term has fed back into the technology under a different meaning, since i now hear of run-of-the-mill parallel-bit sample data referred to as pcm. so what gives? any enlightenment out there in emusicland? Eric Harnden (Ronin) Mailing From: <EHARNDEN@AUVM> Physics Dept./Audio Tech. Transmedia Music The American University (202) 347-8050 Wash., D.C
adamson@ITD.NRL.NAVY.MIL (Brian Adamson) (02/12/90)
Pulse code modulation is just a description of how the analog was sampled into a digital signal. People use thsi term in a general (perhaps incorrect) sense when talking about analog-to-digital conversion. Other techniques to do A/D conversion besides PCM could include continuous delta conversion techniques. As far as linear is concerned, that just refers to the fact that there is relationship (mathematically) between the digital data and the analog signal is linear. For example non-linear would indicate that some sort of companding or compression rule had been used when A/D converting the analog signal. An example of this is mu-law or A-law mapping which is used for speech. By optimizing the fit of the digital data to the particualar analog signal (speech) using empirical data, 8 bit analog-to-digital conversion can sound like 12 bit (but only for speech using mu-law or A-law) This is due to the fact that the human ear doesn't necessarily respond in a linear fashion to sound. Of course these mappling rules don't preserve the dynamic range of the input signal (O.K. for speech) so probably want to maintain a linear representation of the analog signal when sampling for hi-fi music purposes. Hope this helps, Brian Adamson